Sahyadri Journal of Research...It gives us immense pleasure to bring out Volume 4, Issue 2 of Sahyadri International Journal of Research (SIJR). The journal covers the wide disciplines

Sahyadri Journal of ResearchVOL. 4 ISSUE 2 ISSN 2456-186X DECEMBR 2018

SIJRJournal

Research

Research Papers

Review Papers

Scientific Articles

International

AdvisorsDr. D L Prabhakara (Director)

Dr. Manjappa Sarathi (Director-Consultancy)

Dr. Umesh M Bhushi (Director- Strategy & Planning)

Dr. R Srinivasa Rao Kunte (Principal)

Dr. Rajesha S (Dean- Academic)

Mr Ravichandra K. Rangappa (Dean- Industry)

Mr. Ramesh K G

Dr. Shriganesh Prabhu, TIFR, Mumbai

Dr. Dinesh Kabra. IITB, Mumbai

Editorial Board Dr. Pushpalatha K - Editor-in-Chief

Dr. Ramakrishna Sharma - Editor

Dr. Navin. N. Bappalige- Editor

Dr. Rathishchandra Gatti - Editor

Dr. Ravindra Babu G - Editor

Dr. Niraj Joshi – Editor

Mr. S N Bharath Bhushan - Editor

Mr. Balaji. N - Editor.

Mrs. Geetha S D - Editor

Mrs. Megha N - Editor

Mrs. Aysha Shabana - Editor

Mrs. Smitha - Editor

Disclaimer: The individual authors are solely responsible for

infringement, if any, of Intellectual Property Rights of third

parties. The views expressed are those of the authors. Facts

and opinions published in SIJR express solely the opinions of

the respective authors. Authors are responsible for citing of

sources and accuracy of references and bibliographies.

Although every effort will be made by the editorial board to

see that no inaccurate or misleading data, opinion or

statements appear in this journal, the data and opinions

appearing in the articles including editorials and

advertisements are the responsibility of the contributors

concerned. The editorial board accepts no liability

whatsoever for the consequences of any such inaccurate or

misleading data, information, opinion or statements.

Contents

SAHYADRI International Journal of Research | Vol 4 | Issue 2 | December 2018

Contents

Editorial

Research / Review Articles

Recent Trends in Point Absorber Wave Energy Converters 1-6

Crystal and surface structural studies on

Sol-gel derived Ga-doped ZnO thin films 7-11

EEG based classification for Alcoholism 12-16

Some Classes of Diophantine Edge Graceful graph 17-21

Discrete Cosine Transform Coefficients for

Kannada Hand-Written Character Recognition 22-26

Classical and Refined Beam and Plate Theories: 27-32

A Blockchain Approach for Eliminating Counterfeit

Drugs in Pharma Supply Chain 33-35

Vol. 4, Issue 2ISSN 2456-186X (Online)ISSN Pending (Print)

Mailing Address:Editor Sahyadri International Journal of ResearchSahyadri campus, Adyar, Mangalore - 575 007, IndiaE-mail: [email protected]: www.sijr.in

SAHYADRIInternational Journal of Research

It gives us immense pleasure to bring out Volume 4, Issue 2 of Sahyadri International Journal of Research

(SIJR). The journal covers the wide disciplines in science, engineering and technology. SIJR is published

biannually and is an open access journal available online. The focus of the Journal is to motivate the

researchers of various disciplines to publish their quality research. The most important disciplines in

which we would focus are: Physics, Chemistry, Applied Mathematics, Electronics and Communication,

Mechanical Engineering, Civil Engineering and Computer Science and Engineering. This issue has the

research and review articles of current trends in various disciplines.

In this occasion I would like to express heartfelt appreciation to all authors and reviewers of the SIJR on

behalf of the entire editorial board and the publisher. It was with the mere co-operation, enthusiasm,

and spirit of the authors and reviewers we could make SIJR a grand success. I thank all the authors in

considering and trusting SIJR as the platform for publishing their valuable work. I also thank for their

kind co-operation extended during the various stages of processing of the manuscript in SIJR.

The reviewing of a manuscript is very essential to assure the quality of the manuscript published in any

journal. The inputs of reviewers are frequently used in improving the quality of a submitted manuscript.

I thank all reviewers for their excellent contributions and support for the journal.

I also wish to acknowledge the contributions made by the dedicated members of our Editorial Board,

the invaluable support given by advisors and the Sahyadri Management, and the hard working,

professional staff of the publishing office.

Finally, I would like to thank the readers of SIJR, for your interest in the journal. We welcome your

valuable feedback and ideas for further improvement of SIJR.

Editor-in-Chief

Editorial

1SAHYADRI International Journal of Research | Vol 4 | Issue 2 | December 2018

SAHYADRI INTERNATIONAL JOURNAL OF RESEARCH, VOL 4, ISSUE 2, 2018

1

Rathishchandra.R.Gatti*, Ajith.B.S

Mechanical Engineering Dept., Sahyadri College of Engineering & Management, Adyar, Mangaluru-575007 *Email:[email protected]

ABSTRACT Ocean energy still remains as unexplored massive source of renewable energy. Survivability of the structures is the major impediment in harnessing this type of energy. Amongst ocean energy conversion, wave energy is a viable option since the wave energy converters are not detrimental to aquatic life. Point absorber wave energy converters have better structural survivability compared to other wave energy converters due to their small size relative to the wavelength of the incident waves. A brief survey of the commercial deployment of the point absorber wave energy converters is presented in this paper. Point absorbers continue to be viable option for wave energy conversion. Keywords: buoy, ocean waves, point absorbers, renewable energy, wave energy converter, wave energy

1. INTRODUCTION Demand for energy has increased day by day and at the

same time, fossil fuels sources are being increasingly depleted in nature. Hence, there is a huge concern on looking for alternative energy sources. The rapid industrialization has resulted in the increased emission of carbon monoxide and other hazardous gases. Hence it has become pertinent to study the various renewable resources in the field of electricity generation. Due to global industrialization, the world has severely increased global warming phenomena such as rising CO2 levels which have necessitated more focus on extracting electricity from renewable sources. Among renewable energy, ocean wave energy is one of the promising forms of energy with an estimated potential of about 2TW of the world’s total power generation [1].

Ocean Waves are created mainly sue to the oceanic winds and the air-water interactions on the sea surface. Wind blowing over the surface of the ocean push the water due to the geological effects and may travel thousands of miles before striking the land. The size of the waves vary from small ripples to tsunamis and create large energy potential out of which a feasible portion can be converted to useful electrical energy. Solar energy is the main source and cause of wave energy, as the sun’s thermal radiation increases the air temperature which leads to generating winds and finally creates propagating waves along the surface of the ocean [2].

The largest quantity of wave energy received is by Asia and Australasian region as shown in Fig.1[3] with South and North America also receiving fair amount. Wave energy is also significantly available at western seaboard Western and Northern Europe performs well given its relatively small size. Further Mediterranean Sea, Atlantic Archipelagos and Central America have lesser wave energy potential compared to other regions. The total world’s wave energy potential is of 29,500 TWh/yr. Fig1. Distribution of total world's wave energy potential of 29,500 TWh/yr[3].

Recent Trends in Point Absorber Wave Energy Converters


2

2. BENEFITS Compared to other modes of energy generation, ocean waves as a source of renewable energy offers significant advantages. Among all renewable energy sources, ocean waves provide the highest energy density [4]. Waves are generated in the ocean by temperature variation in the atmosphere due to the solar radiation. Nearly 0.1- 0.3kW/m2 of solar intensity in horizontal surface is converted to an average power flow intensity of 2 - 3 kW/m2 of a vertical plane perpendicular to the wave propagation slightly below the surface of water [5]. Negative environmental impact is limited in use. Thorpe [6] explains the potential impact and presents an estimation of the life cycle emissions of a typical near shore device. In general, it is noted that lowest potential impact in offshore devices. Waves are active 24 hours a day, 365 days a year. With minute loss in energy, waves can travel large distances. The wind with high velocity carries the waves to large distance [4]. It is observed that wave power devices can generate power up to 90 % of the time, when compared to about 20–30 % for solar and wind power devices [7, 8].

3. CHALLENGES

The harnessing of Ocean energy is significantly challenging mainly due to the issues that needs to be overcome that would result in enhanced performance of wave power devices and would make wave energy commercially competitive in the global energy market. A major challenge is to extract energy from oscillatory, random and low frequency waves and convert extracted energy into useful electrical energy [1]. As the wave height and period randomly vary , the respective power levels also vary in random. Hence, it is challenging to convert variable input to smooth electrical energy. In offshore locations, wave energy converters align themselves according to the direction of waves that are random in nature. The challenge of effective design of energy extraction devices to withstand extreme wave conditions. The speed of offshore waves is around 30–70kW/m [9]. The extreme saline nature of sea water makes it pertinent to have corrosion resistant coating for the devices that are employed under operating conditions. This results in higher operating costs [4][10].The ecosystem is disrupted during the construction of powerhouse. This affects the fish and also the fishermen whose livelihood depends on fishing. 4. TYPES OF OCEAN WAVE ENERGY CONVERTERS Ocean can be considered as a storage house of various forms

of energy such as heat stored from the solar radiation, energy from wind and surface water, energy from water currents, energy from biomass, energy from difference in salt concentrations and energy stored in the form of uranium resources to name a few. The primary classifications of available and know ocean energy is based on how they can be converted to electrical energy, the most useable energy format for operations as given in Table 1 below. Nuclear resources definitely have the highest potential but owing to the cost of extraction of these resources and subsequent energy extraction in controlled fission reactors. Geothermal energy is the next big bet but is location specific and not abundant. Of these, Ocean thermal energy converters (OTECs) and ocean wave energy converters are reasonably available in most of the oceanic regions and feasible. Of this the OTECs also have significant energy losses due to the inherent nature of heat as a highly susceptible form of energy to expend. Thus, more research is now concentrated on harnessing Ocean wave energy. Also, if one sums up the coastal line of all the landmass regions of tour planet, it is about 5000 kilometers owing to OWEC potential of approximately 5000GW. Only a small percentage of this energy potential can be harnessed owing to geographical and other region specific environmental conditions as well as the premature state of wave energy products.

Table 1 Classification of Ocean energy converters [11]

No. Ocean energy Basic principle of energy transduction

Potential (GW)

1 Ocean thermal energy conversion (OTEC)

Temperature gradient to electric energy

10,000

2 Ocean wave energy conversion(OWEC)

Wave kinetic energy to electric energy

5,000

3 Tidal energy conversion(TEC)

Potential difference between high tides and low tides used to extract energy

200

4 Ocean current energy conversion(OCEC)

Kinetic energy of the ocean currents converted to electric energy

50

5 Ocean salinity gradient energy conversion(OSGEC)

Electric potential between saline water and fresh water separated by a semipermeable membrane

3,540

6 Offshore geothermal energy conversion(OGEC)

Thermal energy from the Geothermal fluid to electrical energy

30,000

7 Ocean bio-mass resources

Ocean biomass as fuel is converted to useful forms of energy including electrical energy.

800

8 Ocean nuclear energy conversion

Uranium resources in ocean used in nuclear power plants.

80,000

Total Ocean energy potential of the world 129,000


3

Due to significant research and development of new prototypes, classification of wave energy converters (WECs) remains vague. According to their location relative to the land, they can be classified as off-shore, near-shore and on-shore types[12][13]. According to their operation on the sea level, they can be classified as emerged, submerged and semi-submerged. According to the structural configuration, they can be classified as freely-floating with the mooring system anchored to the sea bed or bottom-standing. One prominent classification that makes sense is distinguishing the OWECs by the method of energy capture[12][13] as indicated in the Table 2 below. This classification is widely used in the reported research and developed products. Table 2 Classification of OWECs based on Energy capture

[12] No. OWEC type Principle of energy capture 1 Oscillating water

column The air turbine is run by an air column pushed by water column that is pushed up during the incident wave tide.

2 Archimedes effect Relative motion between the air filled buoy and the fixed base that is moored to the seabed.

3 Floating buoy with fixed reference

Float moves with respect to a fixed reference, usually the point that is fixed on the seabed.

4 Floating buoy with moving reference

Float moves with respect to a moving basement that is not anchored.

5 Overtopping Shore side consisting of turbine that is run by the tide hitting the shore.

6 Impact Submerged oscillating WEC that oscillates during impact force.

Another classification based on current deployment literature is class A to class I classifications from European Marine Energy Centre [13] as per the table 3. 4. ABOUT POINT ABSORBERS A point absorber is a buoyant WEC with single or multi-degree of freedom body oscillating relative to a fixed member[14]. The build of the point absorbers are small compared to the average wavelength as decided by the analysis of the wave characteristics. Point absorbers are submerged WECs designed to convert the surface waves to generate electricity. Point absorbers are built custom to their locational wave characteristics and hence a detailed investigation of the

wave parameters such as wave frequency, wave length, maximum wave displacement and wave period to name a few. Also, in general, their size is designed to be about 0.05 to0.1 times the most common wavelength as per annual wave data characteristics. This is followed by wave structure interaction analysis to determine the structural stability and natural frequency of the proposed point absorber system. It has been found that maximum power is achieved when the natural frequency of the point absorber is in resonance with the incident wave frequency. Hence, point absorbers are designed after determining the most commonly occurring incident wave frequencies. Another important design consideration is the structural survivability aspect of the point absorber which is achieved by fluid structure interaction models that achieve greater stabilities of the pitch, yaw and surge forces of the prevalent wave conditions. Very often, the structural survivability is achieved at the expense of energy conversion efficiency. The other design factors to be considered include electro-mechanical transduction mechanism, design for safety and reliability, build costs, life cycle costs and costs of maintenance. Table 2 Classification of OWECs based on Energy capture

[13] Class OWEC type Deployment % A Attenuator 23% B Point absorber 40% C Oscillating wave surge

converter 19%

D Oscillating water column 7% G Bulge wave 1% H Rotating mass 7% E Overstopping/Terminator

3%

F Submerged-pressure differrential

I Others

One classification of PAWECs is based on their locations- offshore PAWECs, near-shore PAWECS and shore-line PAWECs. The design of Offshore PAWECs needs to be robust in terms of structural survivability where the waves are turbulent but wave energy potential is very high in these regions. Near shore PAWECs and shore-line PAWECs often witness flow that are relatively laminar compared to offshore and hence more safe but have lesser wave power potential. Another classification is based on the principal direction of wave energy conversion – heaving, pitching and surging out of which the heaving and pitching are most commonly used.


4

4. RECENT PATENTS IN POINT ABSORBERS Hagmüller, A.W and Levites-ginsburg, M. J (US 20190063395) proposed an intelligent wave energy converter system that is adaptive to the prevailing sea conditions by sensing the wave parameters and being responsive such that the natural frequency of the WEC system is in resonance with the frequency of the sea waves[15]. Point absorbers using elongated inexpensive light weight buoys with self-orientation capabilities to multi-directional wave forces was proposed by Rohrer,J.W (15/863947)[16].These PAWECs are capable of harvesting energies from both heaves and surges of the existing waves. Another feature of this PAWEC is to submerge completely during severe seas to accommodate survivability conditions. Thresher, R. et.al,(US20190040839)[17] proposed wave energy converters that can be geometrically reconfigured by employing actuated geometry components to effectively transfer wave energy to the PTO component.

A method of harnessing energy from the heave motion of the incident waves by estimating and computing the heave excitation followed by subsequent application of the controlled force was suggested by Abdelkhalik, O. et al.( US10197040)[18]. The estimation of the heave excitation can be obtained by measuring the surface pressure and position of the buoy whichis more accurate than the measurement of wave elevation.

Gregory, B.[19] suggested a submerged point absorber that lies between the float at the sea surface and the sea bed by means of two special mooring tauts on either side. It is possible to dynamically control the lengths of these tauts either between sea-surface and point absorber or between point absorber and the sea bed in accordance with the wave height at that instant.

A self-powered computing apparatus integrated within the float was suggested by Sheldon-Coulson, G.A (US2018/042023)[20].This device also includes novel features to economically cool the computing circuits using its close proximity of the sea water and oceanic wind waves.

Sidenmark, M. (SE2018/050599)[21] suggested a PTO device of PAWEC which consists of a float and a drive unit. The drive unit comprises a mechanism to convert linear motion of the buoy to convert to rotary motion, rotary input shaft, variable transmission unit and energy storage device. The variable transmission provides adaptive control on storage or retrieval of the stored power and also the force applied from the drive unit to the float.

Point absorbers using elongated inexpensive light weight buoys with self-orientation capabilities to multi-directional wave forces was proposed by Rohrer,J.W (15/863947)[22].These PAWECs are capable of harvesting energies from both heaves and surges of the existing waves. Another feature of this PAWEC is to submerge completely

during severe seas to accommodate survivability conditions. Todalshaug, J.H[23] introduced a PAWEC that consists of a

float oscillationg relative to a fixed point and a negative spring connects the float to the fixed point. Here, the negative spring helps to apply positive force in the direction of displacement as the float moves away from its mean position.

Yang,Y.(US 15/301823)[24] proposed a wave energy converter that has a special unidirectional rotor consisting of multiple lift type and/or drag type blades and a vertical shaft perpendicular to the incident waves. 4. COMMERCIAL DEVELOPMENTS OF POINT ABSORBERS Seabased, a Swedish company in collaboration with Swedish agency and Fortum has developed its PAWECs at Sotenas, Ghana, Aland and Islandsberg[25]. In their Waveparks a multitude of PAWECs are connected to the linear generators beneath the sea that convert them into electrical energy and send to subsea switchgears that are connected to the grid. SINN Power GmBH[26] recently launched its second phase of renewable energies project at Guinea for its client Guinea Gold PLC. SINN Power WECs are arrays of PAWECs that reciprocate up and down in a constrained structure. Each PAWEC reciprocates a rod that acts as driver for the generator unit. Another WEC concept that uses hydraulic pistons placed on a point absorber was proposed by Ocean Grazer, a dutch startup[27]. The startup is currently planning to develop hybrid offshore energy converters by integration wind and other feasible renewable sources with ocean wave power extraction. Flansea from Belgium[28] introduced an offshore PAWEC that works based on the bobbing effect of the point absorber buoy on the cable. It was custom designed for deployment in the Southern North sea wave environment. SDE from Israel [29] developed SDE Waves power plant in 2010, which is based on hydraulic rams connected to the generators are operated by the pumping motion of buoys. One of the SDE waves power generation produced peak power of about 40kWh during its two year operational tenure (2008-2010). The WaveEL buoy [30] developed by a Swedish company Waves4Power is PAWEC designed for survivability with shorter horizontal dimensions compared to the average wavelength of the incident ocean waves. This buoy has a long vertical tube with a piston connected to the power conversion system. The moors attached to the point absorbers secure them to their location but also allow them for free vertical motion to extract energy. SeaRaser from Alvin Smith, UK [31] consists of a point absorber float tethered to a piston pump anchored to seabed. The point absorbers generate pressurized water during their


5

oscillations. The pressurized water is sent through pipes to onshore hydraulic generators to generate electricity. Oceanus2 launched by Seatricity Ltd,UK [32] consists of a gimbal mounted piston pump supported by the buoy with the pump moored to the seabed. The point absorber reciprocates vertically up and down thus driving the pump and developing hydraulic pressure. The pressurized water is either sent to reverse osmosis desalination plant as input raw material or hydraulic turbines for power generation. OE Buoy developed by Ocean Energy, Ireland [33] consists of only a simple moving part and has potential rated capacity of 125 MW. Other prominent Point absorbers include the AquaBuoY from SSE Renewables [34], Lysekil project from Uppsala University[35], Wavestar Buoy from Wavestar A/S[36],CETO Wave power from Carnegie, Australia [37], Wavebob from Ireland [38], Atmocean from US[39] and PowerBuoy from Ocean Power Technologies, US[40] to name a few.

3. CONCLUSION It has been observed that there is considerable research and working prototypes being developed across the world to harness wave energy converters. This is due to the massive availability of wave energy that is available and can be utilized for growing energy needs. PAWECs and other wave energy converters are safer options without significantly hampering the aquatic life and the ocean environment. Survivability of the PAWECs is stillan issue that will affect the wave energy extraction process and hence stability needs to be critically addressed in the PAWEC design.

REFERENCES

4. CONCLUSION It has been observed that there is considerable research and working prototypes being developed across the world to harness wave energy converters. This is due to the massive availability of wave energy that is available and can be utilized for growing energy needs. PAWECs and other wave energy converters are safer options without significantly hampering the aquatic life and the ocean environment. Survivability of the PAWECs is stillan issue that will affect the wave energy extraction process and hence stability needs to be critically addressed in the PAWEC design.

REFERENCES [1] B. Drew, A. R. Plummer, and M. N. Sahinkaya, “A

review of wave energy converter technology,” Proceedings of the IMechEA: Journal of Power and Energy, vol. 223, no. 8, pp. 887–902, 2009.

[2] Mørk G et al (2010), Assessing the global wave energy potential. In Proceedings of OMAE2010 (ASME), 29th International Conference on Ocean, Offshore Mechanics and Arctic Engineering. Shanghai, China.

[3]World Energy Council, World Energy Resources 2016,7-8.

[4] Cornett AM (2008) A Global Wave Energy Resource Assessment. Proceedings of ISOPE, 2008 (March), p.9.

[4] Clement, A., McCullen, P., Falc.ao, A., Fiorentino, A., Gardner, F., Hammarlund, K., Lemonis, G., Lewis, T., Nielsen, K., Petroncini, S., Pontes, M.-T., Schild, B.-O., Sjostrom, P., Soresen, H. C., and Thorpe,T. Wave energy in Europe: current status and perspectives. Renew. Sust. Energy Rev., 2002, 6(5), 405.431.

[5] Falnes, J. A review of wave-energy extraction. Mar. Struct., 2007, 20, 185–201.

[6] Thorpe, T. W. A brief review of wave energy, Technical report no. R120, Energy Technology Support Unit (ETSU), A report produced for the UK Department of Trade and Industry, 1999.

[7] Pelc, R. and Fujita, R. M. Renewable energy from the ocean. Mar. Policy, 2002, 26(6), 471–479.

[8] Power buoys. The Economist, 19May 2001. [9] Polinder,H. and Scuotto,M.Wave energy converters and

their impact on power systems. In Proceedings of the 2005 International Conference on Future power systems, 2005, pp. 1–9.

[10] Leijon, M., Danielsson, O., Eriksson, M., Thorburn, K., Bernhoff, H., Isberg, J., Sundberg, J., Ivanova, I., Sjöstedt, E., Ågren,O.,Karlsson, K. E., andWolfbrandt, A. An electrical approach to wave energy conversion. Renew. Energy, 2006, 31,1309–1319.

[11] Rao,S.S, Parulekar,B.B,Energy Technology,Khanna Publishers,2009.

[12] Dolores E M, José S LG, Vicente N. Classification of Wave Energy Converters. Recent Adv Petrochem Sci. 2017; 2(4): 555593. Dolores E M, José S LG, Vicente N. Classification of Wave Energy Converters. Recent Adv Petrochem Sci. 2017; 2(4): 555593.DOI: 10.19080/RAPSCI.2017.02.555593.

[13] EMEC, Wave devices, accessed from www.emec.org.uk

[14] Faizal, M., Ahmed, M. R., & Lee, Y. H. (2014). A design outline for floating point absorber wave energy converters. Advances in Mechanical Engineering, 6, 846097.

[15] Hagmüller, A.W and Levites-ginsburg, M. ,Sea Wave Energy Converter Capable of Resonant Operation,USPA20190063395.


6

[16] Rohrer,J. W.,Ocean wave energy converter capturing heave, surge and pitch motion ,US8581432B2.

[17] Thresher, R.,Lawson, M.,Tom, N. ,Cotrell, J. ,Yu, Y.,Wright, A.,Wave Energy Conversion Incorporating Actuated Geometry,USPA 20190040839.

[18] Abdelkhalik, O. ,Robinett III, R. D. ,Zou, S.,Bacelli, G. ,Wilson, D. G. ,Korde, U. ,Optimal control of wave energy converters,US10197040.

[19] Gregory, B. , Wave energy converter, US10190568. [20] SHELDON-COULSON, G. A.,MOFFAT, B. L.,Self-

Powered Computing Buoy, WO/2019/018226. [21] SIDENMARK, M., Power Take Off Device

comprising c Variable Transmission for Use in a Wave Energy Converter,WO/2018/226152.

[22] Rohrer,J. W.,Wave Energy Converter With Concurrent Multi-Directional Energy Absorption, USPA20180306165.

[23] Todalshaug, J. H., Wave energy converter with negative spring, US10082127.

[24] Yang, Y.,WAVE ENERGY CONVERTER, USPA 20180202412.

[25] Accessed information from https://www.seabased.com/ [26] Accessed information from

https://www.sinnpower.com/ [27] Accessed information from https://oceangrazer.com/ [28] Accessed information from http://www.flansea.eu/ [29] Accessed information from

https://www.energydigital.com/renewable-energy/sde-power-india-sea-wave-power-plants

[30] Accessed information from https://www.waves4power.com/

[31] Accessed information from https://www.ecotricity.co.uk/our-green-energy/our-green-electricity/and-the-sea/seamills

[32] Accessed information from http://seatricity.com/technology/

[33] Accessed information from http://www.oceanenergy.ie/oe-technology1/platform

[34] Vicinanza, D., Margheritini, L., & Frigaard, P. (2007). Aquabuoy wave energy converter. Department of Civil Engineering, Aalborg University: Aalborg, Denmark.

[35] Haikonen, K., Sundberg, J., & Leijon, M. (2013). Characteristics of the operational noise from full scale wave energy converters in the Lysekil project: Estimation of potential environmental impacts. Energies, 6(5), 2562-2582.

[36] Ransley, E. J., Greaves, D. M., Raby, A., Simmonds, D., Jakobsen, M. M., & Kramer, M. (2017). RANS-VOF modelling of the Wavestar point absorber. Renewable Energy, 109, 49-65.

[37] Mann, L. D. (2011). Application of ocean observations & analysis: The CETO wave energy project. In Operational Oceanography in the 21st Century (pp. 721-729). Springer, Dordrecht.

[38] Sharkey, F., Conlon, M., & Gaughan, K. (2011, November). Investigation of wave farm electrical network configurations. In World Renewable Energy Congress-Sweden; 8-13 May; 2011; Linköping; Sweden (No. 057, pp. 2222-2229). Linköping University Electronic Press.

[39] Kithil, P. W. (2008). Reducing hurricane intensity by cooling the upper mixed layer using arrays of Atmocean, Inc.'s wave-driven upwelling pumps.

[40] Edwards, K., & Mekhiche, M. (2014). Ocean Power Technologies Powerbuoy®: System‐Level Design, Development and Validation Methodology.

https://www.seabased.com/

https://www.sinnpower.com/

https://oceangrazer.com/

http://www.flansea.eu/



https://www.waves4power.com/

https://www.ecotricity.co.uk/our-green-energy/our-green-electricity/and-the-sea/seamills

https://www.ecotricity.co.uk/our-green-energy/our-green-electricity/and-the-sea/seamills

http://seatricity.com/technology/

http://www.oceanenergy.ie/oe-technology1/platform


ISSN: 2456-186X, Published Online February, 2019 (http://www.sijr.in/)

7

Felcy Jyothi Serrao1*, K. M. Sandeep2, S.M. Dharmaprakash3

1 Department of Physics, Sahyadri College of Engineering and Management, Mangalore, India-575007. 2 Department of Physics, Bearys Institute of Technology, Inoli, Mangalore, 574153, India 3 Department of studies in Physics, Mangalore University, Mangalagangothri, India-574199

*E-mail: [email protected]

ABSTRACT Highly transparent gallium-doped zinc oxide (GZO) thin films were prepared on glass substrate by optimized sol-gel spin coating technique. Zinc acetate and gallium nitrate were used as precursors for Zn and Ga ions respectively. Combined effects of doping concentration and thermal annealing temperatures in air on crystal and surface structural properties of the GZO films were investigated by X-ray diffraction (XRD) and atomic force microscopy (AFM) respectively. The X-ray diffraction studies revealed the polycrystalline nature of the films with hexagonal wurtzite phase and showed compressive stress along c-axis. The compressive stress was found to increase with increase in gallium concentration. The grain size of the films was calculated using the Scherrer formula. It was found that the grain size, lattice constant, and surface roughness were significantly affected by the doping concentration and annealing temperature. The GZO films deposited with doping concentration of 1 at% Ga, annealed at 500oC exhibit optimum characteristics and could be used for the display applications and in optoelectronic devices.

Keywords: Ga doped ZnO, Doping Concentration, spin-coating, annealing temperature

1. INTRODUCTIONIn the last decade, among various II-VI semiconductors, zinc oxide (ZnO) has received considerable attention due to its novel properties and used as a promising alternate transparent conducting oxide (TCO) material to replace indium tin oxide (ITO). ZnO is an inexpensive n-type semiconductor having hexagonal wurtzite structure with a direct energy wide band gap of 3.37eV and large exciton binding energy of 60 m eV at room temperature which are widely used as transparent electrodes for optoelectronic devices like light emitting diodes, flat panel displays, solar cells etc., [1]-[4]. Recently, ZnO-based TCO thin films have engrossed much attention due to their interesting opto-electrical properties. Group III metal elements such as In, Al and Ga have considered as the most suitable doping elements. Among these impurities, gallium is of great interest because the ionic and covalent radii of Ga3+ (0.62Å, 1.26Å) are much closer to the size of Zn2+ (0.74Å, 1.31Å) than to those of Al (0.5Å, 1.26Å) and In (0.81Å, 1.44Å). In addition to this, Ga3+ dopants are more resistant to oxidation and less reactive than Al3+ dopants. Therefore, the addition of Ga to the ZnO host lattice reduces the deformation and stress in the ZnO lattice [5]. GZO thin films have been synthesized by various techniques such as magnetron sputtering, chemical vapour deposition, pulsed laser deposition

(PLD), sol-gel and spray pyrolysis [6]-[9]. Among these techniques, the sol-gel method has discrete advantages of homogeneity, controllability of compositions, simplicity and low cost.

In sol-gel method, the concentration of the precursor solution and annealing temperatures have a major role in defining the properties of the film. Therefore, in the present work, we report the combined effects of doping concentration and thermal annealing temperatures in air on crystal and surface structural properties of the GZO films.

2. EXPERIMENTALSol is prepared by taking zinc acetate dehydrate and gallium nitrate as starting materials in appropriate quantity. 2-methoxyehanol and Monoethanolamine were used as a solvent and sol stabilizer respectively. Three different dopant concentrations (1at%, 2at%, 3at%) were selected. The total sol concentration was retained at 0.5 mol L-1 and the molar ratio of MEA to zinc acetate was maintained at 1.0. The resulting mixture was aged for 2 days at 30oC. The Ga:ZnO films were deposited on a glass substrate (Corning1737) at 3000 rotations per minute (rpm) for 30s by the spin coating method at 30 oC.

Crystal and surface structural studies on Sol-gel derived Ga-doped ZnO thin films

SAHYADRI INTERNATIONAL JOURNAL OF RESEARCH, VOL 4, ISSUE 2, 2018 8

The resulted films were pre-heated at 150oC for 15 min, to evaporate the solvent and to eliminate organic residuals. Multiple spin-bake process is followed to achieve the desired thickness of the films. The resulting films were annealed for an hour at various temperatures

The structural characteristics of GZO thin films were explored by Rigaku Miniflex 600 Table Top Powder X-ray diffractometer using Cu Kα radiation with wavelength 0.154059 nm. Debye-Scherrer formula is used to estimate the crystallite size of the GZO films using XRD data. The topography of the film surfaces was studied using the atomic force microscope (AFM-Nanosurf Easy Scan 2). Ellipsometer is used to measure the thickness of GZO films. SHIMADZU 1800, a UV-visible spectrophotometer, is used to study the optical properties of the films the in the wavelength range 300-800nm.

3. RESULTS AND DISCUSSION

X-ray diffraction patterns (XRD) of GZO films at different Gadopant concentrations are shown in Fig.1. The dominantpeaks are due to ZnO (002) planes, indicating that all the GZOfilms have polycrystalline nature with a preferred orientationalong the c-axis i.e. [001] direction (JCPDS file no.36-1451)[10]. The intensity of the (002) plane of the films decreasedwith increasing Ga dopant concentration. This clearly specifiesthe degradation in the film quality, which may be due to thelattice deformation produced by the substitution of the Gaatoms for Zn sites in the ZnO host lattice (Fig.2) and the stressdeveloped by the smaller radius of Ga3+ ions (0.62Å)compared with Zn2+ ions (0.74Å) [11]. Details thatcharacterizes the growth is obtained from the XRD data. Theseinclude the average grain size, lattice strain, lattice stress anddefect density. The grain sizes of the GZO were estimatedusing the Debye-Scherrer formula [12]

θβλ

cos9.0

=D (1)

Where, D is the grain size, λ is wavelength of X-ray (0.154059nm), β is the full width at half maximum of the peaks in radians and θ is the diffraction angle. The strain along c-axis ( ) and the dislocation density ( ) of Ga:ZnO films werecomputed using the equations

4cosθβε = (2)

2

1D

=δ (3)

The stress in the direction of the c-axis is calculated based on biaxial strain model and using the following formula, which is valid for a hexagonal lattice [13]

o

o

ccc −

−= 233σ (4)

Where, c and are the lattice constants of the Ga:ZnO thin films and strain free ZnO thin films respectively. The lattice constant c can be computed by the following formula:

( ) )3

412

2

2

22

2 cl

akhkh

d+

++= (5)

The grain size, strain, dislocation density, c-parameter and calculated stress of the GZO thin films are presented in Table 1. It can be seen that the film with doping concentration 1 at%shows the better quality. The negative sign of the stressindicates that the stress of GZO films is compressive.Moreover, there is a slight increase in the compressive stresswith the Ga doping concentration and also the expansion of thec-lattice parameter indicating there is more interstitial Ga3+ inZnO lattice.

Fig.1: XRD patterns of GZO thin films with different gallium concentrations


Fig.2: Schematic representation of hexagonal wurtzite a) ZnO structure b) ZnO:Ga structure

Fig.3 exhibits the XRD patterns of GZO films, post-annealed at different temperatures. All the films showed hexagonal wurtzite crystal structure with a preferential growth of (002) plane, oriented along c-axis. The intensity of the peaks increases with increase in post-annealing temperature. The (002) peak positions of GZO film shifts to a larger diffraction angle as the post heating temperature is increased. The lattice constant ‘c’ and strain along c -axis were calculated and tabulated in Table 2. The negative sign of the strain indicates that strain in the films are compressive and decreases with annealing temperatures. The FWHM of the (002) plane decreases with increasing annealing temperature, indicating that the increase in crystallite size. The grain size of the films can be estimated using the XRD patterns and Debye-Scherrer’s formula and are shown in Table 2.

Fig.3: XRD patterns of the Ga:ZnO films annealed at various temperatures

Fig.4a shows the AFM images of gallium doped ZnO thin films at different doping concentrations. It can be realized that the surface morphology of the film changed with the dopant concentration. The surface roughness of the GZO films decreased from 18.268 nm to 5.318nm when the doping concentration was increased from 1at% to 3 at%. At 1 at% Ga, the largest grains were observed. This result is in agreement with the result of XRD. The AFM surface morphology (Fig.4b) indicates that the post-annealing treatment enhanced the grain size and morphology of the GZO film surface, significantly.

Fig.4a: AFM images of Ga:ZnO thin films (a) ZnO: Ga 1at% (b) ZnO: Ga 3at%

Fig.4b: AFM images of Ga:ZnO thin films annealed at (a) 300oC (b) 500oC

Optical transmittance spectra between 300 to 800 nm wavelengths of the Ga:ZnO thin films at different Ga dopant concentrations, annealed at 500oC are shown in Fig.5. It is crystal clear that all the GZO films show the average transmittance above 90% in the visible wavelength range with a sharp absorption edge in the UV region. It shows that the GZO film would be a good material for the display applications. A slight blue shift was observed with increasing Ga concentration indicates the broadening of the optical band gap. This deviancy can be described by the Burstein-Moss effect and is associated with carrier concentration [14]. The optical band gap (Eg) was analyzed using the relation [15]

(αhν)2 = c (hν - Eg) (6)

Where, h is Planck’s constant, ν is the frequency of the incident photon and c is the constant for direct transition.


Table 1: Structural and optical properties of GZO films with different Ga concentrations

Table 2: Properties of GZO films annealed at different temperatures

Fig.5: Optical Transmittance spectra of GZO filmswith different Ga doping concentrations

Fig.6 shows the plots of (αhν)2 against hν for the Ga:ZnO films at various Ga concentrations. The variation of the band gap energy (Eg) and average transmittance (T%) of GZO films with different Ga concentrations are shown in Table 1. It is clear that the direct optical band gap value (Eg) raised from 3.271 eV to 3.280 eV, when the concentration of Ga is increased from 1 at% to 3 at%. This widening of the optical band gap is associated with Burstein-Moss effect.

Fig.6: Plot of 2)( ναh Vs νh for ZnO: Ga (1, 2 and 3 at %) thin films

From the optical transmission spectra of GZO films annealed at different temperatures (Fig.7), it can be seen that all the GZO thin films exhibited a good transmittance in the 400 to 800 nm region (>89%). The optical band gap values (Eg) of GZO films can be estimated by the absorption coefficient (α) and photon energies (hν) and are presented in Table 2. It can be observed that the energy gap rises as the annealing temperature is elevated from 300oC to 500oC.

Ga (2θ) (002) FWHM (o) D ( nm) dhkl (Å) c (Å) ε (10-3) δ (1015) σ (GPa) T( %) Eg (eV)

1at% 34.405 0.335 24.825 2.6046 5.2092 1.396 1.623 -0.1163 89.97 3.271 2at% 34.403 0.379 21.941 2.6047 5.2094 1.579 2.076 -0.1252 93.39 3.276 3at% 34.388 0.393 21.160 2.6058 5.2116 1.637 2.233 -0.2236 94.10 3.280

Tan (oC) 2θ ( o ) FWHM (o ) D (nm) d (Ao) C (Ao) (%) T (%) Eg (eV)

300 34.398 0.353 23.570 2.605 5.210 0.065 89.338 3.253 400 34.453 0.331 25.140 2.601 5.202 -0.088 91.020 3.279 500 34.519 0.323 25.768 2.596 5.192 -0.280 93.705 3.289


Fig.7: Optical Transmittance spectra of GZO filmsAnnealed at different temperatures

4. CONCLUSIONThe Ga- doped ZnO thin films were prepared by simple sol-gel spin coating method. The Influences of doping concentration and post-annealing temperatures on the structural and optical properties of the films were investigated. We observed that the properties of GZO thin films significantly affected by the doping concentration and annealing temperature.

5. ACKNOWLEDGEMENTThe authors gratefully acknowledge Department of Physics, Mangalore University for providing facilities for the characterization of thin films and technical support to carry out the work.

REFERENCES 1. W. S. Han, Y. Y. Kim, B. H. Kong, H. K. Cho,

“Ultraviolet light emitting diode with n-ZnO/Ga/i-ZnO/p-GaN:Mg heterojunction”, Thin Solid Films,Vol: 517, pp.5106, 2009.

2. P. K. Song, M. Watanabe, M. Kon, A. Mitsui A, Y.Shigesato Y, “Electrical and optical properties ofgallium- doped zinc oxide films deposited by dcmagnetron sputtering”, Thin solid Films, Vol:411,pp.82-86, 2002.

3. G. X. Liang, P. Fan, X. M. Cai, D. P. Zhang and Z. H.Zheng, “The Influence of Film Thickness on theTransparency and Conductivity of Al-Doped ZnO ThinFilms Fabricated by Ion-Beam Sputtering”, Journal ofElectronic Materia, Vol: 40, pp.267_273, 2011.

4. Q. Zhang, C. Dandeneau, X. Zhou, G. Cao, “ZnONanostructures for dye sensitized Solar cells”, Adv.Mater., Vol: 21, pp.4087_4108, 2009.

5. K. Yim, H. W. Kim, C. Lee, “Effects of annealing onstructure, resistivity and transmittance of Ga dopedZnO films”, Materials Science and Technology, Vol.:23, pp.108-112, 2007.

6. C. E. Kim, P. Moon, I. Yun, S. Kim, J. M. Myoung ,H.W. Jang , “ Process estimation and optimized recipesof ZnO:Ga thin film characteristics for transparentelectrode applications”, J. Bang. Expert Syst. Appl.,Vol: 38, pp. 2823_2827, 2011.

7. J. S. Park, J. P. Kim, Y. R. Noh, K. C. Jo, S. Y. Lee, H.Y. Choi, J. U. Kim, “X-ray images obtained from coldcathodes using carbon nanotubes coated with gallium-doped zinc oxide thin films”, Thin Solid Films, Vol:519, pp.1743_1748, 2010.

8. S. J. Henley, M. N. R. Ashfold, D. Cherns, “The growthof transparent conducting ZnO films by pulsed laserablation”, Surf. Coat. Technol., Vol: 177, pp.271_276,2004.

9. J. Hu, R. G. Gordon, “Textured aluminum-doped zincoxide thin films from atmospheric pressure chemical-vapor deposition”, J. Appl. Phys., Vol:71, pp. 880,1992.

10. H. H. Shin, Y. H. Jong, S. J. Kang, “Influence of thesubstrate temperature on the optical and electricalproperties of Ga-doped ZnO thin films fabricated bypulsed laser deposition”, J. Mater. Sci: Mater. Electron,Vol: 20, pp. 704_708, 2009.

11. P. K. Nayak, J. Yang, J. Kim, S. Chung, J. Jeong, C.Lee and Y. Hong, “Spin-coated Ga-doped ZnOtransparent conducting thin films for organic light-emitting diodes”, J. Phys. D: Appl..Phys., Vol: 42,pp.035102, 2009.

12. A. R. Babar, P. R. Deshamukh, R. J. Deokate, D.Haranath, C. H. Bhosale, K.Y. Rajpure, “Galliumdoping in transparent conductive ZnO thin filmsprepared by chemical spray pyrolysis”, J. Phys. D:Appl. Phys., Vol: 41, pp.135404, 2008.

13. M. C. Jun, S. U. Park and J.H. Koh, “Comparativestudies of Al-doped ZnO and Ga-doped ZnOtransparent conducting oxide thin films”, NanoscaleResearch letters, Vol: 7, pp.639, 2012.

14. J. Zhou and Z. Y. Zhong, “Structural and optoelectricalproperties of Ga‐doped ZnO semiconductor thin filmsgrown by magnetron sputtering”, Cryst. Res. Technol.,Vol: 47, pp.944, 2012.

15. D. Raoufi, T. Raoufi., “The effect of heat treatment onthe physical properties of sol–gel derived ZnO thinfilms”, Appl. Sur. Sci., Vol: 255, pp. 5812_5817, 2009.

12 SAHYADRI INTERNATIONAL JOURNAL OF RESEARCH, VOL 4, ISSUE 2, 2018


EEG based classification for Alcoholism Nishitha Lakshmi, Rani Adhaduk, , Nidarsh Nithyananda, S Rashwin Nonda and Pushpalatha K *

Department of Computer Science and Engineering, Sahyadri College of Engineering & Management, Mangaluru-575007

*E-mail:[email protected]

ABSTRACT Alcoholism is a tendency to continually rely on alcohol. Unchecked ingestion leads to gradual deteriorating mental health of the abusers. To study the changes in brain activity, Electroencephalography (EEG) is one of the acute and low cost methods. In this study, a sampling rate of 16 is experimented on the input EEG signals. The sampled data is extracted for features using mean. The best result obtained is an accuracy of 89.49% with Support Vector Machine (SVM) classifier using linear kernel.

Keywords: Alcoholism, Artificial Neural Network, Cosine Similarity, EEG Signals, K-Nearest Neighbors, Mean, Support Vector Machine

1 INTRODUCTION Alcohol is usually considered a `social beverage'. Unfortunately, the abuse of alcohol has caused societal rot. Alcoholism dulls brain activity and causes muddled physical reactions. Innumerable cases are recorded of liver cirrhosis deaths and many of the road traffic crash deaths caused by uncontrolled alcohol consumption. A healthy society refers to the perfect state of being wholly well in terms of the physical, emotional, spiritual aspects that surrounds the well-being of the existence of the society. This state of serenity of a society is disrupted when crime and diseases run rampant. As humans are social beings, these transgressions cause hindrance in the progress of a dynamic society and negatively affect mental health. Brain signal study is ideal to obtain the insights about mental health and activities as they are the blueprint of all nervous activity. Electroencephalography (EEG) is one of the most effective brain signal recording system that helps to study the brain activities. EEG is a neuro-electricity activity that collects information in bulk that represents the psychological and physiological state of the human body by a conductive medium. Out of norm EEG signals represent irregular brain activities. By analyzing these EEG signals, we gain an understanding about occurrences of abnormal brain activity which plays an important role in diagnosis of different mental disorders like epilepsy, schizophrenia, addiction. EEG signals are also employed in criminal psychology studies where abnormal brain waves are significant in analyzing the violent crime patterns, lie detection and deception. In this paper, we provide performance analysis on various classification algorithms on our data set to diagnose abnormal brain activity and classify input EEG signals into control or alcoholic groups. This paper is structured as follows. Section II recounts the related work published on EEG signal analysis. Section III illustrates our proposed work. Section IV shows experiments

undertaken and results obtained. The concluding remarks for the proposed system are given in section V.

2 Related Work Over the past few years, EEG has gained popularity in BCI or Brain-Computer Interface applications. EEG has been successfully used for diagnosis and treatment of mental abnormalities, brain-neuro-degenerative diseases and criminology studies. EEG can also be used to diagnose numerous neurological disorders such as dementia, brain tumors [15], Parkinson’s disease, Alzheimer’s and countless others.

With the advancement of technologies, the crime scenes also require a newer method wherein investigators can use modern ways to look for clues which might not be visible to the traditional method of collecting evidence. Based on the results that were published by the researchers on this topic, application of EEG significantly improves the accuracy of criminal identification to over 70%. This proves that EEG can also be used in the forensics discipline [1]. Yasmeen et.al., [21] suggested a model which analyzes EEG signal in seizure detection utilizing wavelet transform with statistical parameters. They took two different data sets, each having a different sampling rate. One had 128Hz whilst the other had 1024 Hz. Using discrete wavelet transformation, feature extraction was done, and a multi-layered neural network was used to differentiate between normal brain signals and brain signals indicating seizures.

It is proven true that the emotions are affected by the alcohol intake and influence adverse effects on human abilities to think, act and behave. These progressions or harms appear in the brainwave recording of an EEG. The target behind this research [20] was to demonstrate the unfavorable impacts of alcohol on


the brain. The EEG signals were denoised utilizing Independent Component Analysis and classified using Probability Neural Network to successfully recognize the brain signals influenced by alcohol.

Support vector machine [10]-[14], artificial neural networks [18,9,5,19], k-nearest neighbors [16,17] and many other machine learning methods [2]-[9] are a wide assortment of classification methods for seizure prediction and other disorders.

Therefore, from perusal of the related works we perceive that EEG recordings contain significant information regarding mental activities and are ideal in diagnosis of mental and chronic disorders. We also studied various feature classification algorithms suitable for EEG signals, which are employed in our experimental study. The goal of this paper is to obtain the most efficient classifier by comparing the accuracy rates resulted by experimental procedures undertaken.

3 Proposed Study To analyze the performance of the classifiers, following phases are employed i.e. sampling, feature extraction with mean. Sampling allows the reduction in dimension of our large data set. The sampled data set with the optimal sampling rate is extracted for features using mean. These extracted features are fed as training data for various classifiers and their performance is analyzed.

3.1 EEG Data Brain cells communicate with each other by sending messages as electrical impulses. Transportation of these impulses from one neuron to another, causes ionic drift between the two, which is recorded by the electrodes in electroencephalogram. Based on the frequency of EEG signals resulting due to conscious and sub-conscious brain activities, EEG signals are divided into four categorical wave forms i.e. alpha, beta, theta and delta. Delta(<4Hz) is found in the adults under slow-wave sleep. Theta (4 - 7 Hz) associated with weariness in teenagers and adults. Alpha (8 - 15Hz) wave form represents relaxed state in the adults and is associated with inhibition control. Beta (16 - 31Hz) is associated with state of high computation and alertness [22].3.2 Feature ExtractionFeature extraction methods distill the sampled data for its characteristic attributes. Features are the representative parameters of input sampled patterns that facilitate differentiating between the samples of input feature patterns.

To analyze a representative subset of data points and identify the feature patterns, we employ the sampling method. In statistics, sampling is a process of selection of a subset of individuals from a group of observations (called population) to represent the whole population. Sampling also brings about

reduction of the dimension of our data set, making it simpler for ease of interpretation.

Mean: Mean is a scale of central tendency, which ascertains the feature around which central grouping occurs. We calculate the Mean value of a sample using the equation 1.

μ = 1𝑛𝑛∑ 𝑥𝑥i𝑛𝑛

𝑖𝑖=1 (1)

where xi is sample value and n is the number of samples.

3.3 Classification We aim to compare the performance of classifiers SVM, ANN, KNN and Cosine Similarity, thus obtain the most favorable classifier on our classifier. Extracted features are fed into classifiers to classify the abnormal EEG signals and finally sort, if the input EEG signals belong to the control group or alcoholic group.

3.3.1 Support Vector Machine Collection of affiliated supervised learning techniques, proposed for classification. SVM maps input vector to a higher dimensional space where a maximal separating hyperplane is constructed. An SVM outputs a map of the sorted data with the margins between the two as far apart as possible using kernel functions. The function of kernel is to take data as input and transform it into the required form. SVM has many kernel functions and the ones used in our study are as follows:

Sigmoid Kernel: K (xi, xj) = tanh (γ𝑥𝑥𝑖𝑖𝑇𝑇 x𝑗𝑗 + r)

RBF Kernel: K (xi, xj) = exp (-γ|𝑥𝑥𝑖𝑖 − 𝑥𝑥𝑥𝑥|2),𝛾𝛾 >0 (3)

Linear kernel: K (xi, xj) = xi

T xj (4)

Polynomial Kernel: K (xi, xj) = �γ𝑥𝑥𝑖𝑖𝑇𝑇𝑥𝑥𝑗𝑗 + 𝑟𝑟�d, 𝛾𝛾 > 0 (5)

𝑛𝑛

where xi and xj training vectors xT is the transpose of training vector, k (xi, xj) is a kernel function and ᵞ, r, d are the kernel parameters.

3.3.2 K-Nearest Neighbors A non-parametric approach, which classifies the training data point according to majority of its nearest neighbors. Performance of KNN depends on the number of nearest neighbor values i.e. k = 2, 3, 4, 5. To classify the neighbor, we made use of Minkowski metric distance as given in equation 6. D(𝑋𝑋, 𝑌𝑌) = ( ∑𝑖𝑖 =1|𝑥𝑥𝑖𝑖 − 𝑦𝑦𝑖𝑖 |p)1/p (6)


where X = (x1,x2,...,xn), Y= (y1,y2,...yn ) and p is the parameter measure.

3.3.3 Artificial Neural Network A framework of machine learning algorithms motivated by the working of biological neural networks. It consists of single or more layers to perform complex tasks. Each layer is made up of artificial neurons (nodes) and node connectors called edges. Artificial Neural Network comprises of three layers: input layer, hidden layer and the output layer. The inputs to the artificial neural network are stored in the artificial neurons of the input layer along with its respective weights, which is assigned based on its relative importance. The output of each artificial neurons is obtained only when a non-linear function called activation function is triggered by the weighted sum of its inputs, provides a result exceeding the threshold value.

Activation functions utilized in this experiment are:

ReLU: Rectified Linear Unit, takes a real-valued input and calculates a threshold at zero (replaces negative values with zero) using the equation 7. 𝑓𝑓(𝑥𝑥) = 𝑚𝑚𝑚𝑚𝑥𝑥(0, 𝑥𝑥) (7)

where x is a real sample value.

Tanh: It takes a real-valued input and transforms it into the range [-1, 1] by using the equation 8.

tanh (x) = 2�1+𝑒𝑒𝑒𝑒𝑒𝑒(−2𝑒𝑒)�

- 1 (8)


Sigmoid: takes a real-valued input and transforms it into the range [ 0, -1] by using the equation 9.

𝜎𝜎(𝑥𝑥) = 11+𝑒𝑒𝑒𝑒𝑒𝑒(−𝑒𝑒)

(9)


3.3.4 Cosine Similarity A similarity measure which is obtained by observing the cosine angle between the feature vectors. The outcome of the cosine similarity is bound in [0,1] where similarity score of 1 denotes vectors having same orientation and similarity score of 0 denoting vectors being relatively oriented at 90 degree. Cosine similarity then gives a useful measure of how similar two signals are likely to be in terms of their label i.e. alcohol or control using the following equation 10.

cos 𝜙𝜙 = 𝐴𝐴.𝐵𝐵|𝐴𝐴|.|𝐵𝐵|

(10)

A and B are sample vectors of EEG data, |A| and |B| represent the cardinality of two vectors.

4 Experimental ResultsWe experimented using HP Pavilion 15-au007tx with theconfigurations of RAM - 8GB DDR4, CPU - Intel i5 6thgeneration. The coding for the experiment were done inPython2 language. We have led our analyses by taking the data set from an open source, contributed by Henri Begleiter at the Neurodynamics Laboratory at the State University of New York Health Center at Brooklyn [23]. Placement of the electrodes was done in the internationally accepted 10-20 method. Data set comprises of EEG signal recordings from 64 electrodes from each subject which were sampled at the rate of 256 Hz for a second with 30 trials each. Each subject belonged to either one of the groups: control or alcoholic. Training and testing data comprise of 468 trials and 480 trials respectively. Training and testing trials were randomly combined in the ratio of 80:20.

Table 1. Accuracy values of Classifiers with Mean Classifier Kernel/Activation

Function Accuracy

SVM Linear 89.47 RBF 62.45

Sigmoid 52.11 Polynomial 73.68

ANN Logistic 86.84 Relu 85.78 Tanh 85.26

Identity 84.36 KNN K=2 64.09

K=3 66.31 K=4 59.47 K=5 62.10

Cosine Similarity - 68.21

4.1 Performance evaluation of classifiers with sampling rate 16

With sampling rate of 16 on an electrode, we get one feature from every 16 feature samples. Hence from 64 electrodes we get 16 X 64 = 1024 values, where the input data of size (948 X 1024) is fed to the classifiers SVM, ANN, KNN and Cosine Similarity. Following results are obtained.

We experiment with the SVM classifier and analyze the performance of four of its kernel functions. Therefore, from our analysis, we see in Table 1 that the linear kernel function on SVM gives an accuracy of 89.47%. Performance of ANN is analyzed with various activation functions where logistic function gives an accuracy of 86.84%, as seen in Table 1. When experimented with 2, 3, 4 and 5 neighbors of KNN classifier, it


is evident that 3 nearest neighbors are suitable for our data set, giving an accuracy of 66.31ᵞᵞ%. The data set is lastly analyzed for the cosine similarity method, which gives an accuracy of 68.21%. Hence, we conclude that the best performance on our experimental input data is given by linear kernel of SVM with an accuracy of 89.47%.

4.2 Comparison of computational time taken by EEG classifiers From the previous results obtained, as seen in figure 1, SVM gives the highest accuracy score of 89.47% and the second highest score i.e. 86.84% is given by ANN with logistic function. Hence, we compare the computational time for these two classifiers. From the Table 2, it is observed that SVM takes the minimal time of 0.74 seconds. Therefore, we conclude that SVM classifier is the most efficiently suited classifier for our data set.

Table 2. Computational Time taken by Classifiers in seconds

Classifiers Time (s) ANN 5.11 SVM 0.74

5 Conclusion Alcoholism is an addiction risen with a need for reliance on a substance to make crucial decisions and to cope with events. Alcoholism predominantly dulls brain activity and triggers numerous societal illnesses. The aim of this paper is to analyze EEG signals and classify them as belonging to control or alcoholic group, efficiently. We conducted performance analysis of SVM, ANN, KNN, Cosine Similarity classifiers on our data set. Performance of the classifiers is checked for sampling rate of 16 and mean aids for feature extraction. Therefore, SVM classifier with linear kernel function, obtains the highest accuracy of 89.47%.

REFERENCES [1] Chan, Hsin-Te, et al. `Àpplying EEG in criminal

identification research." 2017 InternationalConference on Applied System Innovation (ICASI).IEEE, 2017.

[2] Adeli, Hojjat, Ziqin Zhou, and Nahid Dadmehr.`Ànalysis of EEG records in an epileptic patient usingwavelet transform." Journal of neuroscience methods123.1 (2003): 69-87.

[3] Harikumar, R., and P. Sunil Kumar. ``Dimensionalityreduction techniques for processing epilepticencephalographic signals." Biomedical andPharmacology Journal 8.1 (2015): 103-106.

[4] Shahid, Arslan, et al. `Èpileptic seizure detectionusing the singular values of EEG signals." 2013 ICMEInternational Conference on Complex MedicalEngineering. IEEE, 2013.

[5] Ghosh-Dastidar, Samanwoy, Hojjat Adeli, and NahidDadmehr. ``Principal component analysis-enhancedcosine radial basis function neural network for robustepilepsy and seizure detection". IEEE Transactions onBiomedical Engineering 55.2 (2008): 512-518.

[6] Gandhi, Tapan, et al. `Èxpert model for detection ofepileptic activity in EEG signature". Expert Systemswith Applications 37.4 (2010): 3513-3520.

[7] Supriya, Supriya, et al. ``Weighted visibility graphwith complex network features in the detection ofepilepsy". IEEE Access 4 (2016): 6554-6566.

[8] Ahmadi, Amirmasoud, Vahid Shalchyan, andMohammad Reza Daliri. `À new method for epilepticseizure classification in EEG using adapted waveletpackets". 2017

[9] Srinivasan, Vairavan, Chikkannan Eswaran, andNatarajan Sriraam. `Àpproximate entropy-basedepileptic EEG detection using artificial neuralnetworks". IEEE Transactions on informationTechnology in Biomedicine 11.3 (2007): 288-295.

[10] Bayram, K. Sercan, M. Ayyuce Kızrak, andBulent Bolat. ``Classification of EEG signals by usingsupport vector machines." 2013 IEEE INISTA. IEEE,2013.

[11] Bhuvaneswari, P., and J. Satheesh Kumar.``Support vector machine technique for EEG signals".International Journal of Computer Applications 63.13(2013).

[12] Zavar, M., et al. `Èvolutionary modelselection in a wavelet-based support vector machinefor automated seizure detection". Expert Systems withApplications 38.9 (2011): 10751-10758.

[13] Guler, Inan, and Elif Derya Ubeyli.``Multiclass support vector machines for EEG-signalsclassification". IEEE Transactions on InformationTechnology in Biomedicine 11.2 (2007): 117-126.

[14] Subasi, Abdulhamit, and M. Ismail Gursoy.


`ÈEG signal classification using PCA, ICA, LDA and support vector machines". Expert systems with applications 37.12 (2010): 8659-8666.

[15] Murugesan, M., and R. Sukanesh. ``Towardsdetection of brain tumor in electroencephalogramsignals using support vector machines".InternationalJournal of Computer Theory and Engineering 1.5(2009): 622.

[16] Huang, Jie, et al. `Àn Improved kNN Basedon Class Contribution and Feature Weighting." 201810th International Conference on MeasuringTechnology and Mechatronics Automation(ICMTMA). IEEE, 2018.

[17] Bablani, Annushree, Damodar Reddy Edla,and Shubham Dodia. ``Classification of EEG Datausing k-Nearest Neighbor approach for ConcealedInformation Test".Procedia computer science 143(2018): 242-249.

[18] Guler, Nihal Fatma, Elif Derya Ubeyli, andInan Guler. ``Recurrent neural networks employingLyapunov exponents for EEG signalsclassification".Expert systems with applications 29.3(2005): 506-514.

[19] Classification Technique Based on ANN forEEG Signals". IJCSIT (2014).

[20] Rachman, Nurindah Tiffani, HandayaniTjandrasa, and Chastine Fatichah. `Àlcoholismclassification based on EEG data using IndependentComponent Analysis (ICA), Wavelet de-noising andProbabilistic Neural Network (PNN)".2016International Seminar on Intelligent Technology andIts Applications (ISITIA). IEEE, 2016.

[21] Yasmeen, Shaguftha, and Maya V. Karki.``Neural network classification of EEG signal for thedetection of seizure". 2017 2nd IEEE InternationalConference on Recent Trends in Electronics,Information & Communication Technology(RTEICT). IEEE, 2017.

[22] Kirmizi-Alsan, Elif, et al.``Comparativeanalysis of event-related potentials during Go/NoGoand CPT: decomposition of electrophysiologicalmarkers of response inhibition and sustainedattention." Brain research 1104.1 (2006): 114-128.

[23] S. U. O. N. Y. H. C. NeurodynamicsLaboratory, `ÙCI Machine Learning Repository," 13October 1999. [Online]. Available:https://archive.ics.uci.edu/ml/datasets/EEG+Database.


Some Classes of Diophantine Edge Graceful graph ABSTRACT In this paper we consider only finite graphs. A graph G (V, E) is said to be Diophantine edge graceful graph, if there exists a bijection such that the labels of every set of m edges satisfy a linear Diophantine equation. In [2], we constructed Diophantine equations and used the solutions for edge graceful trees of the form (m, h) trees. In this paper we extend the results of [2] to Ramanujan graphs. A d-regular graph is called Ramanujan graph, if the absolute value of second highest adjacency eigen value is bounded above by 2. A graph is graceful if its m edges are labelled by a unique positive integer from 1 to m consecutively. A major unproven-conjecture in graph theory is the graceful tree conjecture or Ringel-Kotzig conjecture, which hypothesizes that all trees are graceful. For the class of complete (m, h) trees, the authors [2] proved that they are Diophantine edge graceful. The main results of this paper are Diophantine edge graceful graph of complete graphs kn ( n odd) and the n-prism graphs. It is interesting to observe that these two classes of graphs are also Ramanujan graphs and the solutions of the corresponding Diophantine equations again yield edge-graceful graphs. Keywords: Diophantine edge graceful graph, Ramanujan graph .

Consider a graph , with and If is a bijective mapping and if be defined by

for , and if bijective, then the induced map gives an edge

graceful labeling. A graph G (V, E) is said to be Diophantine edge graceful graph, if there exists a bijection such that the labels of every set of m edges satisfy a linear Diophantine equation. In this paper, we for any given graph G (may be complete graph, n –prism graph or, a Ramanujan graph), we construct a specific Diophantine equation and solve it and label the vertices consecutively as a graceful Graph. If there are m edges in G, then edges are labelled from 1, 2, 3, ..., m . The labelling gives an edge graceful Graph. Definition : Let G be a simple graph. Let A(G) = adjacency matrix of G = { (i, j) = 1, if ij is an edge and (i,j) = 0, otherwise.}. If A (G) is real symmetric, then sum of eigen values becomes zero, if not average of eigen values be non-zero, say m. Definition : Let G be a simple graph, then Energy of G = E (G) = sum of absolute differences between an eigen value and the average m of the eigen values.

But if the average is zero, in the case of A(G) is a real symmetric Matrix, then E(G)=sum of absolute values of eigen values of A(G), because in this case, m=0.. We observe that every complete graph Kn is (n-1)-regular graph. If G is a d-regular graph, then it is called a Ramanujan graph. If L is the modulus of the highest eigen value of A (G) = adjacency matrix of G and if L ≤ . Also the n-prism graph G is a 3-regular graph. The class of generalised Petersen graphs, n-Petersen graph, for n= odd positive integer, turns out to be Ramanujan Graphs and also have minimal energy. In section 1, we get results concerning those graphs that are Ramanujan graphs which admit Diophantine edge graceful labelling this includes the explicit construction of the Diophantine equations, relevant to the chosen graph G. The interesting feature is that the solutions of Diophantine equations are labelled for the different Hamiltonian cycles in G. Theorem 1, concerns the complete graphs

for odd Theorem 2, concerns the technique of Diophantine edge labelling of all n-prisms. In section 2, we study Diophantine edge graceful labelling of Generalised n-Petersen graphs. If G is an n-prism and if P is a Generalised Petersen graph P, the special property: energy of (P) ≤ Energy (G), where G is any n-prism is given in theorem 3.

In section3: theorem 4 proves edge-gracefulness for the complete bipartite graph , that are not necessarily Ramanujan graphs and we have also results on Wheel graphs proved.


1. INTRODUCTION


2. SECTION ONE In this section , we consider the Diophantine edge graceful graph which is also Ramunajan graphs. Definition: A complete digraph is a directed graph in which every pair of distinct vertices is connected by a pair of unique edges . THEOREM 1: Every complete graph Kn (n odd) is a Diophantine edge graceful. The Diophantine equation satisfied by Hamiltonian cycles of Kn is -------------------- (1) Where s is the number of Hamiltonian cycles in Kn (n odd). Proof: We know that for a complete graph Kn (n odd), we have (n-1)/2 Hamiltonian cycles. Consider a complete graph Kn, with n=2k+1, k >0 vertices. Then it will have k number of Hamiltonian cycles. We have to prove that every k Hamiltonian cycles with n vertices satisfy the Diophantine equation (1). We observe that in every Kn, with k Hamiltonian cycles have nk edges. We divide nk edges into k number of equation s having n variables. Also we multiply 2 to every first variables so that each k equation with n edges satisfy (1) . (i.e.) will satisfy the Diophantine equation as follows =

= = = = = = = = . Hence the theorem.

Example 1:

In K7, we have 3 Hamiltonian cycles. Each Hamiltonian cycle satisfy the Diophantine equation: Here s=3, n=7 then 7x32 + (3+1)2=79 OBSERVATIONS: 1. Every complete graph Kn (n odd), the Diophantine equation can be given in the form 2. For every complete graph Kn (n odd), the Diophantine equation also satisfy . Definition:A prism graph is a graph that has one of the prisms as its skeleton. THEOREM 2: Every n-prism graph is Diophantine edge graceful; the n graceful edge labels satisfy the Diophantine equation (i)When n is even: (ii)When n is odd:

Proof is similar to theorem 1 Example: Consider the graph n-Prism for n=4 and n=5

https://en.wikipedia.org/wiki/Directed_graph

https://en.wikipedia.org/wiki/Graph_(discrete_mathematics)

https://en.wikipedia.org/wiki/Prism_(geometry)

SAHYADRI INTERNATIONAL JOURNAL OF RESEARCH, VOL 4, ISSUE 2, 2018 19 In fig 2(a), the Diophantine equation satisfied by the set of 4 edges is . Also, in fig 2(b), the Diophantine equations satisfied by the set of 5 edges are . GENERALISED PETERSEN GRAPHS: The Petersen graph is an undirected graph with 10 vertices and 15 edges. It is a is the complement of the line graph of . The generalized Petersen graphs are a family of cubic graphs formed by connecting the vertices of a regular polygon to the corresponding vertices of a star polygon. They include the Petersen graph and generalize one of the ways of constructing the Petersen graph. The generalized Petersen graph family was introduced in 1950 by H. S. M. Coxeter and was given its name in 1969 by Mark Watkins.

We observe that Generalised Petersen graphs satisfy the same Diophantine Equations of n- Prisms for each n. For n=5 which is the Petersen Graph satisfy the Diophantine equation same as in example 2(b): .

If X is a connected k-regular graph, we may arrange the Eigen values as

It is not difficult to show that −k is an eigen value of X if and only if X is bipartite in which case its multiplicity is again equal to the number of connected components. Any eigen value is referred to as a nontrivial eigen value. The maximum of the absolute values of all the non-trivial eigen values is denoted as λ(X). A Ramanujan multigraph is a k-regular graph satisfying

. ------ (1) Considering the above graphs, i.e. Complete graphs, n-prism graphs and generalized Petersen graphs, we observe that all these graphs are regular and satisfy the condition (1). So all these graphs are also Ramanujan Graphs. Considering the above graphs, we observe that all the graphs satisfy the conditions of Ramanujan Graphs. If G is any simple graph, then Energy(G)= sum of absolute values of eigen values of A(G) = the adjacency matrix of G = {(i ,j) = 1, if ( i, j) is an edge in G, and (i, j) = 0, otherwise}.review the main points of the paper, do not replicate the abstract as the conclusion. A conclusion might elaborate on the importance of the work or suggest applications and extensions. THEOREM 3: For n a positive odd integer, the n-prism and the Generalised n-Petersen graphs are both 3-regular graphs and they also satisfy the conditions of Ramanujan graphs. If G is an-n prism and P is n-Petersen graph, then Energy (G)>= Energy (P).The n-Petersen graph has minimal energy for n=3, 5, 7, 9... Example: Let G= a 7-prism and P = a 7-Petersen graph. Let A(G) , A(P) are the adjacency matrices of G and P respectively :then A(G)= [ ... ] and A(P)=[...] be 14x14 adjacency matrices. Then the eig(A(G)) = spec(A(G))=( -2.8017 -1.445 -0.8019 0.247 0.555 1 2.247 3) each with multiplicity 2 , ,except 1 and 3 occurring once. Evidently, the 7-prism is 3-regular and satisfies the conditions of a Ramanujan Graph. Since average of absolute values of eigen values = 1.4426 = non-zero and Energy(A(G)) = sum(abs((eig(A(G))-1.4426) ) = 26.5257. On the other hand, Eig (P-7petersen) = {3,-2.3319, -2.3028, -2.1007, -2.0000, -0.9089, -0.6180, 0.0000, 0.849, 1,0000, 1.3028, 1.7108, 1.5180, 1.5457} and average of

https://en.wikipedia.org/wiki/Undirected_graph

https://en.wikipedia.org/wiki/Vertex_(graph_theory)

https://en.wikipedia.org/wiki/Edge_(graph_theory)

https://en.wikipedia.org/wiki/Complement_graph

https://en.wikipedia.org/wiki/Line_graph

https://en.wikipedia.org/wiki/Line_graph

https://en.wikipedia.org/wiki/Cubic_graph

https://en.wikipedia.org/wiki/Regular_polygon

https://en.wikipedia.org/wiki/Regular_polygon

https://en.wikipedia.org/wiki/Star_polygon

https://en.wikipedia.org/wiki/Petersen_graph

https://en.wikipedia.org/wiki/H._S._M._Coxeter

20 SAHYADRI INTERNATIONAL JOURNAL OF RESEARCH, VOL 4, ISSUE 2, 2018 absolute values of eigen values = 1.455 and Energy =Sum(abs(A(P)-1.466))=24.5451..and E(A(P)) = 4.5451< E(G) = 26.5257. Thus Energy (Petersen-7 prism) < Energy (7-Prism).and so the Petersen-7 prism has LEAST Energy among 7-Prisms. In general, Energy of an n-Petersen graph is minimum among that of a general n-prism, for n=3, 5, 7, 9, .... 3. SECTION TWO: In this section, we consider the graphs which are only Diophantine edge graceful graphs. Definition: A complete bipartite graph is a graph whose vertices can be partitioned into two subsets V1 and V2 such that no edge has both endpoints in the same subset, and every possible edge that could connect vertices in different subsets is part of the graph. That is, it is a bipartite graph (V1, V2, E) such that for every two vertices v1 ∈ V1 and v2 ∈ V2, v1v2 is an edge in E. A complete bipartite graph with partitions of size |V1|=m and |V2|=n, is denoted Km,n THEOREM 4: Every complete bipartite graph is an edge graceful graph: The Diophantine equation satisfied by the mn edges fall in one of the three cases

(i) If m is even, then n equations with m edges satisfy

(ii) If n is odd then m equations with n edges satisfy

(iii) If both m and n are odd, then m equations with n edges satisfy , m > 1.

Remark: The Complete Bipartite graph is not Ramanujan Graph in general, but if m=n ,the graph is a Ramanujan Graph in trivial sense. Definition: A wheel graph is a graph formed by connecting a single universal vertex to all vertices of a cycle. Theorem 4:

All Wheel graphs are Diophantine edge graceful graphs. The Diophantine equation satisfied by the n edges is:

(a) For n even, we have

(b) For n odd, we have

Fill the text from your manuscript in different sections. Fill the text from your manuscript in different sections. Fill the text from your manuscript in different sections. Fill the text from your manuscript in different sections. Fill the text from your manuscript in different sections. CONCLUSION: Labelled graph has many applications due to which it one of the topics of current interest. Here we have given five classes of graphs which are Diophantine edge graceful. This can be carried out for other classes of Graph also. ACKNOWLEDGEMENT: Mrs. Sunita D’Silva is grateful to Dr.S.A.Mariadoss for suggesting the topic and guiding the research work. The authors are grateful to the management for encouraging research in the department. REFERENCES

[1] Gallian Joseph A. (1998),”A Dynamic Survey of Graph Labelling”, Electronic Journal of Combinatorics.

[2] A.Mariadoss and Sunita,” Diophantine edge graceful graph” Recent Patents on Computer Science. Volume 9, 3 Issues, 2016

https://en.wikipedia.org/wiki/Bipartite_graph

https://en.wikipedia.org/wiki/Universal_vertex

https://en.wikipedia.org/wiki/Cycle_graph


[3] B. Gayathri, S. Kousalya Devi, k-even edge-graceful labelling of the graph < K1, n: K1, m>, Bharathidasan University Journal of Science and Technology, in press

[4] M. Ram Murty, Ramanujan Graphs, J. Ramanujan Math. Soc. 18, No.1 (2003) 1–20

[5] Gayathri. B and Subbiah. S - Strong edge graceful labeling of some trees presented in the National Conference at Jamal Mohamed College, Trichy On March 27 -28 (2008).


ISSN: 2456-186X, Published Online Frbruary, 2019 (http://www.sijr.in/)

22

Discrete Cosine Transform Coefficients for Kannada Hand-Written Character Recognition

Pateel G P1*, Sunil Kumar P2, Megha N3 1-3 Electronics and Comm. Dept., Sahyadri College of Engineering & Management, Adyar, Mangaluru-575007

*E-mail:[email protected]

ABSTRACT An offline handwritten Kannada word recognition system using Support Vector Machine (SVM) as classifier is described in this paper. The character recognition system generally involve three major steps viz, preprocessing, feature extraction and classification. In our work in the preprocessing section some of the image processing techniques such as RGB to gray conversion, Binerization, Line segmentation and character segmentation of scanned document are implemented. In the feature extraction section Discrete Cosine Transform coefficients are generated and used to as feature vectors, Later these features (Discrete Cosine Transform) given as inputs to Support Vector Machine (SVM) classifier individually. There by we obtained results. In order to evaluate the performance of our proposed Optical Character Recognition (OCR) system, 1050 samples of Kannada alphabets written by various people in various styles are made used. Part of this data set is used to train the SVM and remaining part is used to test the performance of SVM. We achieved satisfactory recognition rate of around 86%.

Keywords: Keywords—Image processing, feature extraction and SVM classification

1. INTRODUCTIONCharacter recognition is method of recognizing characters

from scanned image and converts it into American Standard Code for Information exchange or alternative equivalent machine editable form. This improves the interface between man and machine in numerous applications. In future days, Kannada character recognition system would possibly functions a key issue to form paperless atmosphere by digitizing and process existing paper documents. This method presents an innovative technique to acknowledge written Character. This method will be classified in 2 classes.1) Offline character recognition—method 2) Online character recognition--method

The offline character recognition method will more split into holistic segmentation approach. In holistic approach word is treated as whole and processed however in segmentation approach every character is separated then processed, in offline recognition method, document is initial created, digitized, hold on in pc then it's processed. Just in case of on--line character identification method, characters square measure processed whereas it's beneath creation. Change of manually written characters is critical for making a few imperative records identified with our history, for example, original copies, into machine editable shape so it can be effectively gotten to and saved. To lessen the exercise in futility associated with composing articles e.g. Kannada daily

papers. Valuable under tight restraints processing in banks, all sort of shape preparing frameworks, written by hand post address determination and some more.

Kannada is the official dialect of Karnataka,, More than thirty million individuals talk Kannada as their primary language . Around eleven million individuals utilize Kannada as the second dialect. Kannada has got its own content derived from Brahmi content. Kannada has a base arrangement of 49 Characters. They are ordered into three categories: Swara (vowels), Vyanjana (consonants), and Yogavahakas. There are 13 vowels, 34 consonants and 2 Yogavahakas. The Fig 1.1 represents the Kannada varnamale.


25

2. METHODOLOGY

Fig.1: Methodology

2.1 Image Pre-Processing When an image is fed into the MATLAB handwriting recognition system from the scanner, it is vital to process the image using standard image-processing techniques for easy and appropriate data acquisitions. Noise is the most common portion of the image that has to be discriminated and removed. The following preprocessing techniques as shown in Fig. 1 are used to remove noise and extract individual character handwritten data in MATLAB.

2.1.1 Gray Scale of RGB image The need for converting to Grayscale is to reduce the processing time for the algorithms; RGB Images are not required to processing. 2.1.2 Removing of Small Object

The gray image contains some small object, so it remove all small object using BWAREAOPEN operation, this operation will remove all small pixel object and it remove small pixel object based on user need and here in our work it removing all small object whose size less than 50 pixel.

2.1.3 Binarization of image

Binarization converts gray-scale image into binary image. During binarization the gray image pixel values with intensity greater than half of the full intensity will be made as ‘1’, which means white and gray image intensity pixel values with intensity less than half of the full intensity will be made as ‘0’ , which means black.

2.1.4 Inversion Inversion is the process of changing binary image pixel

value 1 to 0 which means white color is changed to black and binary image pixel value 0 to 1 which means black color to white. This process is important in extracting a character efficiently from image if it as only one color which is distinct from the background color.

2.2 Segmentation Segmentation is the process of extraction an individual character from a document this is done in two steps. 1) Line segmentation. 2) Character segmentation. As Kannada is a non cursive script, and individual character in word are isolated. Spacing between the characters can be used for the segmentation. Line segmentation extracts lines from a given image.

Steps to be followed for the line Segmentation is as follows: 1. Scan the image horizontally and identify the non-zero

rows between zero rows. 2. Extract the non-zero row from the image that acts as

line segment. 3. Repeat the step 1 and 2 for the remaining image until

all lines are extracted from the image.

2.3 Feature Extraction The feature extraction is the process of extracting unique-

important properties of an image in the form of feature vector which describes about the characteristics of an image. It is one of the most important components for any recognition system, since the classification/recognition accuracy is depending on the features. Well known and simple and efficient feature extraction method is Discrete Cosine Transform (DCT) features extraction for handwritten basic Kannada characters recognition system is proposed. A brief description about zoning is given below.

For a image f (x, y), its second DCT rework is

outlined as follows:

(1) Where

(2) Sample Discrete Cosine Transform features extracted for few characters are shown in the following table 1.

Sl. No.

1 23.12 20.12 18.34 17.5

2 0.855237

1.455498 -2.4183 -1.9366

Algorithm-for-DCT--FEATURES

S-1: Read-the pre-processed input image.

S-2: Compute--DCT for the input--image (binary image).

S-3: Convert--DCT coefficient matrix into 1d zigzag

column array.

S-4: Choose-the- first- 50- Discrete-Cosine-Transform

(DCT) - coefficients-- as features.


24

3 0.114048

2.501746

2.145956 -1.70152

4 0.324902 -1.5406 3.99131

5 -9.36609

5 0.430636

1.339451

4.524884

0.523709

6 1.480387

2.881773 -3.31648 -3.66928

7 0.739882

0.544732

0.256137 -0.6354

8 6.024383

5.034343

2.005787 -0.21909

9 2.173903

1.552997

1.645417 -1.98539

10 -1.1501 -3.83777 4.097103 -3.2439

11 -2.71565 2.139438 -5.34735 1.14437

3

12 -0.98202 0.803318

2.527517

0.435731

13 -4.43894 -3.73824 -0.62207 1.969878

14 -1.08616 -2.72938 2.482452

3.357671

15 -1.61976 -4.20701 -1.20904 -2.64659

16 -1.77838 3.878677 4.32767 -0.43862

17 -4.10753 -0.40107 0.937606 -0.9497

18 -3.25963 -3.11966 -0.3425 0.45798

19 1.616914

1.280556 -7.10238 0.58376

2

20 -0.30078 -0.00612 -1.62683 -0.94998

21 -3.15011 -1.94019 -0.7803 4.289144

22 -3.69199 -4.08389 -1.51703 -0.05086

23 -0.32347 0.839073 -0.20973 -0.32129

24 -4.47463 -6.69927 4.107965 -0.5547

25 0.711191

0.672846

1.057774

2.453092

MEAN 0.091317 0.16367 0.13885

3 0.19332

7

VAR 18.52608

13.76344 8.74148 11.0985

2 The Table-1 represents the Discrete Cosine

Transform features extracted for four sample characters which show the difference between mean\variance of one character with the other and the difference is enough to get classification accuracy in case of Discrete Cosine Transform feature extraction method.

2.4 Classification and Recognition After the feature extraction, the major task is to make decision to classify the character to which class it belongs. There are various classifiers that can apply in recognition. The most important and more effective classifier is Support Vector Machine (SVM).Support vector machine (SVMs) is a supervised learning method used for classification. Where SVM's are a relatively new learning method used for binary classification. The basic idea is to find a hyper-plane which separates the N-dimensional data perfectly into its two classes. SVM commonly used with linear, polynomial, RBF and sigmoid kernels. A multiclass SVM classification has been used in the proposed system with different kernels of 1) linear, 2) polynomial, 3) RBF, 4) sigmoid and it achieves very high recognition accuracy. The final step is the recognition which is matching the selected class by the SVM with the character and finds the desired character in the Kannada alphabets.

3. EXPERIMENTAL RESULTS AND DISCUSSION The experiments were carried out in Matlab-2015 on a 64-BIT 2.67 GHz INTEL i3 processor, with 2 GB RAM. The dataset consisted of 1050 samples out of which 462 selected samples were used for training and the remaining samples were used for testing. The classification is done using SVM. Fig 3 represents the image sample.

Fig 2 proposed image sample

The following table-2 and 3 gives a summary of the results:

The dataset consisted of 1512 samples out of which

462 selected samples were used-for-training and 1050 selected

samples were-used-for-testing. The classification is done using

SVM.

Experiment 1:

Training samples=462 Test samples=1050

Class Trained samples

Tested samples

Correct classification %

11 25 21 84

11 25 21 84


25

11 25 21 84

11 25 21 84

11 25 24 96

11 25 24 96

11 25 16 64

11 25 18 72

11 25 23 92

11 25 13 52

11 25 14 56

11 25 18 72

11 25 18 72

11 25 20 80

11 25 18 72 11 25 14 56

11 25 23 92

11 25 13 52

11 25 19 76

11 25 16 64

11 25 14 56

11 25 18 72 11 25 12 48 11 25 13 52

11 25 19 76 11 25 15 60

11 25 15 60

11 25 16 64

11 25 14 56

11 25 9 36

11 25 8 32

11 25 19 76

11 25 17 68

11 25 21 84

11 25 2 8

11 25 15 60

11 25 19 76

11 25 14 56

11 25 14 56

11 25 4 16

11 25 20 80

11 25 16 64

Average 75.6190

4762 Table-2

Experiment 2:

The dataset consisted of 1512 samples out of which 1050 selected samples were used-for-training and the remaining samples were-used-for-testing. The classification is done using SVM. The bellow Table-3 represents the percentage of recognition rate of individual characters and the average percentage of recognition is 86%. From the observation of the table 2 and 3 the recognition rate can be improved by using effective feature extraction method and also be improved by using more number of samples for training the SVM.



Tested samples


25 11 11 100

25 11 10 90.9090

25 11 11 100

25 11 11 100

25 11 11 100

25 11 11 100

25 11 11 100

25 11 11 100

25 11 11 100

25 11 10 90.9090

25 11 6 54.5454

25 11 9 81.81818

25 11 7 63.63636

25 11 8 72.72727

25 11 11 100

25 11 8 72.72727




Tested samples


25 11 10 90.90909

25 11 11 100

25 11 10 90.90909

25 11 10 90.90909

25 11 10 90.90909

25 11 10 90.90909

25 11 9 81.81818

25 11 9 81.81818

25 11 11 100

25 11 9 81.81818

25 11 8 72.72727

25 11 9 81.81818

25 11 4 36.36363

25 11 11 100

25 11 11 100

25 11 11 100

25 11 10 90.9090

25 11 6 54.5454

25 11 2 18.18181

25 11 11 100

25 11 10 90.9090

25 11 10 90.9090

25 11 10 90.9090

25 11 11 100

25 11 10 90.9090

25 11 11 100

Average 86.7965 Table-3

4. CONCLUSION

Recognition of individual character from the handwrittendocument using image processing techniques, feature extraction methods and finally Support Vector Machine (SVM) as classifier, is implemented in this paper. The recognition rate is around 86%. This work-was--basically focused on the method that extracts the features efficiently-from a single separated characters image i.e, Discrete Cosine Transform features and also the-method that recognizes the character efficiently i.e, Support Vector Machine (SVM). There by we achieved satisfactory recognition rate for the Kannada hand written words.

The proposed character recognition system of our work can be used to recognize hand written documents of the other languages with suitable modifications. Image de noising and enhancement techniques can be incorporated in the preprocessing section for the degraded image documents.

REFERENCES 1. Rampalli R., Ramkrishnan, Angarai G., “Fusion of

Complementary Online and offline Strategies forrecognition of Handwritten Kannada Characters “Journal of Universal Computer Science, 17(1) . pp 81-93. 2011

2. Niranjan S. K, Vijaya Kumar, Hemantha Kumar“unconstrained handwritten Kannada characterrecognition” International Journal of Database Theory and Application, Vol.2, No. 4, pp 290- 301,2009

3. Ragha, L. R., Sasikumar, M.: “Feature Analysis forHandwritten Kannada Kagunita Recognition “.International Journal of Computer Theory and Engineering, IAC-SIT 3(1), pp. 1793-8201 ,2011

4. Aradhya M.,Niranjana S.K.,Hemantha kumar G.,“Probabilistic Neural Network based Approach forHandwritten Character Recognition” Special Issue of IJCCT, Vol.1 Issue 2,3,4 pp.9- 13,2010

5. B.V. Dhandra, Mallikarjun Hangarge and GururajMukarambi. “A Zone Based Character RecognitionEngine for Kannada and English Scripts”. Elsevier Science Direct, pp. 3292-3299. 2012

6. Kunte Sanjeev R., Sudhaker Samuel. “A simple andefficient optical character recognition system for basicsymbols in printed Kannada text”. Sadhana, Vol. 32, Part 5, pp. 521-533. ,2006

7. G.G. Rajput,Rajeswari Horakeri , “ Zone basedHandwritten Character Recognition using crack code andSVM ” ,International Conference on advances in Computing, Communication and Informatics(ICACC)-2013

8. S.A.Angadi and Sharanabasavaraj.H.Angadi “structuralfeatures for recognition of hand written kannada characterbased on svm” International Journal of ComputerScience, Engineering and Information Technology (IJCSEIT), Vol. 5,No.2, ,pp.25-32 ,April



27

Rajesh A Shetty1*, Dushyanthkumar G L1 and Deepak S A2

1 Mechanical Engineering Dept., Sahyadri College of Engineering & Management, Adyar, Mangaluru-575007 2School of Mechanical Engineering, Reva University, Bengaluru, Karnataka - 560064

*Email: [email protected] ABSTRACT In this paper, a brief review of classical and refined beam and plate theories has been presented. For easy understanding of the concept of beam and plate theories, the discussion has been started with an introduction to the simplest, yet extensively used classical beam and plate theories. Later, the discussion has been extended to the refined or shear deformation beam and plate theories which are considered to be more accurate and analytically more acceptable compared to the classical theories. The equations governing the elementary or classical theories are highly simple, hence are could be discussed in undergraduate and postgraduate courses. Whereas, the displacement functions and governing differential equations connected with refined or shear deformation theory formulation are considerably complex in comparison to the classical theory formulation. The inclination towards refined beam and plate theories is developed due to the difficulty in formulation of thick beams and plates using classical theories. The classical theories could be used only for the investigation of slender beams and thin plates as these theories can capture only the bending deflections. However, in case of thick beams and plates, the contributions from both bending and shear deflections need to be taken into account in the theory formulation. This could be achieved by using refined beam and plate theories. In this regard, a brief review of various well-known refined beam and plate theories has been presented in this paper. Also, a brief discussion on pros and cons of various theories has been presented. Keywords: Beam and Plate Theories, Classical Theories, Refined Theories, Shear Deformation Theories, Bending Deflection, Shear Deflection.

NOMENCLATURE A - Beam cross-sectional area b - Beam width D - Plate flexural rigidity E - Modulus of elasticity G - Shear modulus h - Beam height or plate thickness I - Moment of inertia k - Shear coefficient t - Time variable u, v - In-plane displacements in x & y-directions u0, v0 - Axial displacements in x & y-directions w - Lateral displacement in z-direction wb - Bending component of lateral deflection w ws - Shear component of lateral deflection w w0 - Lateral displacement about mid-plane x, y, z - Cartesian coordinates q - Intensity of transverse distributed load

xφ

- Rotation of the normal about y-axis

- yφ

- Rotation of the normal about y-axis

-

ρ

- Density of the material

La - La-

1. INTRODUCTION Beam Theories: Beams are the structural elements which are primarily designed for withstanding applied transverse loads by developing bending stresses. In practice, beams are the 3-dimensional elements. However, for the analytical modeling purpose, the beams can be considered as one-dimensional elements by taking into consideration of the large longitudinal dimension in comparison to the small cross-sectional dimensions. Most of the beam theory equations are derived by considering beams as one-dimensional elements. The theoretical study of beams using beam theories is an approximate, but easy and effective way of investigation of behavior of beams subjected to transverse loads.

A vast literature is available with regard to the static, vibration and buckling analysis of beams using beam theories. Some of the important theories proposed by researchers for the study of beams are, Euler-Bernoulli Beam Theory (EBT) [1], Timoshenko Beam Theory (TBT) [2], Levinson Beam Theory (LBT [3]), Reddy Beam Theory (RBT) [4] and so on. Euler-

Classical and Refined Beam and Plate Theories: A Brief Technical Review


28

Bernoulli Beam Theory, proposed in 18th century, is a basic or elementary theory in the domain of beam theories. Timoshenko Beam Theory, proposed in 1921, is a first-order shear deformation beam theory. Levinson Beam Theory (1981) and Reddy Beam Theory (1984) are the higher-order shear deformation beam theories. The first-order and higher-order shear deformation beam theories are the refined version of Euler-Bernoulli Beam Theory, which are developed by researchers to overcome the drawbacks posed by the elementary beam theory. A large number of research articles are available on beam analysis using Euler-Bernoulli Beam Theory and other just mentioned first-order and higher-order shear deformation beam theories. Plate Theories: Analogous to beams, plates are also the structural elements which are designed for resisting applied transverse loads by developing bending stresses. The main difference between the beam and plate elements is that, for theoretical study purpose, the beams are considered as one-dimensional elements, whereas, plates are treated as two-dimensional elements. This simplification is possible as, in case of plates the in-plane dimensions are large compared to plate thickness. Hence, while formulating a plate theory, this factor could be considered and the plate equations could be derived by simplifying plate as a two-dimensional element. The theoretical study of plates using plate theories is an approximate way of investigating the static and dynamic behavior of plates.

Classical Plate Theory (CPT) [5], proposed in 19th century, is the basic or elementary theory in the domain of plate theories. Classical Plate Theory is also known as “Kirchhoff-Love Theory.” The refined plate theories or shear deformations plate theories available in the domain of plate theories are, Reissner Plate Theory (1945) [6], Mindlin Plate Theory (1951) [7], Levinson Plate Theory (1980) [8], Reddy Plate Theory (1984) [9], Refined Plate Theory (2001) [10], New First-order Shear Deformation Plate (2007) [11] and so on. Reissner Plate Theory is a stress based theory. The formulation of the theory begins with assumptions of the stress functions. Mindlin Plate Theory and New First-order Shear deformation plate theory are the displacement based first-order shear deformation plate theories. Levinson Plate Theory, Reddy Plate Theory and Refined Plate Theory are the higher-order shear deformation plate theories. The refined plate theories could be used for the plate analysis whenever the results predicted by Classical Plate theory are unsatisfactory. In literature, a vast research articles are available pertaining to the analytical study of static, vibration and buckling behavior of plates using classical and refined plate theories.

In this article, a brief review of various important beam and plate theories has been presented. To begin with, a discussion on classical beam and plate theories has been presented. The drawbacks of Euler-Bernoulli Beam Theory and Classical Plate Theory have been discussed in detail. Also, the importance of shear deformation theories over the classical

theories has been highlighted. Further, a short discussion on formulation part of various beam and plate theories such as displacement functions and governing differential equations has been presented.

2. REVIEW OF BEAM THEORIES

2.1 Euler-Bernoulli Beam Theory Euler-Bernoulli Beam Theory (EBT), also known as “Elementary Beam Theory” is the first beam theory formulated for the analysis of beams. It was first proposed in 18th century. The expressions for displacements of a transversely loaded beam given by EBT are as follows [12, 13]:

dxdwzu 0−= (1)

dydwzv 0−= (2) )(0 xww = (3)

The governing differential equation used for obtaining the transverse deflection of a beam subjected to static bending given by EBT is

)(40

4

xqdx

wdEI = (4)

The governing equation is a fourth-order differential equation with 0w as an unknown. By integrating Eq. (4) and applying the beam end conditions, the value of transverse deflection 0w can be determined. In literature, it has been discussed comprehensively that the deflections predicted by EBT are only accurate, if the beams under consideration are slender beams. In case of short or thick beams, EBT underestimate the deflection values. EBT can only capture the bending deflection. However, the contribution of shear deflection cannot be ignored in case of short or thick beams.

Further, the governing equation for transverse vibrations of a beam given by EBT is

),(20

2

40

4

txqtwA

xwEI =

∂∂

+∂∂ ρ (5)

By considering the effects of rotary inertia, the Eq. (5) can be written as follows:

),(220

4

20

2

40

4

txqtx

wItwA

xwEI =

∂∂∂

+∂∂

+∂∂ ρρ (6)

As discussed earlier, EBT is prone to underestimate the deflections in case of thick beams. This means, EBT assumes beams are stiffer. Hence, the frequencies predicted by EBT are higher. Also, the buckling loads predicted by EBT are also higher. The drawbacks associated with EBT resulted in the development of refined beam theories which could formulate a beam by considering the effects of shear deformation also.

2.2 First-order Shear Deformation Beam Theories To overcome the drawbacks posed by elementary beam theory, a new class beam theories known as, First-order shear deformation beam theories are came into existence. These theories can consider the effects of shear deformation also in the beam formulation in addition to the bending deformation. The primary drawback linked with first-order theories is the


29

difficulty in deciding the correct shear coefficients. The formulation of first-order theories is such that, the theories cannot satisfy the shear stress free surface conditions at the top and bottom surfaces of the beams. To account for this it is necessary to use suitable shear coefficients to slightly improve the results yielded by TBT in comparison to exact theory and other shear deformation beam theories. 2.2.1 Timoshenko Beam Theory Timoshenko Beam Theory (TBT), proposed in 1921, is a highly popular and extensively used first-order shear deformation beam theory. The expressions for displacement of a beam given by TBT are [12],

xzu φ−=

(7)

((7

),( yxww o=

(8) The above displacement expressions contain two unknown

functions. According to TBT, the differential equations governing the beam behavior are given by

0),(20

2

20

2

=−

−

∂∂

−∂∂ txq

xwkGA

twA xφρ

(9)

00

2

2

2

2

=

−∂∂

−∂∂

−∂∂ φφφρ

xwkGA

xEI

tI

(10)

In the above equations, k is the shear coefficient. The value of shear coefficient k is reported in literature for various cross-sections of the beams. The above differential equations need to be solved to obtain two unknown variables. The lateral displacement predicted by TBT involves both the bending deflection as well as the shear deflection. However, the use of shear coefficient is the main drawback linked with the TBT. Also, TBT predicts constant shear stress across the thickness of the beam. Therefore, TBT cannot satisfy the shear stress free surface conditions. These just discussed some of the drawbacks linked with first-order theories led to the development of a new class of theories known as “Higher-order shear deformation theories.”

2.3 Higher-order Shear Deformation Beam Theories

To deal with ambiguity in deciding suitable shear coefficients in case of first-order shear deformation theories, the interest is later switched over to higher-order shear deformation theories. In case of higher-order shear deformation theories, the displacement field of the theories can predict realistic transverse shear stress distribution through the beam height. Also, the displacement field can satisfy zero shear stress surface conditions. Hence, the need for using shear coefficients in the beam formulation can be avoided.

2.3.1 Levinson Beam Theory Levinson Beam Theory (LBT), proposed in 1981, is a higher-order shear deformation beam theory. This beam theory does not use a shear coefficient as the displacement field of the theory can predict realistic shear stress distribution across the beam thickness. The displacement field is [3],

∂∂

+

−+=

xw

hzzuu xx

02

0 34 φφ

(11)

),(0 yxww =

(12)

(12)

The displacement filed of the theory involves two unknown functions. The derived differential equations of the Levinson Beam Theory are given by

),,(32

20

20 tyxq

twA

xwAG

x−

∂∂

=

∂∂

+∂∂ ρφ

(13)

−∂∂

∂∂

=

∂∂

−∂∂

∂∂

+

∂∂

+

φρ

φφ

45

451

32

02

2

20

20

xw

tI

xxwEI

xxwAG

(14)

The above two coupled differential equations need to be solved to determine the two unknown variables. The beam deflections and frequencies predicted by LBT are quite accurate. Unlike Timoshenko Beam Theory, LBT does not require a shear coefficient. The derivations of this theory are easy to understand as the differential equations of LBT are derived by Newtonian approach. Further, it is easy to understand the physical meaning of the boundary conditions used in case of beam analysis using LBT.

2.3.2 Reddy Beam Theory Reddy Beam Theory (RBT), proposed in 1984, is a higher-order shear deformation beam theory. This theory is classified under the category of third-order theories as the displacement field of this theory involves third order terms. The theory can predict realistic shear stress distribution across the beam thickness. The displacement field of the theory is [4],

∂∂

+

−+=

xw

hzzuu xx

02

0 34 φφ

(15)

),(0 yxww =

(16) The displacement field of Reddy’s Beam Theory is same as that of LBT. But, the difference between these two theories lies in the approach used in deriving the governing differential equations. In case of RBT, the governing equations are derived by energy principles, whereas, in case of LBT the differential equations are derived by Newtonian approach. The governing equations of RBT are given as:

2

22

21

tuAp

xw

xu

xAE

∂∂

=+

∂∂

+∂∂

∂∂ ρ

(17)

∂∂

−∂∂

∂∂∂

−∂∂

=

+

∂∂

−∂∂

∂∂

−

∂∂

++

∂∂

+∂∂

∂∂

∂∂

2

2

2

3

2

2

2

2

2

2

2

10516

211

10516

211

158

21

tI

xtwI

xtwA

qx

EIxwEI

x

xwGA

xw

xu

xwAE

x

φρρρ

φ

φ

(18)

2

2

2

2

2

2

10516

10568

158

10516

10568

txwI

tEI

xGA

xwEI

xEI

x

∂∂∂

+∂∂

−=

∂∂

++

∂∂

−∂∂

∂∂

−

ρφ

φφφ

(19)


There are three coupled differential equations. The number of variables can be reduced to two by ignoring the variable corresponding to the axial displacement. It is to be observed that, the equations governing the RBT are more complex compared to those of LBT, even though both the theories are formulated based upon the same displacement field. RBT is considered to be more accurate compared to LBT as the equations of RBT are derived through variable consistent approach.

2.3.3 Single Variable Refined Beam Theory This Single Variable Refined Beam Theory (SVRBT) [14] involves only one unknown function. This is also a higher-order shear deformation beam theory. In the expressions for displacements, we have third order terms. No need of using a shear coefficient. The displacement field is

∂∂

∂∂

+∂∂

−×

−

+

+∂∂

−=

xw

tI

twEI

hz

hz

Ebxwzu

bb

b

2

2

2

3

3

2103)1(2

ρ

µ

(20)

∂∂

+∂∂

−+

+= 2

2

2

2

5)1(12

twI

xwEI

Ebhww bb

b ρµ

(

(21)

Even though theory is a higher-order theory, the formulation of the theory involves only one unknown function. Most of the equations of this theory have strong similarity to the equations of Euler-Bernoulli Beam Theory (EBT). Therefore, the beam analysis using this theory involves slightly higher effort required in case of beam analysis using EBT.

The governing equation of this theory is derived by using the gross equilibrium equations. The differential equation governing the beam behavior is given by

)(4

4

xqdx

wdEI b =

(

(22)

The above differential equation is similar to the governing equation of EBT. By solving the above differential equation, we can obtain the value of bending deflection (i.e. wb). Next, the total lateral deflection ‘w’ can be calculated by using the following equation:

2

22

5)1(

dxwdhww b

bµ+

−=

(23)

The analysis of beams using SVRBT is similar to that of EBT as most of the equations of this theory are as simpler as those of EBT.

For the case of vibration analysis, the derived differential equation of SVRBT can be written as follows:

),(5

)1(125

)1(121

4

42

2

2

22

4

4

4

txqtw

EI

twA

txwI

xwEI

b

bbb

=∂∂+

+

∂∂

+∂∂

∂

++−

∂∂

µρ

ρµρ

(24)In the above equation we have only one unknown variable.

By solving the above equation for unknown variable (wb), we can study the vibration behavior of a beam under

consideration. The above differential equation includes both the effects of shear deformation as well as rotary inertia.

3. REVIEW OF PLATE THEORIES

3.1 CLASSICAL PLATE THEORY Classical Plate Theory (CPT), proposed in 19th century, is a basic and easy to use plate theory available in the literature plate theories. This theory is also known as “Kirchhoff-Love Theory.” The formulation of the theory along with properly prescribed boundary conditions was ready by approximately around 1850. Even though the governing equation of the theory was available since 1815, the theory could not be used to its full potential till 1850, due to the difficulty in understanding of the free edge boundary conditions associated with the theory. The convincing explanation about boundary conditions is obtained after Kirchhoff used energy principles to derive the governing equations and boundary conditions of CPT. The displacement field of CPT is given by [12, 13],

xwzu∂∂

−= 0 (25) y

wzv∂∂

−= 0 (26)

),(0 yxww = (27) The displacement field of the theory involves only one

variable (w0). Using the above expressions for displacements, we can write the expressions for strains and stresses of a given plate. Next, the governing differential equation of CPT can be written as follows:

),(022 yxqwD =∇∇

(

(28)

where 2

2

2

22

yx ∂∂

+∂∂

=∇

The above differential equation is for the case of static analysis of a plate. By solving the above equation, one gets the solution for static deflection of a plate. Next, the governing equation for the case of vibration of a plate can be written as:

),,(20

2

022 tyxq

twhwD =

∂∂

+∇∇ ρ

(29)

In the above equation, the effects of shear deformation and rotary inertia are ignored. The use of above equation for the case of thick plates would result in the overestimation of frequencies. To overcome this drawback, one can use the shear deformation plate theories.

3.2 First-order Shear Deformation Plate Theories 3.2.1 Mindlin Plate Theory Mindlin Plate Theory (MPT), proposed in 1921, is a first-order shear deformation plate theory. The formulation of this theory is analogous to that of Timoshenko Beam Theory (TBT) proposed in 1921. Difference is that TBT is one-dimensional theory, whereas, MPT is two-dimensional theory. As discussed in case of beams, first-order plate theories predict constant shear stress across the plate thickness. Therefore, one


will have non-zero shear stresses at the top and bottom surfaces of the plate. This theory uses a shear correction factor. The displacement field of the MPT can be written as [12, 13]:

xzu φ−= (30) yzv φ−= (31) ),( yxww o= (32)

There are three unknown variables are involved in the displacement field of MPT. For the complete plate analysis using MPT, one needs to solve the MPT equations to determine these three variables. The differential equations connected with MPT for the case of vibrations of a thick plate can be written as:

012

21

21

2

230

2

2

2

2

2

=∂∂

−

−∂∂

+

∂∂∂+

+∂∂−

+∂∂

th

xwkGh

yxyxD

xx

yxx

φρφ

φµφµφ(33)

012

21

21

2

230

2

2

2

2

2

=∂∂

−

−

∂∂

+

∂∂∂+

+∂∂−

+∂∂

th

ywkGh

yxxyD

yy

xyy

φρφ

φµφµφ

(34)

),,(20

2

20

2

20

2

tyxqtwh

yxyw

xwkGh yx

=∂∂

+

∂∂

−∂∂

−∂∂

+∂∂

−

ρ

φφ(35)

The above equations are coupled differential equations. By solving above three equations, we can have the complete solution of a given plate. In above equations, the notation ‘k’ denotes the shear coefficient. Based upon the literature, one can decide the value of shear coefficient. The shear coefficient is needed to approximately map the results of Mindlin Plate Theory with those of exact theory and other shear deformation plate theories.

3.3 Higher-order Shear Deformation Plate Theories 3.3.1 Levinson Plate Theory Levinson Plate Theory (LPT), proposed in 1980, is a higher-order shear deformation theory. This theory is a third-order plate theory. In case of LPT, no shear coefficient is required. The displacement field of this theory involves three unknown variables. The displacement field of LPT is [8],

∂∂

+

−+=

xw

hzzuu xx

02

0 34 φφ (36)

∂∂

+

−+=

yw

hzzvv yy

02

0 34 φφ (37)

),(0 yxww =

(38)

Using the above displacement expressions, one can obtain

the strains and stresses of a given plate under consideration. The differential equations related to Levinson Plate Theory can be written as follows:

( ) 20

2

02 ),,(

32

twhtyxqwGh

∂∂

=++∇ ρψ (39)

( ) ( ) ( )

∂∂

−∂∂

=

∂∂

+−

∇

∂∂

−∂∂

++∇−

xw

th

xwGh

wxx

D

xx

x

02

230

022

4603

22111

52

φρφ

ψµφµ(40)

( ) ( ) ( )

∂∂

−∂∂

=

∂∂

+−

∇

∂∂

−∂∂

++∇−

yw

th

ywGh

wyy

D

yy

y

02

230

022

4603

2

2111

52

φρφ

ψµφµ(41)

Similar to MPT, in case of Levinson Plate Theory also we have three coupled differential equations. The above differential equations have been derived by Newtonian approach.

3.3.2 Reddy Plate Theory Reddy Plate Theory, proposed in 1984, is a higher-order shear deformation plate theory. Likewise Levinson Plate Theory, this theory also does not require a shear coefficient. The displacement field of the Reddy Plate Theory is same that of Levinson Plate Theory. However, Reddy’s theory formulation differs from Levinson theory in the approach used in deriving the governing equations. Levinson has used Newtonian approach, whereas, Reddy has used energy principles to derive the governing differential equations. The displacement field is given by [9],

∂∂

+

−+=

xw

hzzuu xx

02

0 34 φφ (42)

∂∂

+

−+=

yw

hzzvv yy

02

0 34 φφ (43)

),(0 yxww =

(44) The governing differential equations of Reddy’s Plate Theory are not presented here as they are too lengthy. Those equations are available in the Ref. [].

3.3.3 Refined Plate Theory Refined Plate Theory (RPT), proposed in 2001, is a two variable shear deformation plate theory. The analysis of plates using RPT involves two variables. The two variables involved are bending deflection (wb) and shear deflection (ws). The displacement field of RPT can be written as [10]:

xw

hz

hzh

xwzu sb

∂∂

−+

∂∂

−=3

35

41 (45)

yw

hz

hzh

ywzv sb

∂∂

−+

∂∂

−=3

35

41 (46)

),(),( yxwyxww sb +=

(47) The unknown variables in the above equations can be

determined by solving the following two differential equations:

Dyxqwb),(22 =∇∇ (48)


( )D

yxqwh

w ss),()1(5

841 2

222 =∇

−−∇∇

µ

(49)

The above two differential equations are uncoupled and hence, these equations can be solved separately. Therefore, obtaining the solution using RPT is simpler when compared with other shear deformation plate theories. Most of the differential equations associated with shear deformation theories are coupled and hence, solution is harder. Further, RPT displacement field predicts realistic quadratic shear stress distribution across the plate thickness. Therefore, no need to use a shear coefficient in case of RPT. Many equations of RPT are having strong similarity to the equations of CPT. Hence, RPT is easy to understand and RPT equations can be easily dealt with in the similar lines CPT equations.

3.3.4 Single Variable Refined Plate Theory Single Variable Refined Plate Theory (SVRPT) [15] is a theory developed based upon two variable RPT discussed in the preceding section. The displacement field of the SVRPT involves only one variable. Also, the moment and shear force expressions of a plate given by SVRPT are similar to those of expressions given by CPT. One can easily understand SVRPT in the similar lines of CPT. The governing differential equation of SVRPT is given by

),(22 yxqwD b =∇∇

(50)

[5

The above differential equation is strikingly similar to the governing equation of CPT. Using the above equation one can determine the bending deflection. The total plate deflection can be determined by using the following equation:

bb whww 22

)1(5∇

−−=

µ

(51)

Obtaining the bending solution of a plate using SVRPT is almost same as that involved in case of CPT.

For the case of vibrations of a plate, the differential equation is given by

( )

( ) ),,(51

)1(5121

12

4

432

2

2

22

2322

tyxqtwh

twh

wt

hwD

bb

bb

=∂∂+

+∂∂

+

∇∂∂

−

+−∇∇

µρρ

µρ

(52)

The above equation involves the terms pertaining to shear deformation as well as the terms related to rotary inertia.

4. CONCLUSION

In this paper, a short discussion on classical and refined beam and plate theories has been discussed. An attempt has been made to familiarize the reader about the different class of beam and plate theories such as classical, first-order and higher-order theories. The necessity of shear deformation theories in case of thick beam and plate theories has been highlighted. Throughout the paper, the discussion is restricted to the beam and plate theories suitable for the analysis of plates made of isotropic materials. For the case of laminated

plates and plates made of functionally graded materials, the reader is suggested to further explore the beam and plate theory literature. Also, nonlocal beam and plate theories are available for the analysis of micro-scale and nano-scale beams and plates. The literature on applications of beam and plate theories for the case of beams and plates made of smart or intelligent materials is also very interesting.

REFERENCES 1. Y. M. Ghugal and R. P. Shimpi, “A review of refined sheardeformation theories for isotropic and anisotropic laminatedbeams,” J Reinf Plast Compos, Vol. 20, pp. 255_272, 2001.2. S. P. Timoshenko, “On the correction for shear of thedifferential equation for transverse vibrations of prismaticbars,” Philos Mag, Vol. 41, pp. 744_746, 1921.3. M. Levinson, “A new rectangular beam theory,” J SoundVib, Vol. 74, pp. 81–87, 1981.4. P. R. Heyliger and J. N. Reddy, “A Higher Order BeamElement for Bending and Vibration Problems,” J Sound Vib,Vol. 126, pp. 309_326, 1988.5. Y. M. Ghugal and R P Shimpi, “A review of refined sheardeformation theories of isotropic and anisotropic laminatedplates,” J. Reinf. Plast. Compos., Vol. 21, pp. 775_813, 2002.6. E. Reissner, “The effect of transverse shear deformation onthe bending of elastic plates,” J. Appl. Mech., Vol. 12, 69_77, 1945.7. R. D. Mindlin, “Influence of rotary inertia and shear onflexural motions of isotropic, elastic plates” J. Appl. Mech.,Vol. 18, pp. 31_38, 1951.8. M. Levinson, “An accurate, simple theory of the statics anddynamics of elastic plates.” Mech. Res. Commun. Vol. 7, pp. 343_350, 1980. 9. J. N. Reddy, “A simple higher-order theory for laminatedomposite plates,” J. Appl. Mech., Vol. 51, pp. 745_752, 1984.10. R. P. Shimpi, “Refined plate theory and its variants,”AIAA J, Vol. 40, pp. 137_146, 2002.11. R. P. Shimpi, H. G. Patel and H. Arya, “New first-ordershear deformation plate theories.” J. Appl. Mech., Vol. 74, pp. 523_533, 2007. 12. I. H. Shames and C. L. Dym, “Energy and finite elementmethods in structural mechanics,” Washington: Hemisphere Publishing Corporation, 1985, pp.185–204; 257–266; 323–350. 13. C. M. Wang, J. N. Reddy and K. H. Lee, “Sheardeformable beams and plates: Relationships with classicalsolutions,” Oxford: Elsevier Science, pp. 11_23, 2000.14. R. P. Shimpi, R. A. Shetty and A. Guha, “A Simple SingleVariable Shear Deformation Theory for A Rectangular Beam,”Proc. Inst. Mech. Eng., Part C, Vol. 231, pp. 4576-4591, 2017.15. R. P. Shimpi, R. A. Shetty and A. Guha, “A SingleVariable Refined Theory for Free Vibrations of a Plate UsingInertia Related Terms in Displacements.” Eur. J. Mech. A.Solids., Vol. 65, pp. 136-148, 2016.


A Blockchain Approach for Eliminating Counterfeit

Drugs in Pharma Supply Chain

Chaithra Acharya*, Krithika S Udupa, Akshath H S and Keerthan P M Department of Computer Science and Engineering, Sahyadri College of Engineering & Management, Mangaluru – 575007

*Email: [email protected] ABSTRACT The following paper gives a general review of the trending Blockchain technology and the platforms that be able to deploy to build an infallible system in eradicating spurious drugs in pharmaceutical enterprises. Several report analysis indicates that introduction of substandard drugs into the supply chain by unknown sources is one of the significant problems encountered by the drug companies. Applying blockchain throughout the supply chain operations, multiple issues can be resolved as blockchain is reliable, secure and immutable in nature. Overall, this paper aims to explain how blockchain transactions can fix the existing issues faced by the stakeholders Keywords: Anti-Counterfeit, Blockchain, Pharmaceutical Supply Chain, Smart Contract 1. INTRODUCTION

1.1 Blockchain It is a distributed and non-editable database storing blocks of transactions tied together through cryptographic methods across a peer-to-peer network. The underlying architecture allows the shareholders (nodes) connected over the network to share the entries (ledger). Whenever a transaction occurs, all the nodes receive a copy of the transaction since the data gets synchronized across the entire network through peer-to-peer replication. Its distributed nature rules out the need of a central authority as the transactions are visible and validated by the participating nodes of the network.

Figure 1 : Blockchain architecture

1.2 Smart Contract

It is a code promoting the settlement among participants intended to be stored over the blockchain. It is run along with the transaction making transactions tamper proof and trustworthy. The purpose of these smart contracts aims at higher security than traditional contracts. Each smart contract has a contract related address as well as hash which is utilized for caching and recovering contracts in a fool-proof way. 1.3 Supply Chain It comprises of various check-points associated with the production and shipment of assets. Nowadays, a supply chain involves a lot of steps and in many terrestrial places. It results in hard to track events and review issues in each stage. When a participant in a supply-chain introduces substandard goods, the inquiry becomes tedious and usually, neither individual is responsible. Hence both needs a trusted entities to approve a entity a specific quality of the products and the service into consideration. 2. LITERATURE SURVEY Nakamoto.S [1], the paper proposes a new form of digital cash eliminating the need for a assured third party. The system is tamper-proof reducing the frauds existing in the current system. The transactions are validated through the proof-of-work.



After the acceptance by the other nodes, the transactions are attached with the timestamp and added as a block to the existing blockchain. The proposed idea is against the centralized banks but does not achieve the level of privacy provided in the existing systems. M. Crosby et al. [2], the paper discusses the applications of blockchain technology in financial as well as non-financial fields. Non-financial applications like music, IoT, blockchain based counterfeit solutions. It also explores the risks involved in its adoption in real world applications like resistance to change by people, scaling, government regulations etc. In V. Hua et al. [3], the paper suggests how this technology result in an improvement in the assurance factor in supply-chains.

In a supply chain, customers need a certain level of quality of service. Trust is needed for record and also security, authenticity is a must. Therefore, when implementing there are several limitations and need to trust middlemen. So, blockchain is designed for it to overcome this problem. Kentaroh Toyoda et al. [4], the paper implements an experimental system using the Ethereum platform to prevent counterfeit drug supply from cloning of RFID tags. The paper aims at identifying substandard products introduced within the supply chain. With this system still, if the tags are replicated, modification of product ownership is not possible. Thomas Bocek et al. [5], the paper puts forward an integrated system with blockchain and IOT for the supply chain pharmaceutical. It discusses the startup Modium.io which executes such a system. In addition to this, it explains some of the start-ups working on blockchain technology. The system comprises of the front and back end and sensor devices. An Ethereum platform and temperature sensors assure that temperature provided for a system is right. For continuous interactions, the servers are connected to blockchain network, smart contract as well for clients. API allows to create new records in a system. It checks that whether it is immutable by any person within the system. This article infers a blockchain can be integrated with other domain.

In Gendal Brown et al. [6], the paper explains the design and the main ideas of their model: A machine-readable contract code needs an agreement between two or more parties represented by state objects. The transaction is responsible between state objects and their protocol results in no need for centralized authority. This article gives a comparison of a different blockchain platform. The author suggests Croda was built to manage and record business agreements organizations. Soundarya K. et al [7], the paper discusses about the different blockchain platforms that can be used to implement systems which eliminate counterfeit drugs being introduced in supply chains. It also talks about the challenges to be dealt with when using the blockchain technology.

3. PROPOSED SYSTEM

A purpose of this project is building trustable and a robust platform to bypass the counterfeit situations in the supply chain of pharmaceutical industries.

3.1 Overview The aim of the project is to develop a model/prototype for Blockchain to implement a supply chain for pharmaceutical industries to prevent bogus products in the system. With the immutability and transparency obtained by Blockchain, misconducts occurring within the supply chain can be avoided. The different actors are cryptographically appended into a Blockchain network and can add as well as extract details (or records) from the Blockchain. Hence, the patients can be aware of all the related information like the product source, various stages in the supply chain and the holders involved from the data stored in the Blockchain.

Every participant can enter data at every checkpoint which is reflected across the common network. Due to the visibility factor, each shareholder contains data that have been replicated so no manipulation of data is not possible at any given point.

3.2 Architecture Diagram

Architecture diagrams are used to depict the relationship between different components of a system. It is essential to understand the complete concept of the system. The Figure2 shows the architectural diagram of the proposed system. In the proposed system each participant at each checkpoint uses the API to add the related legitimate data to a checkpoint. The entered data stored in a database through a server. It smartly manages the retrieval of data stored in the database. The smart contract is created in IDE and further assembled and blockchain is deployed into it.

4. IMPLEMENTATION

4.1 Components The following components are required for the proposed system:

Figure 2 : Architecture diagram of the proposed system


• Blockchain Platform is needed to make and executesmart contract between participants(nodes) connectedthrough it. There is numerous accessible to an open-source platform of such as MultiChain, BigChain DBEthereum, etc.

• Cloud provides a platform for running virtualmachines and servers and act as different nodes for theblockchain. Some of cloud service providers areAWS,Azure.

• Smart-contract is needed to store and validate thedetails provided in each node of the supply chain.Transaction is done only if policy is agreed.

• Database is needed to store the information entered bythe clients on the application and also to storeinformation related to the used smart contracts.

• Server is needed to establish interaction amongblockchain network, the participants, and smartcontracts.

• Application is needed to provide an interface to enterdetails by participants and also to view the enteredinformation.

4.2 Blockchain Platforms

4.2.1 Ethereum This platform is decentralized which allows building decentralized applications also know as DApps. It also helps to run smart contracts over a built blockchain. The smart contracts overcome the possibility of fraud or third-party interference. It is a public blockchain and can be used in financial as well as non-financial applications. It is adaptable and flexible unlike the Bitcoin protocol. The cryptocurrency used for transactions is named Ether. It supports various programming languages.

4.2.2 Hyperledger Fabric It is a standardized, open, enterprise-grade distributed ledger framework by IBM alongside with The Linux Foundation. Hyperledger Fabric allows for components such as membership services and consensus protocols. The application logic of the system comprises of a chain code (similar to smart contract). Its applications include Banking, Finance, Internet of Things, Supply chain etc., Hyperledger Fabric can be used as a public as well as a private blockchain and it supports Python, Go languages.

4.2.3 MultiChain It helps to rapidly build blockchain applications that can be used for financial transactions. It is an example for a permissioned blockchain. It supports Windows, Linux and Mac servers provides a simple API and a CLI to preserve and set up the chain. Users have the ability to customize the platform accordingly. It supports Python, JavaScript, Ruby, C#, and PHP languages.

4.2.4 Quorum It is created by J.P. Morgan which is a version of Ethereum, basically an enterprise-focused version of. It is also an example of a permissioned network. Applications which requires higher throughput and greater speed for processing of the transaction uses this type of platform. It provides higher performance and privacy when compared to Ethereum. It has extra features like transaction privacy and new consensus mechanisms like Raft-based Consensus and Istanbul BFT. C++, Go and Python are the languages supported by Quorum. 4.2.5 Corda By using this application we can build interoperable networks for permissioned applications. It includes a feature like managing and recording an agreement between two or more participants as per the existing legal guidelines and conventions without a central authority. It provides strict privacy between the participants within the blockchain network. Java is the language supported by Corda.

5. CONCLUSION

The following approach aims at solving the existing issues in the pharmaceutical supply chain using blockchain. Its immutability and reliability factor helps in eliminating the introduction of counterfeit and substandard drugs into the supply chain. Since all the transactions are visible to each and every actor in the supply chain, there is no need of a central authority to supervise the operations.

REFERENCES

[1] Nakamoto S. “Bitcoin: A peer-to-peer electronic cashsystem”, 2008.

[2] M. Crosby, Nachiappan, P. Pattanayak, S. Verma, V andKalyanaraman.”Blockchain technology: Beyond bitcoin”AIR-Applied Innovation review Issue no. 2 pp 6-19. June2016.

[3] A. V. Hua and Jorgen S. Notland, “Blockchain enabledtrust & transparency in supply chains” ntnu school oftiø4530. 2016.

[4] .K. Toyoda, T.Mathiopoulus, I.Sasae, and T.Ohtsuki “ANovel Blockchain-based product ownershipManagement System(poms) for anticounterfeits in thepost supply chain”. IEEE Acess 2017.

[5] T.Bocek, B. Rodrigures, T.Strasser, and B, Stiller,“Blockchains everywhere- a use case of blockchains inthe pharma supply chain”, In IFIP/IEEE InternationalSymposium on Integrated Network Management 2017.

[6] R.G. Brown, J.Carlyle and M.Hearn, “Croada Anintroduction”,2016

[7] Soundarya, K.Priyanaka Pandey and R.Dhanalakshmi,“A counterfeit solution for Pharma Supply Chain”, April2018

.

Guidelines for Submission

• Authors are requested to provide full details for correspondence: postal

address, phone numbers and email address. (The email address of

corresponding authors will be published along with the article).

• Authorsarerequestedtopreparetheirsoftcopyversionintextformatsand not in

PDF version.

• Authorsarerequestedtofollowthemanuscripttemplatewhilepreparingthe

manuscript.

SIJR is devoted to the publication of original research work. The journal also

accepts review papers and scientific articles. SIJR welcomes the submission of

manuscripts that meet the journal standard which includes novelty, significance

and excellence. Every volume of the journal will have two issue in a year. All

articles published in the SIJR are peer-reviewed.

Contact for any query or Information on

SAHYADRI International Journal of Research

Dr. Pushpalatha K - Editor-in-Chief

or

Dr. Navin. N. Bappalige - Editor

Dr. Ravindra Babu G - Editor

Dr. M B Savitha –Editior

Mr. S N Bharath Bhushan - Editor

Mr. Balaji. N - Editor.

Mrs. Geetha S D - Editor

Mrs. Megha N - Editor

Mrs. Aysha Shabana - Editor

Email: [email protected] Web: www.sijr.in

Sahyadri Journal of ResearchInternational

SAHYADRI International Journal of Research

PublisherManjunath Bhandary - President

Bhandary Foundation, Sahyadri Campus, Adyar, Mangaluru - 575 007

PrintersPrakash Offset Printers, 164, Industrial Area, Baikampady, Mangaluru - 575 011

| Vol 4 | Issue 2 | December 2018

mailto:[email protected]

http://www.sijr.in/

Sahyadri Campus, Adyar, Mangaluru - 575 007

COLLEGE OF ENGINEERING & MANAGEMENTSAHYADRI

(Affiliated to VTU, Belagavi and Approved by AICTE, New Delhi)

Empowering Young Minds

+ 91-824-2277222 | [email protected] | www.sijr.in | www.sahyadri.edu.inSAHYADRI Journal of Research | December 2015 | 3

Sahyadri College of Engineering & Management established in 2007 is recognized by AICTE,

UGC, MHRD & MSME Government of India, affiliated to VTU, Government of Karnataka, NAAC

accredited with “A” Grade and accredited by Institution of Engineering (India) IEI, certified by ISO

and also by DSIR as Scientific and Industrial Research Organization (SIRO). The College offers B.E.,

MBA, M.Sc. (Engineering), Doctoral and Research Programs. The total staff and student strength

at campus is nearly 4000.

WD IE TT HI D G

E R

R A

C D

C E

A

Documents

Sahyadri Journal of Research...It gives us immense pleasure to bring out Volume 4, Issue 2 of Sahyadri International Journal of Research (SIJR). The journal covers the wide disciplines