Borelli Ita Revisto

7/27/2019 Borelli Ita Revisto

http://slidepdf.com/reader/full/borelli-ita-revisto 1/4

1

Tool Wear Diagnosis in Machinig Process

João Eduardo BorelliDepartment of Mechanical Engineering, Technological Institute of Aeronautics, CTA, São José dos Campos, SP, CEP:12228-901, Brazil

[email protected]

Luís Gonzaga TrabassoDepartment of Mechanical Engineering, Technological Institute of Aeronautics, CTA, São José dos Campos, SP, CEP:12228-901, [email protected]

Adi lson GonzagaDepartment of Electrical Engineering, EESC - USP, São Carlos, SP, CEP:13560-970, [email protected]

Reginaldo Teixeira CoelhoDepartment of Mechanical Engineering, EESC - USP, São Carlos, SP, CEP:13560-970, [email protected]

Abstract: During a machining process, solicitations of several natures are applied to the cutting tool. Temperature, pressure and the speed of the ofthe material chips in the cutting area trigger wear processes of the tool. This, by its turn, directly influence the finish quality, dimensionaltolerances and even the process safety. Researches demonstrate that the tool wear largely responds for the flaws of the manufacture systems, and

its detection is a necessary condition for automation of the machining processes. This paper presents a new and alternative form to diagnose thetool wear during the machining processes by using a fuzzy algorithm and infrared images. For this purpose a system composed by three moduleshas been developed and commissioned, as follows: (1) infrared images acquisition ,(2) image processing, and (3) decision making. The acquisitionmodule is composed by an infrared camera, a frame grabber board and a host microcomputer. It captures and digitizes the images during the process. The processing module is constituted by a software that extracts the image feature based upon the first order image histogram. Thedecision making module operates through a fuzzy algorithm that has as inputs the images features and as output an index which maps the actualtool wear. The results shown in this paper are very encouraging, since the system has been effective, with success rate superior to 90%. Keywords: diagnosis, tool wear, image processing, fuzzy logic, automation.

Introduction

The knowledge of the tool wear is one of the most important factors for controlling and automating the machining processes inmanufacturing systems. Tool wear is related to parameters, such as cutting speed, feed rate depth of cut and arise as a consequence ofmechanisms triggered by temperature and stress on the tool surface. It responds for several failures during the machining operationsas well and can strongly affect tool and machine performance, surface quality, among others D´ERRICO, G.E.(1998), AY, H. et al.

(1994), CASTO, S.L. et al. (1994), ABRÃO A.M. et al. (1996), SILVA, M.B.; WALLBANK, J. (1999).Research works have extensively been searching for new variables to describe the tool wear in a way the it can be inferredthroughout the monitoring of different signals, which might be related to the wear mechanisms during the machining processes. Oneof the most important parameter is the temperature at the cutting region HUTTON, D.V.; F.HU.(1999), LEE, J.M. et al.(1994),CHOI, D. et al. (1999), JEMIELNIAK, K. (1999), ELZER, J.; PFEIFER.T. (1990), CHOUDHURY, S.K.; RATH. S. (2000).

The present work presents an innovative method for measuring tool wear in real time with the use of fuzzy logic and image data processing. To achieve this goal, a system has been developed, constituted by three modules: infrared image acquisition, image processing and interpretation and decision making.

The system developed yielded a customized solution with the following features: user friendly interface, automatic calculationsof temperature at different regions and with different materials, machining images displays, isotherms plotting , automatic featureextraction and automatic decision making KALPAKJIAN, S.(1991), SHAW, M.C.(1984), ELZER, J.; PFEIFER.T (1990),CHOUDHURY, S.K.; RATH. S. (2000).

The technical specifications of the experiments made are described below. The image acquisition is done using the infraredcamera AGA Thermovision 720 adapted to a coupling system and a frame grabber (MÍCROVIDEO DC30 Míro). LabView was thedata acquisition system used running on hardware from National Instruments. The workpiece used a cold draw AISI 1045 steel,

machined with a inserts WNMG 06 04 08-PM, P15 clamped in a tool support MWLNL2525-06. Machining operation used a latheINDEX GU-600, 22 Kw, maximum rotation 5000 rpm. Cutting conditions was: depth of cut 0.2 and 0.4 mm, feed rate 0.07, 0.25 and0.5 mm/rev and cutting speed 295, 396 and 497 m/min.

Development of the System

The infrared image yields a map in gray levels of all material involved in the cutting process: workpiece, tool, chip and cuttingfluid. The luminosity of each pixel in the image corresponds to the peak energy for the different wavelengths of the infraredspectrum, coming from the temperature during the cutting process. The gray level map is converted into temperature map using acalibration curve which was drawn previously using the same camera and the different materials that could be used in the processBENNETT C.O.; MYERS.J.E.(1978), KOULIANG, W.(1983), SOLOMAN, S.(1998), LIN, J. et al. (1990).

When controlling a machining process, it is essential to identify cutting tools that are no longer capable to meet the requiredspecifications due to wear. Information about the wear progress and the best features to be considered as input to the system wereobtained from studying the images for different stages of the tool wear, new, worn and at the final life stage.

The method used for feature extraction in this work was the statistical method. It describes the model by statistical rules whichsteer the distribution and the list of gray levels. The statistical data can be computed indirectly from the image histogram and the process features chosen were the gray level average and standard deviation, kurtosis, energy and entropy.

mailto:[email protected]










2

The diagnosis system developed is based on the fuzzy inferring. The fuzzy algorithm has functions and rules which allow adiscrimination and classification of tool wear state during the cutting process, according to patterns previously established. The fuzzyset was initially proposed by Zadeh (1965) as an extension to the classical sets. The fuzzy set allows to represent vague conceptsexpressed in natural language ZADEH, L.A.(1965), PAL, S.K; MITRA,S.(1992), FRANÇA, C. A. (1999), KLIR,G.J.;YUAN,B.(1995).

Figure 1 shows the evolution of the gray levels in the infrared images, as a function of tool wear. It can be noted that as the toolgoes through crescent wear states: new, good, medium, worn and end of life, the infrared image registers crescent luminosityintensity. The gray levels correspond to temperatures varying from 180 to 1400 °C.

Figure 1. Gray tones as a function of tool wear for a tool at 497 m/min BORELLI, J.E.(2000).

The system developed used five fuzzy sets as input, represented by the five features extracted from the first order infrared imagesacquired during the cutting operations and a output fuzzy set represented by the variable wear. For each fuzzy set pertinencefunctions are defined, also called membership functions, which describe the elements pertinence within each fuzzy set. For thedetermination of these functions, experiments were performed (cutting operations) varying the wear level of the tools. Table 1 showsthe membership functions selected for each fuzzy set.

Table 1. Fuzzy sets and its input memb ership func tion s BORELLI, J.E.(2000).

Fuzzy Sets Membership Functions

LowIntermediate

Mean, Average

HighLowIntermediate

Std. Deviation

HighLowIntermediate

Kurtosis

HighLowIntermediate

Energy

High

LowIntermediate

Entropy

High

The membership functions of the fuzzy set “average” are shown in Figure 2.



3

Figure 2. Membership function s of fuzzy set “average” B ORELLI, J.E.(2000).

Similarly, it was determined the pertinence functions for the other fuzzy sets: standard deviation, Kurtosis, energy and entropy.To represent fuzzy set “tool wear”, it has been chosen the following membership functions: lowest, low, intermediate, high and

highest, as shown in Table 2.

Table 2. Output fuzzy set and respective members hip func tion s BORELLI, J.E.(2000).

Fuzzy Set Membership Functions

lowestlowintermediatehigh

wear

highest

The membership functions of the fuzzy set wear are shown in Figure 3.

Figure 3. Membership functions of fuzzy set Wear BORELLI, J.E.(2000).

The elaboration of the inference rules was the result of intense research study about the phenomenon of interest, that is tool wear,which allowed to correlate the input variables and the output.

Results and Conclusions

The calibration curves were obtained for the different materials. These curves were used by the software to generate plots ofisotherms on the tool, workpiece and material chips providing the user with more accurate data on the tool performance.

The determination of the fuzzy sets, membership functions and inference rules is the diagnosis engine of the system. The systemreceives the input the features matrix as input, and outputs an index corresponding to the tool wear.

For the assessment of the diagnosis system, experiments were performed with various cutting tools with different levels of wearwell known previously (provided by an experienced tool operator), 300 images were tested. The results are presented in Table 3.

Table 3. Validation of the proposed system BORELLI, J.E.(2000).

Quantity ofimages

Stage ofwear

Number of right

answers

Percentage

300 New 300 100%300 Intermediate 287 96%300 End of life 300 100%Average right answers 99%



4

From the data obtained in Table 5, one can conclude that the system has performed accordingly in all stages, starting from theinfrared image acquisition, then image the pre-processing, feature extraction, determination of membership functions, inference ruleselaboration up to the determination of tool wear. The developed system is reliable, with a correct answer in 99% of the cases. Thiswork is an innovative contribution for the automation in manufacturing systems and similar that have temperature relatedcharacteristics.

Acknowledgments

The authors wish to thank CAPES (Coordenadoria de Aperfeiçoamento de Pessoal de Nível superior - Brazil) and FAPESP(Fundação de Amparo a Pesquisa do Estado de São Paulo – Brazil) for their support to the development of the present work.

References

ABRÃO A.M., ASPINWALL, D.K.; NG. E.G.(1996). Temperature evaluation when machining hardened hot work die steel usingPCBN tooling. PCBN tooling – Industrial Diamond Review. v. 56. n.2. pp. 40-44.

AY, H.; YANG, W.; WANG.J.A.(1994). Dynamics of cutting tool temperatures during cutting process . Experimental Heat

Transfer , 7(3).BENNETT C.O.; MYERS.J.E.(1978). Fenômeno de transporte: quantidade de movimento calor e massa. São Paulo.

McGraw-Hill. E.10. 820 p.

BORELLI, J.E.(2000). Diagnóstico do estado de desgaste de ferramentas para o monitoramento de processos de usinagem de altodesempenho. Dissertação de mestrado. São Carlos, Escola de Engenharia de São Carlos, Universidade de São Paulo- Brazil.

CASTO, S.L.; VALVO, E.L.; PIACENTINI, M.; RUISI, V.F.(1994). Cutting temperatures evaluation in ceramic tools: experimentaltexts, numerical analysis and SEM observation. CIRP ANNALS, Manufacturing Technology. Hellwag Publishers LTD,Switzerland

CHOI, D.; KWON, W.T.; CHU, C.N.(1999). Real-time fracture in turning using sensor fusion. The international Journal of

Advanced Manufacturing Technology. v.15. n.5. pp. 305-310. Springer-Varlag. London Limited.CHOUDHURY, S.K.; RATH. S. (2000). In-process tool wear estimation in milling using cutting force model. Journal of Materials

Processing Technology. 99. pp.113-119. ELSEVIER.D´ERRICO, G.E.(1998). An adaptive system for turning process control based on tool temperature Feedback. Journal of Materials

Processing Technology. 78 (1998) 43-47.ELZER, J.; PFEIFER.T (1990). Tool wear. Journal of engineering for industry. Transactions of ASME . n.2. New York.FRANÇA, C. A. (1999). Avaliação da qualidade de placas de madeira através de um sistema de inferência nebuloso baseado em

redes adaptativas. São Carlos, EESC – USP. Tese de Doutorado.

HUTTON, D.V.; F.HU. (1999). Acoustic emission monitoring of tool wear in end-milling using time-domain averaging . Journal of Manufacturing Science and Engineering. Transactions of ASME. February. v. 121. n.1. pp. 8-12. New York.

JEMIELNIAK, K. (1999). Commercial tool condition monitoring systems. The International Journal of Advanced Manufacturing

Technology. 15. pp. 711-721. Spriger-Verlag London Limited.KALPAKJIAN, S.(1991). Manufacturing Processes for Engineering Materials. New York. Addison-Wesley. Second edition, 920 p.

ISBN:0-201-50806-0.KLIR,G.J.; YUAN,B.(1995). Fuzzy sets and Pattern Recognition.KOULIANG, W.(1983). The application of infrared thermometry in metal cutting. Proceedings of the First International Metal

Cutting Conference. March, 17-19, Guangzhou, China.LEE, J.M.; CHOI, D.K.; CHU, C.N. (1994). Real time breakage monitoring for NC turning and drilling. Manufacturing Technology.

CIRP Annals, v. 43/1.pp.81-84. Switzerland, Hallwag Publishers LTD.LIN, J.; CHEN, T.C.; WANG, C.I.(1990). Development of infrared pyrometer with fiber optic for measuring cutting temperature at

high speed cutting. ASME, PED. Monitoring and Control for Manufacturing Process. v.44. pp.17-32 .PAL, S.K; MITRA,S.(1992). Multilayer Perceptron, Fuzzy Sets and Classification. IEEE Transactions on Neural Networks.

vol.3, no.5. pp.683-697.SHAW, M.C.(1984). “Metal cutting principles”, Oxford Science Publications Clarendon Press Oxford.p.274-275.SILVA, M.B.; WALLBANK, J. (1999). Cutting temperature: prediction and measurement methods - a review. Journal of Materials

Processing Technology. Elsevier. v.88. pp.195-202.SOLOMAN, S.(1998). Sensors handbook. USA, McGraw-Hill. ISBN: 0-07-059630-1.ZADEH, L.A.(1965). Fuzzy Sets. Information and Control. vol.8. pp.338-353.

Documents

Borelli Ita Revisto