ReviewArticle ...islanded operation. 1. Introduction With the high penetration levels of renewable in the power network, the traditional centralized network posed many technical issues

Review ArticleA Survey on Microgrid Control Techniques in Islanded Mode

Uthpala N Ekanayake and Uditha S Navaratne

Department of Electrical and Electronic Engineering University of Peradeniya 20400 Peradeniya Sri Lanka

Correspondence should be addressed to Uthpala N Ekanayake uth7ekanayakegmailcom

Received 16 June 2020 Revised 21 September 2020 Accepted 28 September 2020 Published 15 October 2020

Academic Editor Andrea Bonfiglio

Copyright copy 2020 Uthpala N Ekanayake and Uditha S Navaratne is is an open access article distributed under the CreativeCommons Attribution License which permits unrestricted use distribution and reproduction in any medium provided theoriginal work is properly cited

Traditional power networks with generation in upstream and consumption in the downstream were controlled with centralizedcontrols like SCADA However in order to facilitate the penetration of distributed generation the concept of microgrid waspopularized A microgrid can operate both in grid-connected and in islanded modes One of the challenges in the microgridenvironment is to provide both voltage control andmaintain the system frequency while ensuring the stability of the networkispaper is a literature survey focused on different microgrid control techniques with different levels of communication especially inislanded operation

1 Introduction

With the high penetration levels of renewable in the powernetwork the traditional centralized network posed manytechnical issues for maintaining stability and protectionLimited resources and high energy demand have resulted inmany governments to set up policies to promote renewableenergy e energy generation which was limited to the toplevel of the network hierarchy has now expanded to thedistribution level with the increasing use of solar PV andother renewable sources like wind e control and moni-toring once seen in the generation and transmission havenow become a must even at the distribution e concept ofmicrogrids has facilitated the integration of renewable re-sources at this level A microgrid is a small-scale distributionnetwork containing a set of distributed generations (DGs)designed to supply electrical power to a local area (LA)connected to a large network (host grid) as a single con-trollable entity [1] Microgrids became popular because oftheir ability to work in isolation

A microgrid operation can be in two modes When themicrogrid fulfills its energy demand by the main grid it iscalled grid-connected mode and when demand is suppliedfrom its own local generation it is called islanded mode Ingrid-connected mode the main objective of a controller is to

provide energy management while in islanded mode theobjective is to control both its frequency and voltage whilesupporting the energy demand If there are no synchronousmachines to balance demand and supply for island opera-tion the inverter is responsible for providing these controlsespecially the frequency control [2ndash4] PQ (Active Powerand Reactive Power) control is the control strategy used inthe grid-connected mode for power balance and it switchesto Vf (voltage frequency control) when the microgrid is inislanded mode [2] e most common approaches in gridcontrol architecture are centralized control and decentral-ized control architectures e centralized control approachis widely used in traditional power networks mainly at thegeneration level with Supervisory Control Data Acquisition(SCADA) systems [5] Here the control operation isplanned beforehand and is implemented With distributedgeneration decentralized control architectures becamepopular

2 Centralized Control

A central controller is used with distributed sensors actu-ators local controllers and supporting communicationnetwork Data collected from the distributed sensors are sentto the centralized control via a communication network

HindawiJournal of Electrical and Computer EngineeringVolume 2020 Article ID 6275460 8 pageshttpsdoiorg10115520206275460

Once the central controller gathers data it calculates thecontrol variable for each control equipment and sends themback to the local controllers [4 5] Local controls neverperform decision-making and all the decisions are made atthe central controlleris is illustrated in Figure 1 Here theabbreviation LA corresponds to the local area controller andCC to the central controller

3 Decentralized Control

In a decentralized control approach the distribution systemis divided into small networks called microgrids Amicrogrid can have a few local area (LA) networks and itsown controller In a fully decentralized approach each unit(DER) in a microgrid is controlled by its local controller(LC) [4ndash6] An LC supervises a set of LAs that require nocommunication between LAs When the LCs communicateto find a common solution for the overall control problemthe control architecture is called distributed

Islanded microgrid operation is challenging due to theintermittent nature of renewable energy generation eycreate uncertainties in maintaining a stable voltage andfrequency output Hence this shows the requirement of anaccurate load forecasting and load management system witha decentralized nature However a fully decentralized ap-proach is not possible as it requires strong coordinationbetween the LCs

4 Distributed Control

Decentralized and distributed control operates on the sameprinciple except that distributed architecture has commu-nication with local networks (LAs) and LCs A distributedcontrol scheme is illustrated in Figure 2

Distributed controls assign control tasks to different DGsand their interaction ese control tasks are based ondifferent time frame operations and they constitute a controlhierarchy In a network when the number of units increasesdistributed control is challenging unless a control hierarchyis established is approach is used for the coordination ofeach unit is will be discussed further in the next sectione distributed techniques aim to address an underlyingoptimization problem in a distributed manner with limitedcommunication [7ndash9]

Distributed control strategies include model predictivecontrol-based techniques consensus-based techniques andagent-based techniques which will be discussed in latersections e recent interest in research of distributedcontrol strategies shows microgrid island operation andcontrol together with preserving privacy and protecting thesystem from cyberattacks [7]

5 Hierarchical Control

e hierarchical control system has two concepts namelymultilayer and multilevel In a multilayer concept thecontrol is split into layers and each acts at different timeintervals A hierarchical control scheme consists of threelevels of control primary secondary and tertiary [6 7] e

controller has different control layers to coordinate thecontrol process While the upper layers manage the objec-tives globally by providing control set points the lowerlayers perform the controlling In a multilevel conceptcontrol is introduced into local controllers (LCs) and theiraction is coordinated by an additional unit MCC (MicrogridCentral Control) with the Main Grid Central Controller Ahierarchical control-based control architecture is illustratedin Figure 3 [6]

e control levels in the hierarchy are illustrated inFigure 4 [10] It is necessary to ensure the control from onelevel to the lower levels and the communication bandwidthof the hierarchical control is decreased with an increase incontrol level

e main purpose of the primary level controller is toprovide direct control to the electronic converters tomaintain voltage and frequency the secondary level isresponsible to establish coordination among the neigh-boring local controllers and mitigate voltagefrequencydeviations caused at the primary control level In the thirdlevel and the highest level the controller takes all theuncertainties of the power network to determine the op-timal power flow ensuring demandgeneration balance[10 11]

51 Primary Control Primary control is the first level in thehierarchy dealing with the local controllers ey require afast response with minimum or no communication tocontrol problems in the power network e primary pur-pose is islanding detection and voltagefrequency control[6ndash11] Primary control can be categorized into masterslavecontrol and distributed control which requires communi-cation links to be established and other categories of droopcontrols which require no communication Since masterslave control requires communication channels with higherbandwidths and it reduces the reliability in case of failure of asingle point of the network the widely adapted method isdroop control [12] Primary control is performed by VoltageSource Inverter (VSI) controllers

e inverter that interfaces the DERs to the grid works intwo ways It works in Power Control Mode (PCM) whenoperating in grid-connected mode and Voltage ControlMode (VCM) when operating in islanded mode [4] VCMcontrol is used to regulate the output of the VSI where droopcharacteristics are used to control voltage and frequencye droop controls are described using equations (1) and (2)as follows

ω ω0 minus KplowastPg (1)

V V0 minus KqlowastQg (2)

where ω and V are the new steady-state frequency andvoltage values respectively ω0 and V0 are the base fre-quency and voltage values fixed when operating in grid-connected mode Kp and Kq are the active power and re-active power static droop gains and Pg and Qg are the three-phase injected active power and reactive power

2 Journal of Electrical and Computer Engineering

Droop control realizes active power with droop char-acteristics to improve the frequency and reactive power toimprove the output voltage of the DG units However thereare disadvantages involved in droop controls ey are poorin voltage regulation and have high voltage distortion aprobability of a loss of synchronism when resynchronizingwith the main grid [2] and most importantly poor reactivepower sharing among the DERs Droop control guaranteesthe stability of frequency and correct active power sharingHowever it produces errors in reactive power sharing isis highlighted through grid parameters relevant to theregulation problem in [12] If Qi

prime is the reactive power de-livered to the Point of Common Coupling (PCC) and V isthe voltage at PCC then

Qiprime

Vi minus V

Xi1113874 1113875V (3)

According to equation (3) the connection impedance tothe PCC affects the correctness of reactive power sharingApart from this the reactive power setpoint in droopcontrols from an energy management system (EMS) and theincrease in load demand also affects the reactive powersharing

Nondroop methods such as described in [13] use power-sharing methods A decentralized servomechanism con-troller at each DG unit adjusts its voltage independentlyPower sharing is achieved using an internal oscillator thatextracts the phase angle and transmits it to DGs Howeverthis type of method is difficult to implement in largersystems

Recent developments include the modeling of the in-verter as a virtual generator Its goal is to mimic the con-ventional synchronous machine dynamics to improvevoltage and frequency stability [14 15]

Reference [14] describes a Synchronverter that mimicsthe behavior of a synchronous generator employed withmodifications to the software without changing the hard-ware ese modifications can include virtually controllingthe field current so that it delivers a better way of computingreactive power increasing the size of filter inductors to makethe inverter more robust towards small voltage violationsvirtually changing the nominal active mechanical torqueand introducing virtual capacitors to eliminate the influencefrom DC voltages

Reference [15] describes a comprehensive inverter-BESS-based primary control e control of this system canregulate frequency and voltage in microgrid islands in VCMmode independently of the number of parallel operations ofother DERs with the adaptation of a virtual generator istechnique includes an automatic voltage regulator and arotor flux model which models the reactive powervoltagemagnitude similar to that of a synchronous generatorthrough machine flux variation as in equation (4) and aninertia model and a frequency governor that makes thesystem similar to a synchronous generator active power

Data and command flow

LA1 LA2 LA3

LA4LA5 CC

Figure 1 Centralized control architecture All local area networksor microgrids in the main network are controlled by a centralizedcontroller

Microgrid 1 Microgrid 2

LA1 LC1

LC2

LA1

LA3

LA2LA2

LA3

Data and commands flow

Figure 2 Distributed control architecture A controller is assignedto each microgrid that can communicate with each other

Microgrid control 1

LA1 LC1

LA2

LC2

Main grid

Centralcontroller

Microgridcontrol 3

Microgridcontrol 2

LA3MCC

Datacommand flow

Figure 3 Hierarchical control architecture MCC coordinates theactions of LCs with a main grid

Tertiary control

Secondary control

Primary control

Power flow control

Regulating small deviationsof voltage and frequencydue to primary control

Regulating individual voltageand frequency outputs of DGs

Figure 4 Representation of control hierarchy [10]

Journal of Electrical and Computer Engineering 3

frequency dynamics through virtual mechanical dynamic asin equation (5)

dvrefdt

ki QAvR minus Q( 1113857 (4)

dωref

dt K Pgov minus P minus kdωref1113872 1113873 (5)

vref is the initial voltage variation ki is rotor flux model gainwhich is modeled as an integrator and QAVR and Q arereactive power at the output of the AVR and the inverterrespectivelyωref is the rotational speed of the virtual gen-erator K is inertial model gain kd is the damping effect andPgov and P are the active power at the output of the governorand the inverter respectively

Secondary control level is used to mitigate the problemsaroused in the primary control

52 Secondary Control Secondary control is required inboth grid-connected and in islanded modes However thelatter is challenging due to the uncertainties of generation inDGs and the update rate of load dispatch commands [3 4]e main objective at this level hence includes finding theoptimal dispatch of all DERs and restoring the voltage andfrequency deviations produced at the primary control Real-time optimization is required at this level for economicdispatch and unit commitment

Secondary control is the highest hierarchical level inislanded microgrids [3 4] is control operates in a slowertime frame than the primary control reducing the com-munication bandwidth used by the microgrid variablesHence it allows the performance of more complex calcu-lations e secondary control level can be centralized ordistributed In centralized secondary control MCCmanagespower flow and provides reference values for primarycontrol After gathering data it determines the deviationsand decides the secondary control action [16] However thisapproach is highly dependent on communication and ismore expensive and unreliable Recent developments showdistributed control approaches at this level to reducecommunication

In literature [17] presents a distributed secondary control(DSC) scheme for inverters in a network with uncertaincommunication links e discrete-time DSC controllers usean iterative learning mechanism that enables the DERs in themicrogrid to achieve voltagefrequency restoration and activepower sharing Here the secondary control inputs areupdated at the end of an iteration round thus DERs onlyneed to share data with its neighbors in low bandwidth

e secondary controllers are not necessary to be fre-quently operated Hence [18] suggests an event-triggeredapproach e controller is updated only when a localmeasurement error exceeds a certain boundary therebylimiting the communication Because of the existence ofcommunication noise among DGs [19] describes a fullydistributed noise-resilient secondary voltage control ap-proach to select a proper set of control inputs for voltage andfrequency restoration

An optimal operation is sought in secondary controlalso referred to as the energy management of a microgridthrough the implementation of multiagent systems whereindividual DER units are controlled by local agents

53 Tertiary Control Tertiary control is the highest and thelowest level of control for grid-connected microgrids op-erating in hierarchical control schemes It sets long-term setpoints to operate for DER units Moreover it is responsiblefor managing multiple microgrids in a host network [2ndash4]is control level applies to the operation and managingpower flow between the microgrid and the main grid Inislanded mode tertiary control gets disabled Hence it is notdiscussed in detail in this survey

6 Agent-Based Techniques forDistributed Control

Energy management systems with intelligent decisions andadaptive feedback have complex behaviors Hence multi-agent applications became popular in this domain Amultiagent system decomposes a complex problem intosubproblems e subproblem decisions are then taken byan agent attempting to achieve a certain objective An agentcould be any DER a load a transformer or any other deviceor any entity in the network [20ndash22] In islanded operation amultiagent system will control DER units based on theinformation exchanged by the agents through a local con-troller ey perform negotiations and energy trading tomeet load demands [20ndash22] e control and communi-cation architecture of these agents could be in centralizeddistributed or hierarchical control While the agents in localareas are responsible for communicating with each other tosolve the control issues locally there can be coordinationwith a communicative agent among them is is illustratedin Figure 5

Centralized agent architecture consists of agents man-aged by a single controller in a master-slave relationship[22] Distributed agent architecture consists of several agentsallowed to discover global information through communi-cation with their neighbors Hierarchical agent consists ofdifferent layers of agents whose communication is achievedvia a layered architecture e information flow is fromlower layers to higher layers whereas the control flow isfrom higher layers to lower layers Besides information flowin layers different agents in the same layer may also be ableto communicate with each other

61 Centralized Agent Architecture Here a collection ofnonintelligent agents is managed by a central controller in amaster-slave approach As described in [22] the agents arecategorized into two classes reactive and cognitive ecentral controller is a cognitive agent with intelligence re-sponsible for communicating with nonautonomous reactiveagents to establish the desired control plan like managingand disconnecting noncritical loads A diagram in Figure 6describes these two classes of agents


62 Distributed Agent Architecture Here a collection ofagents is handled by a single layer control structure insteadof hierarchical layers e agents are intelligent and au-tonomous and responsible for managing a local area in thenetwork ese agents can discover global informationthrough communication and coordination with theirneighboring agents [22]

63 Hierarchical Agent Architecture In this architectureagents have different authorities over the control actionsWhile some agents havemore authority to sense andmanageover the others Typically the upper layerlevel agents areresponsible for handling a large amount of data andmaintaining the communication schedules for demandgeneration balance e middle layerlevel agents are re-sponsible for making decisions switching between gridconnection and islanded modes minimizing the energylosses and directing the necessary actions for neighboringagents in identifying faults and restoring power e lowerlayerlevel agents are responsible for interacting with sensorsin the microgrid Sensors are responsible for sensing andcontrolling the grid breakers DGs battery energy storagedevices (BESS) and controllable loads [22]

7 Peer-to-Peer Communication

Peer-to-peer communication is considered a technique thatcan be used in multiagent systems Each agent requires ameasurement of its own and its nearest neighbor and re-quires a very low bandwidth for this communication [16]Here all the agents have the same priority and the im-portance of communicating in a peer-to-peer fashion [6]Algorithms used for P2P communication between the agentsinclude gossiping and consensus algorithms Gossip algo-rithms can be used for voltage and frequency control at thesecondary level

ese algorithms aim at sharing data quickly in thenetwork [6] To facilitate the absence of a central controller

these agents require some sort of intelligence making themicrogrid autonomous

Over the past few years extensive research has beenconducted on multiagent systems (MAS) MAS-based en-ergy management system (EMS) architectures have beenwidely used for distributed control and demand-side energymanagement in microgrids Literature [23] presents a MAS-decentralized EMS that employs fuzzy cognitive maps forenergy management [21] presents an outage managementsystem using MAS and in [24] microgrid local controllersare assigned to some specific agents to interact with eachother to achieve a common decision in the restoration ofvoltage All these studies demonstrate the beneficial appli-cation of MAS in microgrids where the playersagents werecooperating and various decision-making methods andcontrol strategies were used to attain the optimal microgridperformance using techniques like Particle Swarm Opti-mization game theory and generic algorithms

8 Model Predictive Control

In the future microgrid a control hierarchy with multiagentwill be coordinated by a centralized optimization problem tofulfill a control goal ese control strategies can be cate-gorized to direct control and indirect control A directcontrol has a two-way communication for accessing theDERs then communicating back the control strategy andgetting feedback On the other hand indirect control needsonly one way to initialize DER control is is where themodel predictive control (MPC) strategies come into playMPC can predict the next control sequence along with acontrol action (eg variation in output of a DER)

Extending MPC to multiagents involves knowing thesubsystem model of a control agent in advance However acontrol agent has no global information but can access onlylocal information erefore predictions are importantabout the future evolution of external variables of neighborsHowever when the horizon becomes longer the accuracycan get lost Research work suggests that this could bemitigated by assuming a constant value over the wholehorizon based on a local measurement or obtained from aneighboring agent It is also possible that assuming pre-dictions over the whole horizon or assuming a model that

LA1

Microgrid

LA2

LA3LA4

MCC

Datacommand flow between local networksDatacommand flow through a controller

Figure 5 Distributed control with agents is diagram illustratesthe data and command flow between the local areas through agentsas well as how the commands are centralized using a microgridcontroller

Reactiveagent

Reactiveagent

Reactiveagent

Reactiveagent

Cognitiveagent

Figure 6 Centralized multiagent architecture is diagram showsthe two types of agents who are responsible for communicating incontrol architecture the same as in Figure 5


predicts the values of the external variables based on thedynamics of neighboring subsystems [25]

e MPC diagram is illustrated in Figure 7e commonly used MPC strategy is called Distributed

Model Predictive Control (DMPC) In this discrete timecontrol methodology an optimal controller is developedbased on the dynamic system model and forecasting eMPC control objective is a real-time optimization problemand is repeatedly computed for all control inputs e op-timization problem can be based on minimizing costs orforcing the system to follow predefined control set pointsbased on a prior state estimation Literature in [26 27]describes different objective functions based on traditionallinear quadratic regulators economic objectives and acombination of both

e MPC scheme has been used to derive linearizedmodels for voltage control in [27] It predicts the voltageprofile in the next time step and adjusts the voltage andreactive power set points accordingly to achieve a smoothvoltage profile in the next step Equation (6) is the objectivefunction as given in [27] based on optimizing the gener-ation cost of a microgrid operating in islanded operation

minF(U k) 1113944k+Nminus1

tk

Cg (t) + 1113944tm

Cm (t)⎛⎝ ⎞⎠ (6)

where Cg (t) is the generation cost from the main gridproportional to power demand Cm(t) is the cost of the mth

conventional distributed generator in the microgrid and Uis a vector containing control actions in the entire controlhorizon given by equation (7)

U(t) uT(k) u

T(k + 1) middot middot middot u

T(k + N minus 1)1113960 1113961

T (7)

Here for any single time step the control signal U(t)includes power outputs of all controllable generators as wellas energy storage

MPC is a type of dynamic control that requires a quickresponse is is illustrated in Figure 8 with other controlstrategies DMPC and the coordination between the localcontrollers can be again categorized into distributed andhierarchical communication

MPC is also used for inverter controls [28 29] as aprimary level control technique Literature in [28] has usedthis technique for decentralized primary control of anislanded microgrid (MG) composed only by power con-verters and without any rotating electrical machine and wastested to an islanded portion of the Smart Polygeneration

Microgrid (SPM) testbed facility at the University of GenoaItaly

In [29] the approach focuses on solving optimizationproblems of MPC with an analytical optimal solution forunconstrained mode and near-optimal solution in a tran-sient state It gives an output with a fixed switching fre-quency increases robustness under system parameteruncertainties and variations and gives a fast-dynamicresponse

9 Consensus-Based Technique

In consensus-based techniques a distributed optimizationproblem is solved to converge DER operating set points to asingle value In [30 31] consensus-based multiagentschemes are used for load restoration fault recovery andfrequency control e literature in [30] describes a dis-tributed voltage secondary control (DVSC) strategy using adynamic consensus protocol Here a voltage correction isprovided to droop control units using consensus-basedprotocol As shown in Figure 9 to regulate the voltage at anode i the responsible agent has a controller that uses adynamic consensus estimate given by equation (8)

_Ve(t) 1113944N

j1aij Vj(t) minus Ve(t)1113872 1113873 + Vi (8)

Here Vi is the measured voltage at node i Ve is theestimate of the average voltages provided at node I and Vj isthe voltage estimate received by the neighboring node j eestimated voltage is then compared with a referencemicrogrid voltage and the error is fed into a PI controllerSince the consensus control is dependent on the degree ofconnectivity between its agents different communicationtopology exists between the DERs e daisy chain topologyloop communication topology and star topology are someof these examples

e principle of defining the consensus method is basedon equation (8) and two theories Here aij is the element ofthe adjacency matrix of the graph and V is the graph nodesin graph theoryese graph nodes must satisfy the principleof distributed consensus e first theory says that the graphconsists of a spanning tree (a subset of a graph which has allthe vertices covered with a minimum possible number ofedges) causing a consensus control and all eigenvalues havezero or positive parts e second theory is that if a graphconsists of a spanning tree all agentsrsquo states will converge tothe external control signal ese are the fundamentals usedin consensus control [32]

Optimizer Model

x = Ax + Buy = Cx + Du

u y

y

r

d

Reference

Disturbance

Figure 7 Model predictive control diagram


Distributed generation can be physically placed in widelyspanned locations e consensus-based method can bringthese generators to a global agreement of a parameter usingonly communication among neighbors [31]

10 Conclusions

e literature survey describes the techniques used formicrogrid control and communication giving relevance tothe islanded mode operation Islanded microgrid controlsare responsible for making decisions on maintaining powerbalance and providing voltage and frequency control isincludes the equilibrium supply surplus and the supplyshortage Droop controls can maintain discrepancies involtage and frequency However they still can deviate fromtheir reference values Recent developments include anondroop control technique at the primary level where theinverters mimic the conventional synchronous machinedynamics Additional control levels are introduced as sec-ondary controls in islanded microgrids to mitigate the de-viations at the primary level While the traditional secondarycontrol methods include a centralized structure the recentlydeveloping concepts focus more on distributed controlstructures Instead of gathering global information thesecontrols can use local data collected from neighbors One ofthe challenges in distributed control is ensuring that localcontrol actions are consistent with the action of others

Agent-based techniques are popularized which can usepeer-to-peer communication strategies as intelligent agentsattempting to achieve an objective using neighboring dataIn a multiagent setting where DMPC has employed eachcontrol agent makes a prediction of its subsystem over ahorizon from local data Due to the constraints of other

subsystems these predictions by their respective agentscommunication between the neighboring agents must beaccurate enough to improve the decision-making processHowever what information to be communicated and whento decide on the control action are still some open researchin this area In the case of frequency restoration the dis-tributed consensus-based control algorithms have demon-strated more accuracy From all the mentioned algorithmsthe consensus-based control used with an agent-basedscheme requires the least communication resources and hasthe fastest convergence time that suits the microgrid sec-ondary control in restoring variations in voltage and fre-quency during islanded operation

Data Availability

e data used to support the findings of this paper areincluded within the article

Conflicts of Interest

e authors declare that there are no conflicts of interestregarding the publication of this paper

References

[1] K V Vidyanandan and B Kamath ldquoGrid integration ofrenewables challenges and solutionsrdquo in Emerging EnergyScenario in India-Issues Challenges and Way Forward Ney-veli Tamil Nadu India 2018

[2] O Palizban and K Kauhaniemi Microgrid Control Principlesin Island Mode Operation PowerTech Grenoble France2013

[3] K C Soni and F F Belim ldquoMicroGrid during grid-connectedmode and islanded mode-A reviewrdquo International Journal of

Planningmaintenance

and schedulingOptimization Dynamic control

for exampleMPC

Basic controlfor example

PIPID controlPlant

Yearly Daily Within minutes Within seconds Time

Disturbances

Figure 8 A control hierarchy with respect to a timelineis diagram illustrates when and where MPC which is a dynamic control is placedin a control hierarchy

Agentvjvi

veaij Imdashs++ +

ndash

Communication path with neighboring agents

Figure 9 Consensus-based control diagram is diagram shows how consensus-based algorithm is used for voltage regulation


Advanced Engineering and Research Development vol 20pp 2348ndash4470 2016

[4] P Borazjani N I A Wahab H B Hizam et al ldquoA review onmicrogrid control techniquesrdquo IEEE Innovative Smart GridTechnologies-Asia (ISGT ASIA) vol 2014 pp 749ndash753 2014

[5] D E Olivares A Mehrizi-Sani A H Etemadi et al ldquoTrendsin microgrid controlrdquo IEEE-PES Task Force on MicrogridControl vol 14 2014

[6] A Shreshtha B P Hayes and R Bishwokarma ldquoPeer-to-peerenergy trading in micromini-grids for local energy com-munitiesrdquo A Review and Case Study of the Nepalese ElectricitySystem vol 7 no 1 pp 131911ndash131928 2019

[7] M Yazdanian and A Mehrizi-Sani ldquoDistributed controltechniques in microgridsrdquo IEEE Transactions on Smart Gridvol 5 no 6 pp 2901ndash2909 2014

[8] H Almasalma J Engels and G Deconinck ldquoPeer-to-peercontrol of microgridsrdquo in Proceedings of the Young Re-searchers Symposium Eindhoven Netherlands 2016

[9] Z A Obaid L M Cipcigan L Abrahim et al ldquoFrequencycontrol of future power systemsreviewing and evaluatingchallenges and new control methodsrdquo Journal Modern PowerSytems and Clean Energy vol 8 2019

[10] N Yusof1 and Z Ali ldquoReview of active synchronization forrenewable powered microgridrdquordquo International Journal ofEngineering amp Technology vol 8 no 17 pp 14ndash21 2019

[11] L Meng Hierarchical control for optimal and distributedoperation of microgrid systems PhD thesis Aalborg Uni-versity Denmark Europe 2015

[12] A Rosini M Minetti G B Denegri and M InvernizzildquoReactive power sharing analysis in islanded AC microgridsrdquoin Proceedings of the 2019 IEEE International Conference onEnvironment and Electrical Engineering and 2019 IEEE In-dustrial and Commercial Power Systems Europe (EEEICIampCPS Europe) IEEE Genoa Italy pp 1ndash6 September 2019

[13] R Moradi H Karimi and M Karimi-Ghartemani ldquoRobustdecentralized control for islanded operation of two radiallyconnected DG systemsrdquo in Proceedings of the 2010 IEEEInternational Symposium on Industrial Electronics (ISIE)London UK July 2010

[14] V Natarajan and G Weiss ldquoSynchronverters with betterstability due to virtual inductors virtual capacitors and anti-winduprdquo IEEE Transactions on Industrial Electronics vol 64no 7 pp 5994ndash6004 2017

[15] M Fusero A Tuckey A Rosini P Serra R Procopio andA Bonfiglio ldquoA comprehensive inverter-BESS primarycontrol for AC microgridsrdquo Energies vol 12 p 381 2019

[16] P Singh P Paliwal and A Arya ldquoA review on challenges andtechniques for secondary control of microgridrdquordquo IOP Con-ference Series Materials Science and Engineering vol 561Article ID 012075 2019

[17] X Lu X Yu J Lai J M Guerrero and H Zhou ldquoDistributedsecondary voltage and frequency control for islandedmicrogrids with uncertain communication linksrdquo IEEETransactions on Industrial Informatics vol 13 no 2pp 448ndash460 2017

[18] Y Wang T L Nguyen Y Xu Z Li Q-T Tran and R CaireldquoCyber-physical design and implementation of distributedevent-triggered secondary control in islanded microgridsrdquoIEEE Transactions on Industry Applications vol 55 no 6pp 5631ndash5642 2019

[19] N M Dehkordi H R Baghaee N Sadati and J M GuerreroldquoDistributed noise-resilient secondary voltage and frequencycontrol for islanded microgridsrdquo IEEE Transactions on SmartGrid vol 10 no 4 pp 3780ndash3790 2019

[20] H-M Kim T Kinoshita and M-C Shin ldquoA multiagentsystem for autonomous operation of islanded microgridsbased on a power market environmentrdquo Energies vol 3no 12 pp 1972ndash1990 2010

[21] R Leo A A Morais R Rathnakumar et al ldquoMicro-grid gridoutage management using multi agent systemsrdquo in Pro-ceedings of the 2 nd International Conference on Recent Trendsand Challenges in Computational Models Tindivanam IndiaFebruary 2017

[22] A Kantamnenia L E Browna G Parkerb et al ldquoSurvey ofmulti-agent systems for microgrid controlrdquo EngineeringApplications of Artificial Intelligence vol 44 2015

[23] C Karavas K Arvanitis and G Papadakis ldquoA game theoryapproach to multi-agent decentralized energy management ofautonomous polygeneration microgridsrdquo 2017

[24] E Rokrok M Shafie-khan P Siano and J P S Catalao ldquoAdecentralized multi-agent based approach for low voltagemicrogrid restorationrdquo Energies vol 22 2017

[25] R R Negenborn B De Schutter and H Hellendoorn ldquoMulti-agent model predictive control of transportation networksrdquoin Proceedings of the 2006 IEEE International Conference onNetworking Sensing and Control (ICNSC 2006) pp 296ndash301Lauderdale FL USA April 2006

[26] R Halvgaard ldquoModel predictive control for smart energysystemsrdquo Technical University of Denmark Department ofApplied Mathematics and Computer Science vol 23 2017

[27] J Ma F Yang Z Li and S Joe Qin ldquoA renewable energyintegration application in a microgrid based on model pre-dictive controlrdquo IEEE Power and Energy Society GeneralMeeting vol 23 pp 22ndash26 2012

[28] F Blanco A Labella D Mestriner and A Rosini ldquoModelPredictive Control for Primary Regulation of IslandedMicrogridrdquo in Proceedings of the 2018 IEEE InternationalConference on Environment and Electrical Engineering and2018 IEEE Industrial and Commercial Power Systems Europe(EEEICIampCPS Europe) IEEE Denmark Europe pp 1ndash6June 2018

[29] H T Nguyen E-K Kim I-P Kim H H Choi andJ-W Jung ldquoModel predictive control withmodulated optimalvector for a three-phase inverter with an LC filterrdquo IEEETransactions on Power Electronics vol 33 no 3 pp 2690ndash2703 2018

[30] Q Shafiee T Dragicevic F Andrade J C Vasquez andJ M Guerrero ldquoDistributed consensus-based control ofmultiple DC-microgrids clustersrdquo in Proceedings of the 201440th Annual Conference of the IEEE Industrial ElectronicsSociety IECON Singapore February 2015

[31] S Ahmed Fuad ldquoConsensus based distributed control inmicrogrid consensus based distributed control in micro-gridclusters clustersrdquo Master esis Michigan TechnologicalUniversity Houghton Michigan 2017

[32] F Aghaee N M Dehkordi N Bayati and A HajizadehldquoDistributed control methods and impact of communicationfailure in ac microgrids a comparative reviewrdquo Electronicsvol 8 no 11 p 1265 2019


Once the central controller gathers data it calculates thecontrol variable for each control equipment and sends themback to the local controllers [4 5] Local controls neverperform decision-making and all the decisions are made atthe central controlleris is illustrated in Figure 1 Here theabbreviation LA corresponds to the local area controller andCC to the central controller

3 Decentralized Control

In a decentralized control approach the distribution systemis divided into small networks called microgrids Amicrogrid can have a few local area (LA) networks and itsown controller In a fully decentralized approach each unit(DER) in a microgrid is controlled by its local controller(LC) [4ndash6] An LC supervises a set of LAs that require nocommunication between LAs When the LCs communicateto find a common solution for the overall control problemthe control architecture is called distributed

Islanded microgrid operation is challenging due to theintermittent nature of renewable energy generation eycreate uncertainties in maintaining a stable voltage andfrequency output Hence this shows the requirement of anaccurate load forecasting and load management system witha decentralized nature However a fully decentralized ap-proach is not possible as it requires strong coordinationbetween the LCs

4 Distributed Control

Decentralized and distributed control operates on the sameprinciple except that distributed architecture has commu-nication with local networks (LAs) and LCs A distributedcontrol scheme is illustrated in Figure 2

Distributed controls assign control tasks to different DGsand their interaction ese control tasks are based ondifferent time frame operations and they constitute a controlhierarchy In a network when the number of units increasesdistributed control is challenging unless a control hierarchyis established is approach is used for the coordination ofeach unit is will be discussed further in the next sectione distributed techniques aim to address an underlyingoptimization problem in a distributed manner with limitedcommunication [7ndash9]

Distributed control strategies include model predictivecontrol-based techniques consensus-based techniques andagent-based techniques which will be discussed in latersections e recent interest in research of distributedcontrol strategies shows microgrid island operation andcontrol together with preserving privacy and protecting thesystem from cyberattacks [7]

5 Hierarchical Control

e hierarchical control system has two concepts namelymultilayer and multilevel In a multilayer concept thecontrol is split into layers and each acts at different timeintervals A hierarchical control scheme consists of threelevels of control primary secondary and tertiary [6 7] e

controller has different control layers to coordinate thecontrol process While the upper layers manage the objec-tives globally by providing control set points the lowerlayers perform the controlling In a multilevel conceptcontrol is introduced into local controllers (LCs) and theiraction is coordinated by an additional unit MCC (MicrogridCentral Control) with the Main Grid Central Controller Ahierarchical control-based control architecture is illustratedin Figure 3 [6]

e control levels in the hierarchy are illustrated inFigure 4 [10] It is necessary to ensure the control from onelevel to the lower levels and the communication bandwidthof the hierarchical control is decreased with an increase incontrol level

e main purpose of the primary level controller is toprovide direct control to the electronic converters tomaintain voltage and frequency the secondary level isresponsible to establish coordination among the neigh-boring local controllers and mitigate voltagefrequencydeviations caused at the primary control level In the thirdlevel and the highest level the controller takes all theuncertainties of the power network to determine the op-timal power flow ensuring demandgeneration balance[10 11]

51 Primary Control Primary control is the first level in thehierarchy dealing with the local controllers ey require afast response with minimum or no communication tocontrol problems in the power network e primary pur-pose is islanding detection and voltagefrequency control[6ndash11] Primary control can be categorized into masterslavecontrol and distributed control which requires communi-cation links to be established and other categories of droopcontrols which require no communication Since masterslave control requires communication channels with higherbandwidths and it reduces the reliability in case of failure of asingle point of the network the widely adapted method isdroop control [12] Primary control is performed by VoltageSource Inverter (VSI) controllers

e inverter that interfaces the DERs to the grid works intwo ways It works in Power Control Mode (PCM) whenoperating in grid-connected mode and Voltage ControlMode (VCM) when operating in islanded mode [4] VCMcontrol is used to regulate the output of the VSI where droopcharacteristics are used to control voltage and frequencye droop controls are described using equations (1) and (2)as follows

ω ω0 minus KplowastPg (1)

V V0 minus KqlowastQg (2)

where ω and V are the new steady-state frequency andvoltage values respectively ω0 and V0 are the base fre-quency and voltage values fixed when operating in grid-connected mode Kp and Kq are the active power and re-active power static droop gains and Pg and Qg are the three-phase injected active power and reactive power




Qiprime

Vi minus V

Xi1113874 1113875V (3)







LA1 LA2 LA3

LA4LA5 CC



LA1 LC1

LC2

LA1

LA3

LA2LA2

LA3



Microgrid control 1

LA1 LC1

LA2

LC2

Main grid

Centralcontroller

Microgridcontrol 3

Microgridcontrol 2

LA3MCC

Datacommand flow


Tertiary control

Secondary control

Primary control

Power flow control






dvrefdt


dωref

























LA1

Microgrid

LA2

LA3LA4

MCC



Reactiveagent

Reactiveagent

Reactiveagent

Reactiveagent

Cognitiveagent








tk

Cg (t) + 1113944tm

Cm (t)⎛⎝ ⎞⎠ (6)



U(t) uT(k) u


T(k + N minus 1)1113960 1113961

T (7)








_Ve(t) 1113944N




Optimizer Model


u y

y

r

d

Reference

Disturbance




10 Conclusions




Data Availability




References




Planningmaintenance


for exampleMPC


PIPID controlPlant


Disturbances


Agentvjvi

veaij Imdashs++ +

ndash





































Qiprime

Vi minus V

Xi1113874 1113875V (3)







LA1 LA2 LA3

LA4LA5 CC



LA1 LC1

LC2

LA1

LA3

LA2LA2

LA3



Microgrid control 1

LA1 LC1

LA2

LC2

Main grid

Centralcontroller

Microgridcontrol 3

Microgridcontrol 2

LA3MCC

Datacommand flow


Tertiary control

Secondary control

Primary control

Power flow control






dvrefdt


dωref

























LA1

Microgrid

LA2

LA3LA4

MCC



Reactiveagent

Reactiveagent

Reactiveagent

Reactiveagent

Cognitiveagent








tk

Cg (t) + 1113944tm

Cm (t)⎛⎝ ⎞⎠ (6)



U(t) uT(k) u


T(k + N minus 1)1113960 1113961

T (7)








_Ve(t) 1113944N




Optimizer Model


u y

y

r

d

Reference

Disturbance




10 Conclusions




Data Availability




References




Planningmaintenance


for exampleMPC


PIPID controlPlant


Disturbances


Agentvjvi

veaij Imdashs++ +

ndash




































dvrefdt


dωref

























LA1

Microgrid

LA2

LA3LA4

MCC



Reactiveagent

Reactiveagent

Reactiveagent

Reactiveagent

Cognitiveagent








tk

Cg (t) + 1113944tm

Cm (t)⎛⎝ ⎞⎠ (6)



U(t) uT(k) u


T(k + N minus 1)1113960 1113961

T (7)








_Ve(t) 1113944N




Optimizer Model


u y

y

r

d

Reference

Disturbance




10 Conclusions




Data Availability




References




Planningmaintenance


for exampleMPC


PIPID controlPlant


Disturbances


Agentvjvi

veaij Imdashs++ +

ndash













































LA1

Microgrid

LA2

LA3LA4

MCC



Reactiveagent

Reactiveagent

Reactiveagent

Reactiveagent

Cognitiveagent








tk

Cg (t) + 1113944tm

Cm (t)⎛⎝ ⎞⎠ (6)



U(t) uT(k) u


T(k + N minus 1)1113960 1113961

T (7)








_Ve(t) 1113944N




Optimizer Model


u y

y

r

d

Reference

Disturbance




10 Conclusions




Data Availability




References




Planningmaintenance


for exampleMPC


PIPID controlPlant


Disturbances


Agentvjvi

veaij Imdashs++ +

ndash








































tk

Cg (t) + 1113944tm

Cm (t)⎛⎝ ⎞⎠ (6)



U(t) uT(k) u


T(k + N minus 1)1113960 1113961

T (7)








_Ve(t) 1113944N




Optimizer Model


u y

y

r

d

Reference

Disturbance




10 Conclusions




Data Availability




References




Planningmaintenance


for exampleMPC


PIPID controlPlant


Disturbances


Agentvjvi

veaij Imdashs++ +

ndash




































10 Conclusions




Data Availability




References




Planningmaintenance


for exampleMPC


PIPID controlPlant


Disturbances


Agentvjvi

veaij Imdashs++ +

ndash


































































Documents

ReviewArticle ...islanded operation. 1. Introduction With the high penetration levels of renewable in the power network, the traditional centralized network posed many technical issues