Page 1

3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

DifferentPowerQualityDetectionAlgorithms UsingWindowingTechnique PriazMadappillil,JoseMathew,AbleEldhose,AnoopThomas M.Tech.student,DepartmentofECE, RajagirischoolofEngineering&TechnologyCochin,India, S.F.O.Tech.,NeST,

s.i n

Asst.professor,DepartmentofECE, RajagirischoolofEngineering&Technology,Cochin,India,

as

df .re

AbstractǦSetoflimitsofelectricalpropertieswhichallowselectricalsystemstofunctionintheirintended manner without substantial loss of performance is called power quality. Today’s electricity consuming equipment’sareverymuchsensitivetopowerqualityproblems;hencepowerqualityisagrowingconcern. Problemslike,sagsandswells,lastonlyforafractionofasecond,butcauseseveredamageandgreatlossin the power systems. Hence the detection of the voltage sags and swells is considered very important. The paper presents three algorithms based on a windowing technique to detect sag and swell using a PIC32 microcontroller. Also their comparisons are made. Simulation results using Mat lab are also presented to verifytheeffectivenessoftheproposedmethods. KeywordsǦpowerquality;sag;swell;windowingtechnique

lib .

I.Introduction

.e d

Power quality determines the fitness of electrical power given to consumer devices. It describes electric powerthatdrivesanelectricalloadandtheload'sabilitytofunctioncorrectly.Withouttheproperpower, anelectricaldevicemaynotfunctionintheintendedmanner,failprematurelyornotoperateatall. NorealǦlifepowersourceisideal;theygenerallydeviateinatleastthefollowingways:

w

w

VariationsinthepeakorRMSvoltage AshortdurationdecreaseinvoltagevaluesǦsag AShortdurationincreaseinvoltagevaluesǦswell Abrupt,verybriefincreasesinvoltageǦspikes HighoverǦvoltagedisturbancesthatlastforaveryshorttimecalledtransients

w

x x x x x

Much of theelectrical equipment in an industry requires highǦquality electricity; it will not tolerate sags, swells, transients, or harmonics as these disturbances cause problems such as malfunctions, instabilities, shortlifetime,failureofelectricalequipment’setc.

VoltageSag PowersystemshavenonǦzeroimpedances,soeveryincreaseincurrentcausesacorrespondingreductionin voltage.Usually,thesereductionsaresmallenoughthatthevoltageremainswithinnormaltolerances.But when there is a large increase in current, or when the system impedance is high, the voltage can drop significantly[5]. Voltagesagsareoneofthemostcommonpowerdisturbances.Equipment’susedintoday’sindustriesare becomingmoresusceptibletovoltagesagsduetoincreaseintheequipmentcomplexity.

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$



VoltageSwell

s.i n

Fig.1VoltageSag

w

w

.e d

lib .

as

df .re

Voltageswellsareshortincreasesinvoltagethatlastsfromacycletoasecondorso,ortensofmilliseconds to hundreds of milliseconds. They are caused by sudden drop in the load on a circuit with poor voltage regulationorduetolooseneutralconnection.Althoughmoreobviousaretheeffectsofasags,theeffectsof avoltageswellareoftenmoredestructive.Itmaycausebreakdownofcomponentsonthepowersuppliesof theequipment,thoughtheeffectmaybeagradual,accumulativeeffect.Itcancausecontrolproblemsand hardware failure in the equipment, due to overheating that could eventually result to shut down. Also, electronicsandothersensitiveequipmentarepronetodamageduetovoltageswell. 

 Fig.2VoltageSwell

w

Restofthepaperincludesthedescriptionofthreealgorithmsbasedonmovingwindowtechniqueinorder todetectthenumberofsagsandswellsoccurredinaninputvoltage.Allthethreealgorithmshavebeen simulatedinMatlabandhavebeencompared.

II.DetectionAlgorithmUsingMovingWindowTechnique ToensurethePowerquality,powerdisturbancedetectionbecomesimportant.Inthispaperbasicallythree algorithmsaredescribedtodetectthetotalnumberofvoltagesagsorswellsthathasoccurred.

MovingWindow Calculations are performed with the samples enclosed by the window in Fig.3. Now as in the Fig.4, the windowismovedbyonesamplesuchthatthewindowenclosesthesamenumberofsamples,butthenew windowdeletesthefirstsampleoftheoldwindowandaddsthenextsampleafterthelastsampleoftheold window.Nowwiththesamplesinthenewshiftedwindow,calculationsareperformed.

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

s.i n



.e d

lib .

as

df .re

Fig.3Windowenclosingafewnumbersofsamples



w

w

Fig.4Samewindowenclosingthesamenumberofsamples butmovedbyonesample,hencemovingwindow.

A.MovingAverageWindowAlgorithm

w

A moving average is used to analyze a set of sample points of a sampled voltage by creating a series of averages of different subsets of the full sample set. It utilizes a window of certain size, taken to be a full cycle for determining the subset. Given a series of samples and a fixed subset or window size, the first elementofthemovingaverageisobtainedbytakingtheaverageoftheinitialfixedsubsetorwindowofthe sampleseries.Thenthewindowismodifiedby"shiftingforward";thatis,excludingthefirstsampleofthe seriesandincludingthenextsamplefollowingtheoriginalwindowintheseries.Thiscreatesanewwindow ofsamples,whichisaveraged.Eachtimetheaverageoverawindowiscalculated,itischeckedtocrossthe threshold value set for causing sag. Every time we shift the window by one sample each, we repeat the procedureoffindingtheaveragevalueoverthewindowandcheckingwhetheritcrossesthethreshold.Ifit is found to cross the threshold value continuously for a few succeeding average values that have been calculated,thenasagisdetected.Similarlyaswellisdetecteditcrossestheswellthresholdcontinuously.

 

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

B.MovingaveragewindowalgorithmusingRMS

s.i n

Themostcommontechniqueusedfordetectingpurposeisthecalculationoftherootmeansquare(RMS) value of the voltage supply. Here similar to the above method a window is used. All the samples in the window are initially squared, then their average is obtained and finally its square root is found. This obtainedRMSvalueischeckedtocrossthethresholdvaluesetforcausingsag.Werepeattheprocedureof calculatingtheRMSvalueanddoingthesagcheckingbyshiftingthewindowbyonesampleeach.Ifitis found to cross the threshold value continuously for a few succeeding average values that have been calculated,thenasagisdetected.Similarlyaswellisdetecteditcrossestheswellthresholdcontinuously.



lib .

as

Rewritetheeq.(1)tothesequence,shownasfollows

df .re

Where N is the number of samples and v (i) is the ith voltage sample. Real nestgroup.net is obtain if the window length N is set to one cycle. In practical application, the data window is sliding along the time sequenceinspecificsampleinterval.Inordertodistinguisheachresult,timeinstantstampslabeledKare addedtoRMSvoltageasindependentvariablei.e.,itmakesRMSvoltagetobeafunctionoftime.



w

.e d

The time stamp k is restricted to be an integer that is equal to or greater than 1. Each value from (2) is obtainedovertheprocessingwindow.RMScalculationcanalsobeperformedbyconsideringsamplesfor onehalfǦcycle.ThewindowlengthhastobeanintegermultipleofoneǦhalfcycle.Anyotherwindowlength willproduceanoscillationintheRMSplotwithafrequencytwicethefundamentalfrequency[1][3].

w

w

Themainadvantageofthistechniqueintermsofcalculation,itissimple,fastandmuchsensitiveinsags andswells.But,thedrawbacksofthistechniqueitisdependenceonthesizeofthesamplewindow.Asmall windowmakestheRMSparameterlessrelevant,asitfollowsthetendencyofthetemporalsignal,andloses themeaningofmeanvalueofpower.Thebasicideaistofollowthevoltagemagnitudechangesascloseas possibleduringthedisturbingevent.ThemoreRMSvaluesarecalculated,thecloserthedisturbingeventis represented.RMScalculationcanalsobeperformedbyconsideringsamplesforonehalfǦcycle.Thewindow lengthhastobeanintegermultipleofoneǦhalfcycle.Anyotherwindowlengthwillproduceanoscillation intheplotwithafrequencytwicethefundamentalfrequency[4].

C.DetectionbyComparingWithAReferenceWindow Thismethodutilizestwowindows,onewindowforthesamplesofoneoftheinputvoltagecyclesandother for samples of normal reference cycle. Here the magnitude of a sample at any one sample point of the currentinputcycleiscomparedwiththatofthesampleatthesamesamplepointonthereferencewave. Thedifferencebetweenthemischeckedtocrossthethresholdsetforcausingsag.Thisisrepeatedforafew successive samples in a cycle. If the difference in values between them crosses the threshold values that havebeensetforcausingsag,thatparticularcyclewillbedetectedtocausesag.Nowthewindowisshifted to the same sample points of the next input cycle. If it continuously crosses the sag threshold for a few

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

s.i n

succeeding cycles, sag is finally detected. Similarly if the difference in magnitudes between the two windowscrossestheswellthresholdcontinuouslyforafewsucceedingcycles,aswellisdetected.

df .re



w

.e d

lib .

as

Fig.5Windowinthereferencecycle



w

w

Fig.6Windowmovingfromonecycletoanotherintheinputvoltage aftercomparingthesamesamplesofeachcyclewiththereferencecycle

III.AlgorithmsonPIC32Microcontroller

A 110v RMS – 50 Hz supply is used. Each cycle is 20mseconds and it is divided into 32 equally spaced samples. Hence each half cycle holds 16samples. Swell is to happen if there is rise in RMS voltage above 120v for at least 320milliseconds. Sag is to happen if there is fall in RMS voltage below 100v for at least 320milliseconds.PIC32EthernetStarterwithPIC32MX795F512Lmicrocontrollerisused.

A.MovingAverageWindowAlgorithm Thewindowsizeistakenas16orahalfcycle.Asexplainedearlier,theaverageofsamplesoverawindowis calculated. This is repeated by moving the window by one sample. Average value obtained over every window is compared with the threshold value. This is repeated so as to obtain the total number of occurrencesofsagandswell.Ifthecalculatedaveragevaluesremainlessthanthatof100vrmsfor320msec

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

ormorecontinuously,thenitiscalledsag.Ifthecalculatedaveragevaluesremainmorethanthatof120v rmsfor320msecormorecontinuously,thenitiscalledaswell.

B.MovingAverageWindowAlgorithmusingRMS The window size is taken as 16 or a half cycle. The RMS of samples over a window is calculated. This is repeatedbymovingthewindowbyonesampleeach.RMSvalueobtainedovereverywindowiscompared withthethresholdvalue.Thisisrepeatedsoastoobtainthetotalnumberofoccurrencesofsagandswell. IfthecalculatedRMSvaluesremainlessthanthatof100vRMScontinuouslyfor320msecormore,thenitis called sag. If the calculated RMS values remain more than that of 120v RMS for at least 320msec continuously,thenitiscalledaswell.

s.i n

C.DetectionbyComparingwithAReferenceWindow

df .re

Hereawindowsizeof16samplesistaken.Magnitudesofeachsamplewithinthewindowofinputcycleare compared with the samples that are at the same position in the reference cycle. It is checked if the differenceinthemagnitudescrossesthethresholdvaluessetforcausingsag.Ifitcrossesthesagthreshold continuouslyforatleastthesamplesof32cycles,i.e.320msecitcausessag.Similarlyifitcrossestheswell thresholdcontinuouslyfor320msec,itcancauseswell.Thisisrepeatedtofindthetotalnumberofsagsand swellsoccurred.

IV.ComparisonResult

w

w

w

.e d

lib .

as

Timetakenbythefirstmethodis11.2usecandthatbythesecondmethodis37.4usec,whilethattakenby thethirdis16,2usec.Asthetimetakenbyandcodesizeofthesecondalgorithmismaximum,itconsumes themaximumpower.Inthefirstandsecondmethodthedetectionofsagorswellisdelayedbyahalfcycle whileinthethirdcase,itisdetectedinthepresentcycleitself.AllthethreealgorithmscantakeinDirect Current,whileitrequiresarectifierifA.C.isgiventheinput.Thethirdalgorithmneedsanextrahardware todetectthezerocrossingthatdeterminesthe16samplesofahalfcycle.  A B C Algorithm ProgramMemory 1.196KB 2.043KB 1.251KB DataMemory 125Bytes 197Bytes 68Bytes Time 11.2usec 37.4usec 16.2usec PowerConsumed Minimum Maximum Medium DelayinDetection 1HalfCycleDelay 1HalfCycleDelay NoDelay 33HalfCyclesNeedto 33HalfCyclesNeedto 32HalfCyclesNeedto For32HalfCyclestobe DetectedasSwell beSwell beSwell beSwell D.C D.C D.C InputVoltageType A.CNeedsRectifier A.CNeedsRectifier A.CNeedsRectifier NeedofZeroCrossing No No Yes Hardware  Table1ComparisonofthreealgorithmsusingPIC32

V.MatlabSimulation A 110v r.m.s – 50 hz supply is used. Each cycle is 20mseconds and it is divided into 32 equally spaced samples. Hence each half cycle holds 16samples. Swell is to happen if there is rise in RMS voltage above 120vfor320milliseconds.SagistohappenifthereisfallinRMSvoltagebelow100vfor320milliseconds.In ordertodetectsagandswell,theinputvoltageisgivenasacombinationofanormalvoltage,i.e.110vRMS,

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

s.i n

aswellcausingvoltage,i.e.inputgreaterthan120vRMSandasagcausinginput,i.e.;voltagelessthan100v RMSfor1secondtime.



df .re

Fig.5Inputsignalgivenforsimulation

VI.SimulationResults

w

w

w

.e d

lib .

as

Timetakenbythefirstmethodis49usandthatbythesecondmethodis95us,whilethattakenbythethird is128us.Inthefirstandsecondmethodthedetectionofsagorswellisdelayedbyahalfcyclewhileinthe thirdcase,itisdetectedinthepresentcycleitself.Aswellrequiresacontinuousriseinvoltagefor320msec or more, which counts up to 32 half cycles. But first and second method needs an extra swell or sag half cycle for its detection. Thus it needs 32+1 low voltage half cycles to detect sag or 32+1 high voltage half cyclestodetectaswell.Thethirdmethoddoesnotrequirethis.Ittakescorrect32halfcyclesonly.[6]

 Table2ComparisonofthreealgorithmsinMatlab

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ


3URFHHGLQJVRIWKH,QWO&RQIRQ,QQRYDWLYH7UHQGVLQ(OHFWURQLFV&RPPXQLFDWLRQDQG$SSOLFDWLRQV,&,(&$

VII.Conclusion Threealgorithmsbasedonawindowingtechniquearedescribedtodetectthenumberofsagsandswells thathasoccurredininputvoltage.ThethreedescribedalgorithmsarecomparedafterrunninginMpLab onto a PIC32 microÇŚcontroller. Also a comparison after simulating in Matlab is also made. In Matlab the secondalgorithmtakeslessertimethanthatofthird,whileitistheoppositeonPIC32.Thereasonisthat squaringtakesmoretimewhenrunningonahardwareasmultiplicationisdoneusingrepetitiveadditions, unlikethatdoneonMatlab.

References

s.i n

[1]SureshKambleandDr.ChandrashekharThorat,“ANewAlgorithmforVoltageSagDetectionâ€?,IEEE– InternationalConferenceonAdvancesinEngineering,ScienceandManagement(lCAESMÇŚ2012)March30, 31,2012

df .re

[2]J.BarrosandE.Perez,â€?ImitationsintheUseofR.M.S.ValueinPowerQualityAnalysisâ€?,IMTC2006– InstrumentationandMeasurementTechnologyConferenceSorrento,Italy,24ÇŚ27April2006 [3]SureshKambleandDr.ChandrashekharThorat,“CharacteristicsAnalysisofVoltageSaginDistribution SystemusingRMSVoltageMethodâ€?,ACEEEInt.J.onElectricalandPowerEngineering,Vol.03,No.01,Feb 2012

lib .

as

[4]K.Daud,A.F.Abidin,N.Hamzah,H.S.NagindarSingh,“NewWindowingTechniqueDetectionofSags andSwellsBasedonContinuousSÇŚTransform(CST)â€?,InternationalJournalofNewComputerArchitectures and their Applications (IJNCAA) 2(4): 531 ÇŚ549 The Society of Digital Information and Wireless Communications(SDIWC)2012(ISSN:2220ÇŚ9085)

.e d

[5]KaiDing,K.W.E.Cheng,X.D.Xue,B.P.Divakar,C.D.Xu,Y.B.Che,D.H.Wang,P.Dong,“ANovelDetection Method for Voltage Sags�, 2006 2nd International Conference on Power Electronics Systems and Applications.

w

w

w

[6]PraizMadappillil,AnoopThomas,“PowerQualityDetectionAlgorithmsUsingWindowingTechnique�, NationalConferenceOnEmergingTrendsInComputing2013.

ZZZWHFKQRIRUXPFR

3DJHRI

ZZZDVGIUHVLQ

Icieca0001  
Read more
Read more
Similar to
Popular now
Just for you