8/2/2019 paper4 -

1/12

FuzzyLogicControlinAutonomousRobotics:

InvestigatingtheMotorolaMC68HC12onaLineFollowingRobot

DavidOlsen

DepartmentofElectricalandComputerEngineering

UniversityofMinnesotaDuluth

1023UniversityDrive

Duluth,MN55812.USA

FacultyAdvisor:Dr.MarianS.Stachowicz

Abstract

Autonomousrobotsystemsrequirecomplexcontrolsystems.FuzzyLogic,amathematicalsystemdevelopedbyProfessorLotfiZadeh,helpstoreducethecomplexityofmodelingnonlinearproblems.Inthe1990s,MotoroladevelopedtheMC68HC12microcontrollerwithnativeFuzzyLogicinstructions.ThisresearchdeterminedtheeffectivenessoftheMC68HC12sFuzzyLogicinstructionsforroboticcontrol.ThisresearchinvolveddesigningaroboticplatformusingtheMC68HC12andtestingbinarylogiccontrolsystemsagainstFuzzyLogiccontrolsystems.Theresearchanalyzedthetwosystemsusingfourcriteria:(1)thesizeofmemoryrequiredtodevelopthecontrolsystem,(2)theeaseofwritingthecontrolsoftware,

(3)howwellthecontrolsystemmanagedthefunctionsoftherobot,and(4)theoverallprocessingpowerofthesystem.TheresultsshowedthatFuzzyLogicuseslessmemorythanbinarylogicandismucheasiertodesign,althoughmoredifficulttoprograminitially.FuzzyLogiccancontrolmorefunctionsoftherobotandhasgreaterprocessingcapabilities.Thepower,easeofuse,andsmallsizeofFuzzyLogicinstructionsma

eFuzzyLogicapracticalsolutiontoautonomousroboticcontrolsystems.

Keywords:FuzzyLogic,robotics,controlsystems,HC12

1.Introduction

Theexpansionofroboticsandmicrocontrollersintothefacetsofeverydaylifeincreasestheneedtodevelopefficientcontrolsystems.Anon-traditionalapproachtocontrolsystemdesignistheuseofFuzzyLogic.

FuzzyLogicextendsfromthetraditionalcrispboundariesofAristotelianlogic(trueorfalse)toincludetheconceptofpartialtruthhavingtruth-valuesbetweencompletelytrueandcompletelyfalse.Dr.LotfiZadehofUniversityofCali

8/2/2019 paper4 -

2/12

orniaBer

eleyfirstintroducedthesefuzzymethodsin1965[1].Thesemethodsallowtheengineertousenaturallanguagetodescribeandimplementthecontrolsystem.FuzzyLogicuseslinguisticvaluestorepresentpartoftherangeanordinarycrispvariablemayassume[2].Forexample,avariablet,thatrepresentstemperature,mayvaryfrom0oCto100oC.Alinguisticvalue,COLDmaybeusedtorepresenttemperaturesfrom0oCto10oCwhileotherlinguisticvaluesrepresentotherranges.FuzzyLogicma

esitpossibletosolvecomplex,ill-definedproblemswherethereisalargedegreeofexpert

nowledgeorthesolutioniseasytodescribelinguistically[3].

ApplicationsofFuzzyLogicareappearinginmanyindustries.FuzzyLogicenablesdesignerstomodelcomplexsystemsmorequic

lyandeffectivelythantraditionalapproaches.Consumerappliances,automobileengines,transmissions,andindustrialsystemsareallusingFuzzyLogictechniques.

MotorolasHC68HC12(HC12)microcontrollerincorporatesseveralFuzzyLogicprimitivesdirectlyinitsinstructionset.TheinstructionsetcontainstheFuzzyLogicoperationsoftrapezoidalmembership,ruleevaluation,andweightedaveragedefuzzification.ThemicrocontrolleralsoincludesotherinstructionsthatarehelpfulinFuzzyLogicapplicationssuchasMIN/MAXinstructionsandtableloo

ups[4].MotorolasHC12allowsthedevelopmentoflow-levelapplicationsthatcanutilizetheuniquefeaturesofFuzzyLogic.

Thegoalofthisprojectistodesignandbuildanautonomouslinefollowing

robotbasedonFuzzyLogictechniques.TherobotusestheMotorolaHC68HC12microcontroller.ThisprojectinvolvesinvestigatingtheHC12sFuzzyLogicinstructionsetandanalyzingitsabilitytocontrolanautonomousrobot.Therobotinthisprojectservesasatestbedforseveralpiecesofsoftware.Thesesoftwareprogramsincludeinitialtestscripts,severalcontrolsystembasedontraditionallogic,andseveralcontrolsystemsbasedonfuzzysystems.Thebestclassicallogicandfuzzycontrolsystemsarecomparedinvarioustests.Thesetestsinvolvedtherobotfollowingablac lineonthefloor.

2.FuzzyLogicontheHC12

Asstatedintheintroduction,MotoroladesignedtheHC12withadvancedcapabilitiestohandleFuzzyLogiccalculations.TheHC12containsfourinstructionsthatarespecifictoFuzzyLogic.Theseinstructionsare:

MEMEvaluatesthetrapezoidalmembershipfunctionsREV/REVWPerformsunweighted/weightedMIN-MAXruleevaluationWAVPerformsweightedaveragedefuzzification

ThefollowingsectiondescribesthebasicsofFuzzyLogicontheMotorolaHC12.ThissectionwillonlyserveasanintroductiontofuzzyprogrammingontheHC12andisnotareplacementforthemanufacturesdocumentation.Thissectionassumesthatthereaderhassome

nowledgeofFuzzyLogic.

2.1fuzzylogicbasics

8/2/2019 paper4 -

3/12

ThedesignofaFuzzyLogicinterfacefortheHC12consistsoftwoparts.First,theusermustdesigna

nowledgebasethatcontainsthemembershipfunctionsandtheruleset.Thesecondpartistheinference

ernelthatta

esthesysteminputsandproducesthesystemoutputsbasedonthe

nowledgebase[5].Figure1showsthebasicstructureoftheFuzzyLogicsystem.

Figure1:Bloc

diagramofafuzzylogicsystem[5].

2.2fuzzificationstrategy(MEM)

Fuzzificationistheprocessbywhichsysteminputsareevaluatedtodeterminethedegreeatwhichtheybelongtoaparticularfuzzyset,onascalefrom00toFFinhexadecimal[5].TheMEMinstructionevaluatestrapezoidalmembershipfunctions.Thesefunctionsdefinethefuzzyset,thefoundationofFuzzyLogic.Afuzzysetisasetwithoutacrisp,clearlydefinedboundary.Itcancontainelementswithonlyapartialdegreeofmembership.Forexample,themembershipfunctionforCOLDcouldequalFFfortemperaturesbelow7oCandslopedownto00toward10oC.TheMEMinstructioncomparesthesysteminputagainstthemembersh

ipfunctiontodeterminethedegreeoftruthofafuzzyinput.

They-axisinFigure2representsthedegreeoftruthfromcompletelyfalse(00)tocompletelytrue(FF).Thex-axisrepresentstherangeofinputvaluesfortheparticularsysteminput.TheMEMfunctionwor

sbyfindingthey-value,givenasysteminput(x-value)andthemembershipfunction.MEMreturnsthepercentageoftruth(y-axisvalue)fortheparticularfuzzyset.Theresultisasetoffuzzyvaluesthatdescribecharacteristicsofinputvariablesinthesystem.

Figure2:Trapezoidalmembershipfunction.Thex-axisrepresentstheinputrangeandthey-axisrepresentsthedegreeoftruth[5].

TodefineatrapezoidalmembershipfunctionfortheHC12,youneedfourvalues:(1)thestartofthetrapezoid,(2)thefirstslope,(3)secondslope,and(4)theendpointofthetrapezoid.Figure2labelsthesepointsandshowsthememoryrepresentation.Theusershoulddefinethetrapezoidalinmemoryasfollows:

LABEL_MFDC.B$40,$D0,$08,$04

Theprogramshoulduseadescriptivelabelbecausetheprogrammerwillneedthislabelduringfuzzification.Forexample,COLD_MFcouldrepresentamembershipfunctionofthefuzzysetCOLDwherethepostfix_MFremindstheprogrammerthatthisvariablerepresentsamembershipfunction.Atrapezoidaldefinitionisneededforeachmembershipfunction.

8/2/2019 paper4 -

4/12

2.3ruledefinitionandevaluation(REV/REVW)

RuleevaluationishowFuzzyLogicperformscalculations.ThefuzzyvaluesproducedbytheMEMfunctionarepassedthroughtherulelisttofindthefuzzyoutput.ThetwotypesofrulesthattheHC12allowsareweighted(REVW),whereeachrulecanhaveadifferentweights;andun-weighted,(REV)wereallruleshaveequalweight.Anexampleofarulelist:

IftemperatureisCOLDandwindisHIGH,thenheatisonHIGH.

IftemperatureisWARMandwindisLOW,thenheatisonLOW.

IftemperatureisHOTandwindisLOW,thenheatisOFF.

AfterthefuzzyinputsareevaluatedwithREV/REVW,thesystemsfuzzyoutputsindicatethedegreetowhichanoutputshouldhaveaspecificvalue.Theseoutp

utsmustthenundergodefuzzificationbeforetheirvaluesareuseful.Creatingtherulelistisactuallyverysimple.Theantecedents(leftsideoftherule)arethefuzzyinputscreatedbytheMEMinstruction(e.g.,atemperaturereadingevaluatedwiththeCOLD,WARMandHOTmembershipfunctions).Theconsequents(rightsideoftherule)arethefuzzyoutputsofthesystem.DuringREV,eachantecedentisjoinedusingthefuzzyandoperator(MIN).Thisminimumvalueiscomparedtothecurrentfuzzyoutputofeachconsequentusingthefuzzyoroperator(MAX),andthemaximumofthesetwovaluesisstoredineachconsequent(fuzzyoutput).Inotherwords,theoveralltruthofaruleisstoredinthefuzzyoutputsandifasubsequentruleistruer,thenthefuzzyoutputsareupdatedtoreflectthisnewvalue.

Therulelistisstoredinmemoryasalistofpointerstofuzzyinputs(ante

cedents),areservedseparatorvalue,alistofpointerstofuzzyoutputs(consequents),andthenanotherseparator.Eachrulefollowsthispatternandtherulelististerminatedbyanendrulereservedvalue.

RULE_LISTDC.BP_TisCold,P_WisHigh,SEPARATOR,P_HeatHigh,SEPARATOR

DC.BP_TisWarm,P_WisLow,SEPARATOR,P_HeatLow,SEPARATOR

2.4defuzzificationstrategy(WAV)

ThefinalstepintheFuzzyLogiccalculationisdefuzzification,whentherawfuzzyoutputsareevaluatedtocreateacompositesystemoutput.Unli

etheinput,thefuzzyoutputmembershipfunctionisnottrapezoidalbutasingleton.Thissingletonindicatesonesystemoutputvalueforeachfuzzyoutput.Theoutputmembershipsingletonsarearraignedinmemoryinthesameorderastheircorrespondingfuzzyoutputs.WAVcalculatesasumofproductsofeachfuzzyoutputva

8/2/2019 paper4 -

5/12

luetimesitssingletonvalueandasumofallofthefuzzyoutputvalues.ThefirstsumisdividedbythesecondusingEDIVtoproduceanoverallvaluethatisthedefuzzifiedoutputofthesystem.Defuzzificationcreatesaweightedaveragesystemoutputbasedonthetruthofthefuzzyoutputs.

2.5fuzzyinference

ernel

Theinference

ernelcontainsalloftheinstructionsthatma

eupthefuzzysystem.The

ernelutilizesthe

nowledgebasetocreateasystemoutputfromgivensysteminputs[5].Allofthefuzzyinstructionsrequireproperinitializationoftheaccumulators,indexregisters,andfuzzyvalues.Thesedetailsarebeyondthescopeofthispaperandcanbefoundinthemanufacturesdocumentation.Thesampleprogramlisting,AppendixA,alsoincludescommentsondevelopingthefuzzy

ernel.

ReaderswantingmoreinformationareencouragedtorefertotheFuzzyLogicSupportchapterintheMotorola68HC12CPU12ReferenceManualforadetailedexplanationoftheFuzzyLogicinstructionset.

3.ControlSoftware

TotestifFuzzyLogicisasuperiorsolutiontotheproblemsofautonomousroboticcontrol,thisprojectinvolvedcreatingtwocontrolsystems.Thefirstsystemusestraditionallogictocontrolthesystem.ThissystemdoesnotuseanyofthemicrocontrollersembeddedFuzzyLogicfunctions.ThesecondrobotreliessolelyontheHC12sFuzzyLogicinstructions.ThissecondsystemusesFuzzyLogicforallofitscontrolprocessing.BothcontrolsystemsonlyusedtheHC12son

boardRAMarea(512bytes)forprogramstorage.

3.1traditionalcontrolsystem

ThetraditionalcontrolsystemdoesnotuseanoftheFuzzyLogicinstructionsontheHC12.Itreliesontestingeachinputastrueorfalseandthenusesanif-elseprogrammingstructuretodeterminethecorrectsystemoutput.BecauseofthelimitedsizeofRAMavailableforprogramming,thebinaryrobotreliesonanexternalchiptoconverttheanalogsignalsfromthesensorstoabinary,ono

roff,form.Thecontrolstructureissimplifiedtoreducethenecessaryamountofmemoryneededforstorage.Thetraditionalcontrolsystemusesthefollowingrules:

CentersensoronGoStraightLeftsensoronTurnRightRightsensoronTurnLeftNosensorsonStop

8/2/2019 paper4 -

6/12

Themotorcontrolofthissystemislimitedtoonespeedforstraight,left,andright.Becausethesystemreliesonanexternalchiptoconverttheanalogsensorvalues,itisunabletodetectwhenthesensorispartiallyontheline.Whilethissimplifiesprogramming,itreducestheeffectiveresolutionofthesensors.

3.2fuzzycontrolsystem

ThesmallamountofRAMalsolimitsthefuzzycontrolsystem.However,thefuzzyinstructionsareabletohandlefarmorecontrolsystemcomputationsthanthetraditionalsysteminasimilaramountofmemory.First,thefuzzysystemusestheanalogvaluesfromthesensors.Thisgivesthefuzzyrobotthebenefitofbeingabletotellexactlywhenthesensorisleavingthelineandtobemorerobusttochangesintheenvironment.Second,thissystemhasthreelevelsofspeedforeachdirection.Thisgivesthefuzzycontrolsystemimprovedturningability.

Thefuzzycontrolsystemta estheanalogvaluefromthesensorsandassigns

aleveloftruthtoafuzzyvalue.Thiscontrolsystemusedtwofuzzymembershipfunctionsforeachsensor:ONandOFF,tocreateatotalofsixfuzzyinputvalues.

Fuzzycontrolsystemsarebasedonrules.Thecontrolsystemusestheserulestoevaluatethefuzzyvaluesandcreateafuzzyoutput.Table1showstherulesbaseduponthetruthofthefuzzyinputvaluesandthecorrespondingfuzzyoutputsusedinthiscontrolsystem.Inthetable,thereareONandOFFfuzzyvaluesforeachofthethreesensors.ThesystemcombinesthetruthofthethreefuzzyinputsforeachruleusingthefuzzyMINoperator.Thisminimumvalueisstoredinthefuzzyoutputsunlesstheoutputsalreadyhaveagreatervaluestoredfromapreviousrule.Thesystemfindstheoutputthatbestfitthefuzzyinputs.

Table1.FuzzyRuleList.Thistableshowstherulelistthatevaluatestheoutputspeedanddirectiongiventhefuzzyinputvariables.ThisrulelistisevaluatedusingtheREVinstructionontheMC68HC12.

FuzzyInputsFuzzyOutputsLeftCenterRightSpeed

DirectionOffOffOffStopLineFinderOnOff

8/2/2019 paper4 -

7/12

OffSlowHardLeftOnOnOffMediumLeftOnOnOnSlowLineFinderOffOnOffFastStraightOffOn

OnMediumRightOffOffOnSlowHardRightOnOffOn

StopLineFinder

4.HardwareImplementation

Thisprojectusesacustomrobotdesignedtotestthevariouscontrolsystems.Thisrobothasalinesensorandtwomotors.TherobotcanalsohandleIRdistancesensors,butthedistancesensorsarenotusesinthetestingportionofthi

sproject.

4.1baseandmotors

Thetestrobotconsistsofaplywoodbaseandusesdifferentialsteering.Themotorsaretwoservos,modifiedtoallowcontinuousrevolution.Servosareusefu

8/2/2019 paper4 -

8/12

lforsmallrobotsbecausetheservohasanintegratedmotorcontrolleranditprovidesadequatetorqueforasmallrobot[6].Ballcasterssupportthefrontandbac

oftherobot.Thisallowstherobottoturnonitscenter.Arechargeablebatterypac

powersthemotors,linesensor,andthemicrocontroller.

4.2linesensor

ThecustomlinesensorisacollectionofthreeIRphototransistor,transmitterpairsthatsensewhentherobotincenteredontheline,totheleft,andtotheright.ThesesensorsfeedanalogdataintotheHC12sA/Dportsforprocessing.TheanalogvaluesrepresenttheIRreflectivityofthesurfacebelowthelinesensoratthreepoints.BecausethesensorcontainsIRtransmitters,sensorwor

swithoutambientlighting.

5.Testing

Thetworobotcontrolsystemsweretestedonavarietyoftrac

s.Thesetrac

swerecreatedbyplacingblac

electricaltapeonawooden,carpeted,andcementfloor.Thetrac

sincludedastraightline,a45-degreeangleturn,a90-degreeangleturn,anda12-inchradiusovalshapedring.Thelightinglevelwas eptconstantthroughoutthetrials.Thetestingmeasurementsincludeaqualitativedescriptionoftherobotsabilitytostayonthelinewithoutwiggleorovercorrection.

6.Results

Foreachofthetesttrac

s,bothrobotsmaintainedcontrolandfollowedtheline.Thetraditionalcontrolsystemdidhavemorewiggleasitmoveddownthestraightsection.Thisrobotwasunabletocenteritselfonthelineandinsteadzigzaggeddowntheline.Thefuzzyrobotsystemwasabletoquic

lyfindandholdthetruecourseoftheline.Thefuzzyrobotsystemalsomadefewercorrectionsduringtheturnsintheovaltesttrac

.Thetraditionalrobothadtocorrectitscoursemanymoretimesduringtheovalturns.Thefuzzyrobotsystemhadamuchsmoothercoursethroughoutthetrac

.Neitherrobotwasunabletoma

eanyoftheturnsinthetest.

Inadditiontocomparingattheperformanceofthetwodifferentcontrolsystems,thisprojectalsoloo

satthememoryrequirements,theeaseofwritingthecontrolsoftware,andtheprocessingpowerofthesystem.Forthetwosystemstested,theFuzzyLogicsystemusedmorememory.However,italsoprocessedsignificantlymoreinformationandgaveresultswithfinerresolution.Achievingthissamelevelofprocessingusingthetraditionalsystemwouldhavemadethetraditionalsystemsmemoryrequirementsmuchlarger.

Thefuzzysystemisinitiallymoredifficulttoprogrambecausethesyntaxoftheinstructionsandtheformatofthe

nowledgebaseinmemoryisunclear.Ho

8/2/2019 paper4 -

9/12

wever,itismucheasiertomodifythefuzzycontrolsystemtochangesintheenvironmentorsystemoutputsbecauseonlythemembershipfunctionsneedtochange.Forthetraditionalsystem,theentirelogicofthecontrolsystemmayneedtobemodified.ItismoreintuitivetoprograminFuzzyLogicbecauseoftheeaseofcreatingalinguisticsolutiontothecontrolproblem.Thisintuitivenaturema

esFuzzyLogiceasiertoprogramthantraditionallogicforcomplexcontrolproblems.

7.Conclusion

FuzzyLogicrepresentsatremendousadvancementinautonomousroboticcontrolsystems.TheMotorolaMC68HC12sinstructionsetwillma edesigningfuzzysolutionsonmicrocontrollersmucheasierthaninthepast.FuzzyLogiciswellsuitedforcomplexcontrolproblemsli

eautonomouscontrol,butitmaybetoodifficultforsimplecontrolsystems.TheHC12sfuzzyinstructionssimplifythedesigncycleforcontrolsystemdesignersandprovideanalternativetotraditionalcontrolmethodologies.ThefuzzysystemdevelopedbyMotorolaisaverypowerfulsetofinstructionsthatwillhaveadramaticimpactoncontrolsystemsinmanyindustries.

8.References

[1]Reuss,RobertF,etal.FuzzyLogicControlinaLineFollowingRobot.http://www.cs.unr.edu/~simon/fuzzy/fuzzylogic.htm[01/22/2002].

[2]Kandel,Abraham,etal.FuzzyControlSystems.BocaRaton:CRCPress,1994.

[3]Stachowicz,M.S.andBeall,L.,"FuzzyLogicPac

ageforMathematica",Wolf

ramResearch,Inc.,2000.

[4]M.S.StachowiczandC.Carroll,"IntelligentSystemsonMotorola'sMicrocontroller:ATeamDesignWor

shop,"Proceedingsofthe2000InternationalConferenceonEngineeringEducation,Taiwan,August14-18,2000.

[5]MotorolaSemiconductor,CPU12ReferenceManual.MotorolaInc.,1999.

[6]McComb,Gordon.TheRobotBuildersBonanza.NewYor

:MCGraw-Hill,2001.

AppendixA.ExampleFuzzyAssemblyListing

ThefollowcodeexampleliststheoverallstructureofanassemblyprogramfortheHC12thatutilizesFuzzyLogic.Thissimplifiedexampleta

estwosysteminputs,temperatureandwindspeed;andproducesonesystemoutput,heat.Inputvariables,storedin$0800and$0801,havebeenscaledfrom0016toff16.Therangeoftheheatoutputisfrom0016toff16.Refertosection2fordetailsont

8/2/2019 paper4 -

10/12

heFuzzyLogicoperators.

ORG$0900;MemorylocationforFuzzyKnowledgeBase

;###FuzzyConstants###

MARKEREQU$FE;Usedtoseparaterulesections

ENDROEQU$FF;Usedtofinishtherulelist

;###RelativePointers###

P_COLDEQU$00;Pointerstotherelativelocationsofthefuzzy

P_WARMEQU$01;InputandOutputvariables

P_HOTEQU$02

P_CALMEQU$03

P_WINDYEQU$04

P_NoHtEQU$05

P_LowHtEQU$06

P_MedHtEQU$07

P_HiHtEQU$08

;###MembershipFunctions###

COLD_MFDC$00,$76,$00,$05;TemperatureMembershipfunctions

WARM_MFDC$56,$AC,$07,$07;Point1,Point2,Slope1,Slope2

HOT_MFDC$8C,$FF,$07,$00

CALM_MFDC$00,$60,$00,$07;WindMembershipfunctions

WINDY_MFDC$50,$FF,$07,$00

;###OutputSingletons###

No_Heat_FSDC$00;TheFuzzysystemwillta

etheweightedaverage

Low_Heat_FSDC$56;ofthesevaluesbasedonthetruthsofthe

Med_Heat_FSDC$AC;correspondingFuzzyoutputvariables

High_Heat_FSDC$FF

;###FuzzyVariables###

8/2/2019 paper4 -

11/12

COLD_FIDS$01;TheseinstructionsreservebytesfortheFuzzy

WARM_FIDS$01;inputandoutputvariables.

HOT_FIDS$01;(NOTE:Thisvariableorderingiswhatdeterminesthe

CALM_FIDS$01;relativelocationsstoredaboveandcanbethe

WINDY_FIDS$01;sameorderingasthemembershipfunctions.)

NoHeat_FODS$01

LowHeat_FODS$01

MedHeat_FODS$01

HHeat_FODS$01

;###RULEDefinitions###

RULE_START

DCP_COLD,P_WINDY,MARKER,P_HiHt,MARKER;Thissectiondefinestherulelist

DCP_COLD,P_CALM,MARKER,P_MedHt,MARKER;Therulesaredefinedasrelative

DCP_WARM,P_WINDY,MARKER,P_MedHt,MARKER;pointerstothelocationsof

DCP_WARM,P_CALM,MARKER,P_LowHt,MARKER;Fuzzyinputandoutputvariables.

DCP_HOT,P_WINDY,MARKER,P_LowHt,MARKER;MARKERandENDROarecon

stants

DCP_HOT,P_CALM,MARKER,P_NoHt,ENDRO;$FE,$FFrespectively

org$0803;MemorystartinglocationforInferenceKernel

;###Fuzzificationfortemperaturedata###

LDX#COLD_MF;Firstofthreemembershipfunctionfortemperature

LDY#COLD_FI;Locationoffirstfuzzyinputvariablefortemperatur

e

LDAA$0800;Locationoftemperaturedatainsystem

LDAB#$03;Numberofmembershipfunctionsfortemperature

TEMPFUZZ:MEM;Thisloopwillprocesseachofthethreemembershipfunctions

DBNEB,TEMPFUZZ;andcreatethethreefuzzyinputvariables

8/2/2019 paper4 -

12/12

;###Fuzzificationforwinddata###

LDX#CALM_MF;Firstoftwomembershipfunctionforwind

LDY#CALM_FI;Locationoffuzzywind(Theselasttwolinesareoptionalbecause

;AccumulatorXandYalreadyhavethecorrectvalues)

LDAA$0801;Locationofwinddatainsystem

LDAB#$02;Numberofmembershipfunctionsforwind

WINDFUZZ:MEM;Thisloopwillprocesseachofthetwomembershipfunctions

DBNEB,WINDFUZZ;andcreatethetwofuzzyinputvariables

;###RULEEVALUATION###

RULE:LDAB#$04;Firstyoumustclearoutthefuzzyoutputvariables

CLEAROUT:CLR1,Y+

DBNEB,CLEAROUT

LDY#COLD_FI;PointatstartofFuzzyInputvariables

LDX#RULE_START;Pointatstartofrulelist

LDAA#$FF;MustloadAAwith$FF

REV

;###DEFUZZIFICATION###

LDX#No_Heat_FS;Pointtosingletonpositions

LDY#NoHeat_FO;PointtoFuzzyOutputlocations

LDAB#$04;NumberofFuzzyOutputs

WAV;Calculatesumsforweightedaverage

EDIV;Finaldividestep

TFRY,D;MovetheresultstoA:B

STAB$0802;Storesystemoutputs