外文翻译智能超快速的低成本RISC实施镍镉电池充电器Word文件下载.docx

外文翻译智能超快速的低成本RISC实施镍镉电池充电器Word文件下载.docx

《外文翻译智能超快速的低成本RISC实施镍镉电池充电器Word文件下载.docx》由会员分享，可在线阅读，更多相关《外文翻译智能超快速的低成本RISC实施镍镉电池充电器Word文件下载.docx（26页珍藏版）》请在冰豆网上搜索。

Abstract

Thispaperpresentsalowcostreducedinstructionsetcomputer（RISC）implementationofanintelligentultrafastchargerforanickel–cadmium（Ni–Cd）battery.Thechargeremploysageneticalgorithm（GA）trainedgeneralizedregressionneuralnetwork（GRNN）asakeytoultrafastchargingwhileavoidingbatterydamage.Thetradeoffbetweenmeansquareerror（MSE）andthecomputationalburdenoftheGRNNisaddressed.Besides,anefficienttechniqueisproposedforestimationofaradialbasisfunction（RBF）intheGRNN.Hardwarerealizationbaseduponthetechniquesisdiscussed.ExperimentalresultswithcommercialNi–Cdbatteriesrevealthatwhiletheproposedchargersignificantlyreducesthechargingtime,itscarcelydeterioratesthebatteryenergystoragecapabilitywhencomparedwiththeconventionalcharger.

Keywords:

Batterycharger;

Ni–Cdbattery;

Fastcharging;

GA;

GRNN;

RBF

1.Introduction

Abatteryisanelectrochemicaldevicethatconvertschemicalenergycontainedinitsactivematerialsdirectlyintoelectricenergythroughanelectrochemicaloxida-tion–reductionreactions[1].Atpresent,asecondaryorrechargeablebatteryiswidelyusedinmanyapplicationsincludingwirelesscommunicationdevicesandportableappliances.Thistypeofbatteryincludesnickelcadmium（Ni–Cd）,nickelmetalhydride（Ni–MH）andlithiumion（Li-ion）batteries.Amongthese,Ni–Cdbatteryisthemostpopularrechargeablebatteryduetoitslowcost,highnum-berofcharge–dischargecyclesandexcellentloadperfor-mance[2].

Batterychargingisaverycrucialfactorinacontractedmanufacturing,wherealargenumberofbatterypoweredplug-and-playdevicesareoftenproduced.AfullychargedNi–Cdbatterymaybesuppliedbyabatterymanufacturer.However,inmanypracticalsituations,suchasshipmentdelayorfirmwareupgrading,thebatterymustberechargedbythemanufacturerbeforeshippingtoacustomer.Becauseofitssimplestructureandcostcompetitiveness,aconstanttricklecurrentchargestrategy[1,2]isoneofthemostcommonchargingtechniquesforaNi–Cdbattery.

Thetechniquerequiresalongchargetimesinceitemploysaverylowchargingcurrent,i.e.0.1CwhereCisthebatterycapacity,Sometimesthecharginglastslongerthan10h,andthus,thechargerisoftenreferredtoasan‘‘overnightcharger’’[3].Ofcourse,thisisnotasuitabletechniqueforthemanufacturingprocess.

Inordertoreducethechargingtime,amuchlargercur-rentthanthetricklecurrentmaybeutilizedinthechargingsystem.However,thismaycomeattheexpenseofbatterydamagebecausesupplyingahighcurrenttothebatteryforcertainperiodoftimemayresultinexcessiveriseintem-peratureandeventuallycancelachargingreactionofthebattery.AmoresophisticatedalgorithmwasintroducedinRef.[4].Specifically,bydeterminingthebatterychargestatus,thetechniqueselectsasuitablechargestrategysothatminimalchargingtimeisobtained.Besides,implemen-tationofthechargerbasedontheMotorolaTMmicrocon-trollerwassubsequentlypresentedinRef.[5].However,sincethechargerutilizesthemaximumchargecurrentof1C,thechargingtimewasstillapproximately1h.

Infact,chargingthebatterywithachargingcurrentofmorethan1Cispossible,seee.g.Refs.[2,6].Nonetheless,thechargingprocessmustbecarefullyundertakenother-wisetherapidriseinbatterytemperature

cancauseseverebatterydamage.Therefore,atemperaturesensorshouldbeemployedinpracticetomonitorthebatterytemperature.Duringthelastdecade,severalchargerdesignsrelyinguponanintelligentcontrollerhavebeenintroduced.InRef.[7],afastchargingalgorithmbaseduponarelation-shipbetweenthevoltageandtemperatureofthebatterywasproposed.AnothermethodthatutilizesanonlinearcontrollerwassubsequentlyintroducedinRef.[8].Inaddi-tion,fastchargingtechniquesthatrelyuponaneuro-fuzzycontroller[6]andanadaptiveneuro-fuzzyinferencesystem（ANFIS）[2]werepresented.

Basically,anintelligentchargerendeavorstocharacter-izeanonlinearrelationshipbetweenthebatterytempera-tureandvoltageandthebatterychargecurrent.Withthisknowledge,thesystemsalwaysprovidethemaximumallowablechargecurrenttothebatterythatavoidsdamage.Therefore,thechargingtimeisrelativelyshortasopposedtothatoftheconventionalcharger.Nevertheless,lowcostimplementationofthechargerisquestionable,mainlydue

tothecomputationalburdenrequiredforthechargealgorithm.

TheissueofthecomputationalburdenrequiredinanintelligentchargerwasrecentlyaddressedinRef.[9],whereanultrafastchargingtechniquefortheNi–Cdbatteryusingageneralizedregressionneuralnetwork（GRNN）wasintroduced.TheproposedchargeroutperformeditsANFIScounterpartintermsofbothprecisioninnonlineardynamicsapproximationanditsrelativelysimplestructure.Subsequently,aGaussianfunctionintheoriginalGRNNbasedchargerwasreplacedwitharadialbasisfunction（RBF）[10],leadingtoanotherfastcharger.Furtherreduc-tionofthecomputationalcomplexitywasachievedbecausetheRBFcouldbeefficientlyrealizedusingacompactsup-portradialbasisfunction（CSRBF）.TrainingoftheGRNNinbothRefs.[9]and[10]wasachievedbasedupontrialanderror.InRef.[11],ameanstotraintheGRNNusingageneticalgorithm（GA）waspresented.Itwasrevealedthatanimprovementofmorethan10-foldinmeansquareerror（MSE）wasattainedwiththeGAtrainedGRNNcontroller.Atanyrate,theworkpresentedinRefs.[9–11]onlyconcernedthebasicconceptoftheultrafast

chargerbasedonidealcomponentsimulationswithoutconcernfortheaspectofpracticalimplementation.

Thispaperprovidesafullimplementationaccountoftheultrafastchargerforrealapplications.Insteadofusingabulkypersonalcomputer（PC）,thechargerunitisembed-dedwithamicrocontroller.Inaddition,moreefficientpowerelectroniccircuitryisintroducedtominimizehard-warecomplexity.ThisprovidesalowcostRISCimplemen-tationofanultrafastchargingsystemfortheNi–CdbatteryusingtheGAtrainedGRNNcontroller.Besides,thepaperextensivelyexaminestheperformanceofthepro-posedchargerwithcommercialNi–Cdbatteries.

Theremainderofthepaperisorganizedasfollows.Followingthis,theultrafastchargingprinciplesarediscussedinSection2.AnultrafastchargerusingtheGAtrainedGRNNisintroducedinSection3.Hardwareimplementa-tionoftheproposedchargeristhendescribedinSection4.ExperimentalresultswithemphasisonperformanceofthechargerarepresentedinSection5.Finally,conclusionsaredrawninSection6.

2.Ultrafastchargingprinciples

2.1.Ni–Cdbatterycharacteristics

Fig.1illustratestheevolutionofbatterytemperatureandvoltagewhenabatteryissuppliedwithaconstantchargecurrent[2,5].Clearly,thebatterycharacteristicsexhibitstrongnonlinearities.Fromthefigure,thefollowingobservationsaremade.Initially,thebatterytemperature（T）graduallyincreasesastime（t）evolves,i.e.Thetemperaturegradient（dT/dt）isverylow.Asthebatteryapproachesafullychargedstate,thebatterytemperature,however,increasesveryrapidly.Actually,atemperatureriseof60Cmaybenoticedin1min,resultinginthetem-peraturegradientbeingashighas1C/s.Atthisstate,thebatterymayeasilybedamagedifchargingisnotstoppedorslowed.Toavoiddamage,thebatteryshouldnotbeallowedtogobeyond50C

[1].Inadditiontothetemper-ature,thebatteryvoltage（V）immediatelydropsor,equiv-alently,thevoltagegradientbecomes（dV/dt）negative.Thus,twoconditionscanbeutilizedasanindicationofthefullychargedstate,namely（i）thebatterytemperatureisgreaterthan50C（T>

50）or（ii）thevoltagegradientisnegative（dV/dt<

0）.

2.2.Ultrafastchargerdesign

Sinceitsinvention,variousapproacheshavebeenemployedforchargingNi–Cdbatteries.Oneofthemostpopularchargingtechniqueswidelyusedinconsumerproductsistheconstantcurrenttricklechargetechnique,inwhichthechargersuppliesaverysmall,constantcurrentrate（normally0.1C）tothebatteryandreliesonuserinter-ventiontostopthechargewhenthebatteryreturnstofullcapacity.Itisnotedthatchargingabatteryratedat1Ahwith0.1Cimpliesthechargingcurrentof100mAandresultsinthechargingtimebeingatleast10h.Reductionofthechargingperiodsimplyrequiresthechargingcurrenttobelarger.Inpractice,aNi–Cdbatteryiscapableofhan-dlingachargecurrentof8C[2].However,itcannotbesup-plied

withsuchalargecurrentforalongperiod;

otherwisethebatterywillbeseverelydamagedasdiscussedabove.Therefore,thechargingcurrentmustbeadjustedinaccor-dancewiththebatterychargestate.

Akeytoultrafastchargingisanintelligentcontrollerthat,givenmeasuredinputparameters,canproperlydeter-minetheoutput,i.e.thechargingcurrent（1c）.Apparently,thefirsttaskindesigningthecontrolleristoidentifythecontrolinputparameters,whichrepresentthebatterysta-

tus.Typically,thebatteryvoltageandtemperature,togetherwiththeirassociatedgradientsareusedasinputstothecontroller[2,6].Besides,thebatterycurrentmaybeusedtoestimatetheamountofchargeremaininginthebattery[5].Onemightexpectthatmoreparameterswouldprovidebetterperformance.Fromtheimplementationpointofview,however,moreparametersmeansmorecom-putationalcomplexityand,hence,morecomplicatedcir-cuitry.Inourdesign,onlythetemperatureanditsassociatedgradientare