电力市场下水电厂短期优化运行研究

华中科技大学硕士学位论文电力市场下水电厂短期优化运行研究姓名：马建军申请学位级别：硕士专业：水利水电工程指导教师：伍永刚20060512IIIAGCAGCAGC-VaRValueatRiskAGCIVAbstractThecurrentmatureelectricitymarketsoverseasgenerallyincludelong-termcontractmarketandshort-termmarket,whichcomposingmultistagemarkets.Theshort-termmarketisoftencomposedofenergymarketandancillaryservicemarketandsoon.Theoptimizationdispatchproblemforhydropowerplantsturnsintothecoordinativedispatchprobleminthemulti-marketsintheelectricitymarket.TheAGCmarketisthemostimportantpartoftheancillaryservicemarket.Becauseofexcellentperformanceatstartandstopaswellasadjustmentforhydrounit,itisextremelysuitableforparticipatesintheAGCadjustment.Thispaperstudiesthecombinationoptimizationdispatchproblemforhydropowerplantsintheday-aheadenergymarketandAGCmarketandobtainssomevaluableresearchresults.Thispaperfirstintroducesthecurrentworldwideelectricitymarkets,pointingoutthemarketenvironmentforhydropowerplants.Thenthemarketriskfacedbyhydropowerplantsandthemethodsofriskassessment,whicharewidelyusedincurrentfinancialdomain,havebeenanalyzed.Thesearethefoundationsofmakingcoordinativescheduleinthefollowingpartsofthispaper.Solvingthecoordinatedispatchproblemisthemostimportantpartofthispaper.Asthedoubtofusingvarianceforriskassessment,ittakesthevalueatrisk(VaR)astheriskmeasurementindex.TheVaRandexpectedrevenueconstitutedifferentutilityfunctionstodescribemarketbehaviorofdifferentpowerproducer.Amodelforshort-termcoordinativedispatchinenergyandAGCmarketsisestablishedforhydropowerplants,whichisbasedontheforecastedMCPsforenergymarketandAGCmarket.Inthemodel,theconstrainofwaterusing,whichisobtainedbythemediumorlong-termdispatchplan,isconsideredasaweakconstrain,whichisproventobeeffectivetorevisethelongerplan.Inconsiderationofthedifficultyofchoosingappropriateriskfactorinutilityfunctionfordifferentpowerproducer,Monte-CarloMethodisusedforcombinedriskassessmentofpowerpriceandinflow.Theresultcanhelppowerproducerchoosemoreappropriateriskfactor(weightfactor),therebyimprovingthegenerateschedule.Acasestudyisgivenintheendofthispaper,whichisbasedontheThreeGorgesplantandthehistoricpricedataofNewEnglandinAmerican.VKeywords:Electricitymarkets;Hydropowerplants;Coordinativedispatch;Allocatingstrategy;Riskassessment;Valueatrisk(VaR);Utility_____111.1AGCAGC21.21.2.11[1]AGCAGCAGCAGC[2]1111max[()]..[()]()()()mnkttknkkckkkNtTstQNtWtQtNtR===⋅+=∈∑∑∑(1-1)()Qtt()cWtt()kNttk()kkQNkkNtTtkRk23111max{[()]()}..()()()mnkktctknkkkkQNtTWtstNtNtQtD===⋅+=∈∑∑∑(1-2)()Dkk()Ntt[3]412343441Litle1955[4]DynamicProgrammingwithSuccessiveApproximations-DPSAProgressiveOptimalityAlgorithm-POAIncrementDynamicProgramming-IDPDiscreteDifferentialDynamicProgramming-DDDP[5~8]2[9~11]35[12]4EastHall1994Faling[13]1997[14][15][16~17]2AGC-6[18]VaRCVaR2()RUERωσ=−(1-3)UR()ER2Rσωω0ωmaxU[19][20]pricetaker[21]pricetaker0/1[22~23][24][25]71.2.2[26]AGCTMSR10minspinningreserveTMNSR10minnonspinningreserveTMOR30minoperatingreserveAGC[27~29]Markowitz-[27][28][29]Markowitz-[30]VaRCVaR[31]CVaR4[32]8Markowitz-Markowitz-1.3AGCVaRAGC91022.1[33]2.2112.312PJM2.3.1POOLFERC20027StandardMarketDesignSMDSMD11990331PowerPoolPOOLPOOLPOOLPOOL[34]10MW500kWPOOLPOOLPOOLPOOL13POOL1POOLPOOLPOOL2POOL100.5h0.5hSystemMarginalPriceSMPLossofLoadProbabilityLOLPPOOLPoolPurchasePricePPPTableATableBTableAPPPUpliftPOOLPoolSellingPricePSPTableBPSPPPPPOOLPSP5000.5h30.5h2%PPPTableBPPPPSP1990200028%31%142001327NETANEWELETRICTITYTRADINGARRANGEMENTS[35]20012005NETAGBGreatBritainGBBritishElectricityTradingandTransmissionArrangementsBETTABETTA[36]2PJMPJM19971997PJMISO19981999FTR2000[37]1PJM2PJM10min10min30min3LMPHUBsHUBHUBHUBLMPLMPZONEZONELMPLMP4PJMLMP155PJMICAPLSEICAPX%X15%18%ICAPPJMISOICAPLSE6PJM7PJM8PJM$1000/MWh$100/MWh10%LMP9PJMLMPPJMPJM10PJM11PJMFTRFTRFTR319971999IndependentSystemOperatorISO2003163SMD10min10min30min4h110min30min24h1000$/MWhISO2.3.21200220025200220032004120041720045[38]2200411520044200511080%20%18231219100MWA390%42033.1[39~40]1.2.2Markowitz-AGCVaRAGCAGC3.2213.3AGC[41]AGC[42]AGCAGCAGCAGCAGCAGC10minTMSR22AGCISONERCNPCCNEPOOLFERCAGCISOAGCAGCAGCAGCPminSSPBPmaxPalPah03.1AGCAGCAGCAGCACC[43]AGCAGCAGCAGCAGCAGCAGCAGCAGCAGCAGCAGC23[44]AGCISO[45][46]AGCAGCAGCAGCAGCAGC[47]AGCAGC[46][44]AGC1AGCAGCBp1λ2AGCAGCAGCss2λ3AGCAGCBpp−+λppB−−λp4+λ1λbid112λλλ−=−1bidλ1λ−4λ+λ−AGCλ+1λ1λλ+=p∆11()bidpλλ−×∆p∆1bidpλ×∆pλ−×∆112bidλλλ−=−1bidλ1λ−λ1λ−3.4AGCAGC24Markowitz-RiskBenchmarkorReferenceLevelLPMnLPMLowerPartialMomentsLPMnLPMnVaRValueAtRisk25VaRVaR[48~49]t∆cP()1ProbPVaRc∆−=−(3-1)P∆Pt∆VaRVaRVaRVaRNegativeReturnSideVaRVaR3.5pricetakerAGCkAGCkpAGCAGCγAGCakkssγ=⋅kpBkpk2612120.50.5kkBkkkkkkkkBkkkrpssspsλλλγλγλλ−+=++⋅+⋅=+(3-2)kkt∆AGC()121TkkkkkRpsλλ==+∑(3-3)VaR()(,,)UERVaRRct=−(3-4)U)(RER(,,)VaRRcttcR*Rc()*VaRERR=−(3-5)*1()(*)RcfrdrPrR−∞−==≤∫(3-6))(RE)(rfRmaxU(3-7)1kkkkkpspps+≤≤−(3-8)ksvτ≤⋅(3-9)kpkpτAGCv272kkkkqqqqδ≤+≤(3-10)kqkqkqδAGC31()TkkkqqtWδ=+∆≤∑(3-11)Wt∆3-11AGCAGCAGCAGCVaR3-53-6c050%VaRRiskSeekingc50%0=VaRRiskNeutralityc50%100%RiskAversion3-5cVaR2111~(,)expNλλσ2222~(,)expNλλσ(,,)rVaRRcttασ=(3-12)TrWWσ=Σ(3-13)αcrσTnWWWW),,,(21=Σ[49]