基于CVaR风险度量的发电商竞价策略

湖南大学硕士学位论文基于CVaR风险度量的发电商竞价策略姓名：廖菁申请学位级别：硕士专业：电力系统及其自动化指导教师：江辉20070415CVaRIICVaR—VaRCVaRVaRCVaR(VaR)(CVaR)PoolBilateral90%95%VaRCVaRVaRCVaR90%95%CVaRVaRCVaRIIIAbstractRestructuringoftheelectricityindustrycanintroducecompetitiontogenerationcompanies,improverunningefficiencyandoptimizecollocationofresource.Howeveritwillexposetheparticipatingcompaniestophysicalandfinancialuncertainties.Withcompetitionintroducedtothegenerationside,thegenerationcompaniesconcernabouthowtogainmoreprofitsandtrytoreduceanyunnecessarylosses.Generationcompaniesmustbuildoptimalbiddingstrategiestomaximizeprofitswhendifferentconditionsandsituationsofgenerationcompaniesareconsidered.Thepaperfocusesontheriskofgenerationcompany’sbiddingstrategy.Basedontheriskmanagementtheory,ConditionalValueatRiskisusedasriskmeasurementindexforgenerationcompanyinitsbiddingprocess.Firstly,thispapergivesacomprehensiveintroductiontocharacteristicsofelectricitymarket,trademodel,aswellastradestyleandmarketrulesofelectricitymarketstrategicbidding.Andthen,thegenerationcostandtherelevantresearchworksinstrategicbiddingareanalyzed.Secondly,theriskmanagementmethodsofgenerationcompaniesarediscussed,andthebiddingstrategiesofgenerationcompaniesarealsoanalyzed.Thirdly,theexistingresearchaboutriskassessmentisintroduced;theshortcomingoftraditionalriskmeasurementisanalyzed.Andtwonewmethodsofriskassessment—ValueatRisk(VaR)andConditionalValueatRisk(CVaR)areintroduced.ThentheshortcomingsofVaRarepointedoutandthereasonthatCVaRmethodcanmeasureriskmoreeffectivelyisexplained.Finally,tworiskassessmenttechniques—ValueatRisk(VaR)andConditionalValueatRisk(CVaR)areusedasriskmeasurementindexforgenerationcompaniesinthepaper,andtwomodesareestablishedfortherevenueofgenerationcompanyunderPoolandBilateral.ThemarketpriceissimulatedwithMonte-Carlomethod.Bycalculatingthegenerationcompany’sexpectprofit,VaRandCVaR(under90%confidentleveland95%confidencelevel)respectively,thegenerationcompany’sriskunderdifferentbiddingstrategyareanalyzed,andthedifferencebetweenVaRandCVaRmethodsarecompared.Meanwhile,wegotCVaReffectivefrontiercurveunder90%and95%confidencelevel.SimulationresultsshowthatVaRandCVaRmethodcanprovideeffectiveindexesofriskassessmentforgenerationcompanies.KeyWords:Powermarket;Biddingstrategy;Riskassessment;VaR;CVaRI1______2√111.1208019982002[2002]5200212291111CVaR21.2]1[3PX(PowerExchange)ISO(IndependentSystemOperator)TO(TransmissionOwner)AS(AncillaryService)SC(SchedulerCoordinator)]2[1.11.1CVaR4ISO(IndependentSystemOperator)ISOISO1.11.3VaR(ValueatRisk)CVaR(ConditionalValueatRisk)VaRCVaRVaRCVaRVaRCVaRVaRVaRCVaRVaRCVaR1.45[3~16][17]VaR[18]VaR[19]VaRCVaR[20]VaRVaR[21]CVaR6VaRVaRVaR[22]VaR[23][24]VaRVaRVaRCVaR[25~28]VaRCVaRVaRCVaR[29~33][29][30,31]VaR[31]VaR[32]VaR[33]VaRCVaRVaRCVaRPoolBilateralVaRCVaR1.5—VaRCVaR(1)(2)7(3)—VaRCVaR(4)PoolBilateralVaRCVaR(5)VaRCVaR(90%95%)Pool+BilateralVaRCVaR90%95%VaRCVaRVaRCVaRCVaR822.1(Pool)+(Pool+Bilateraltransaction)+2.1.1(Pool)(Poo)Pool2001PoolPJMVictoria[34,35]Pool2.1.2+(Pool+Bilateraltransaction)Pool9P+B+(Pool)(Bilateraltrades)[36][2002]52.1.3+(brokeragesystem)(multilateraltransaction)[37]20013(NETA)(NewElectricityTradingArrangement)[38,39]Pool+NETA97%NordPool71%25%199820012002[40,41]2.2()()CVaR10PJM(Pool)()NETA97%-98%2.2.148(24)(30min1h)482.1[42]2.2.211N-12.1AGC2.2()[43]2.22.2.3CVaR12NETA(Over-the-Counter)PX[11][44]2.2.4AGC2.3(AGC)()2.3AGC13(1)AGC-(2)AGCAGC(3)15min[45][46][47]2.32.3.1(GenerationCost)(2.1)iiiiiiicQbQaQC++=221)((2.1)iCiiQiiaibici2.4CVaR14MW/h)2.42.3.2(AverageCostAC)iiiiiiiiiQcbQaQQCQAC++==21)()((2.2)iACi2.3.3(MarginalCostMC)iiiiiiibQadQQdCQMC+==)()((2.3)iMCi2.5ACMCMW//MWh)2.52.4152.4.1()[48,49]2.4.2—2.4.3(/)/10/2.4.4MCPPABMCP(MarketClearingPrice)MCPCVaR16()()MCPPAB(PayAsBid)MCP—PABPAB[50]MCPPABPAB2.5(Pool)(1)MCPMCPMCPMCPMCP17MCP(2)(3)CVaR1833.13.1.1(1)(2)(3)19(1)(2)(3)(4)3.1.2(1)(2)(3)(4)3.1CVaR203.13.1.33.1.3.1(1)(2)(3)3.1.3.2MBS(Mortgage-backsecurity)MBS()3.1.3.3(1)(2)21(3)3.1.3.43.23.2.1MarineMax·CVaR223.2.2(1)(2)(3)233.3(1)(2)(3)(4)CVaR24(5)(6)3.4(1)(2)()(VaR)(CVaR)—VaRCVaR254VaRCVaR[51]—VaRCVaR[52]VaRCVaRVaRCVaR—CVaRVaRCVaR4.1()H.Markowitz-(M-V)[53]---CVaR26-[54]P.Theodossiou−t[55]VaR(ValueatRisk)(GroupofThirty)1993[35]VaRVaRVaRVaRVaRCVaR(ConditionalValueatRisk)[56~58]VaRVaRCVaRCVaRVaR4.2VaR4.2.1VaRVaR(ValueatRisk)VaR[52]VaR0WRW)1(0RWW+=(4.1)27σµ,β)1(*0*RWW+=(4.2)*R*WrVaR)()(*0*µ−−=−=RWWWEVaRr(4.3))(WEaVaR*0*0RWWWVaRa−=−=(4.4)VaRrVaRVaR)(wf*Wβ*W*Wβ∫∞=*)(Wdwwfβ(4.5)*Wβ−=≤1)(*WwP)()(1**∫∞−≤==−WWwPdwwfβ(4.6))(wf],(*W−∞β−1*W(quartile)VaR4.195%(losingtail)5%VaR5%*W*W=-0.75)(WE=2.500=WrVaR*)(WWEVaRr−==2.5-(-0.75)=3.25aVaR0.750-1-2-3123456785%10%15%20%25%5%4.14.2.2VaRVaRCVaR284.2.2.1VaR95%VaR10000(1)1000010000(2)1000010000(3)(4)5%(5)(4)VaR(1)(2)(3)(1)(2)4.2.2.2[29]29VaR10000(95%)500(99%)100VaR(1)(2)(3)(2)(4)(2)(3)10000VaR4.2.3(1)(2)VaRnVaR(3)VaR(4)VaRCVaR30VaR4.3VaRCVaR4.3.1VaRVaR[25,26](1)(2)VaR(3)VaRVaR()(4)VaRVaR()VaRVaR99%VaR=10001%1000VaRVaRVaRVaRAB4.2ABA95%VaR—CVaR(ConditionalValueatRisk)31AB95%VaR4.2VaRCVaR4.3.2CVaRVaRUryasevSCVaR(Conditi