水电厂优化调度决策支持系统研究

重庆大学硕士学位论文水电厂优化调度决策支持系统研究姓名：徐晓燕申请学位级别：硕士专业：电气工程指导教师：卢继平20050501IIIABSTRACTOptimaldispatchingsystemofhydropowerplantisacomplexproject.Nowthereisnotasystemic,perfectandripeschemetosolveit.Itmusttakethemarketmechanism’sinfectionintoaccountsoastoembodythepriceinfluence.Thesystemhavetheimportantsignificancetoresearchitthatcanimprovethereliabilityandstabilityindexofelectricnetwork.Furthermore,itcanoptimizetheuseofenergyresources,exploittheelectricpowermarketinallsidesandimprovethebenefits.Thepaperproposesthewholestructureandtherelationamongthemodulesofthedecisionsupportsystemofthehydropowerplantoptimizationdispatching.Theimportantmodulesareanalyzedinseparate,whicharetheforecastingsystem,theoptimumdispatchingoperationofhydropowerplantsystemandthemaintenanceschedulingsystemundertheelectricitymarketenvironment.ThepaperresearchesthequestionsofGeZhouBahydroelectricplantfromthetwosidesofeconomyandreliability.Theintegratedpracticalresolventisputtedforwardasanalyticresultsofeconomicaldispatchingoperationandmaintenanceschedulingofplant.Accordingtothemathematicalmodelsandcalculationmethodsbythispaperproposed,whicharebasedondynamicprogrammingalgorithmandgeneticalgorithm,thesoftwareofoptimumoperationofGeZhouBahydropowerplantandthemaintenanceschedulingarecomplied.TheeconomicoperationoftheplantsarecalculatedandtheresultsarecomparedthatindicatethemethodisefficientforeconomicoperationofGeZhouBahydropowerplant,andmoreover,theresultsarethebasisofpracticaleconomicaloperationinthefuture.Thereal-timebalancingbiddingtransactionisanindispensabletransactionforminelectricitymarketandithasdistinguishingfeatures.Thebiddingstrategyofpowergenerationcompaniesinthebiddingprocessbecomesthehotspotintheresearchofelectricitymarket.Forecastingthemarketclearingpriceisthebasisofdecisionmakingforeachparticipantinelectricitymarket.Undertheelectricitymarketenvironment,theorderofgenerationcompaniesisthelargestprofitandprofitsofgenerationcompaniesdepend,toalargeextent,onbiddingstrategiesemployed.Thepaperresearchesthedecisionsupportsystemforpricebiddingofjoint-stockunitinthehydropowerplantandintroducesitsmodulescharacteristicsindifferentpointsindetail,suchasthethoughtofdesignandthearithmeticanditsfunctionalcharacteristics.ThesystemisIIIusedforreferenceofthebiddingstrategyofunitinthehydropowerplant.Furthermore,thispaperpresentsthepricepredictionmodelbasedonradialbasicfunctionneuralnetworksandwavelettransform.Themodelisnotonlybetterthanthetraditiontechnicalanalysismethodbutalsoisavoidingthedefectswhichrelapseintothepartialsmallestpointandconvergencerateislittleofback-propagationalgorithm.Thesimulationresultsoftheexperimentshowthatmodelefficienttoforecastthetrendofmarketclearingprice.Keywords:optimumdispatching,maintenancescheduling,biddingstrategy,electricitypriceforecasting1111.11.1.11.1.212234567812345131.241.1Figure1.1Decisionsupportsystemofthehydropowerplantoptimizationdispatching1.112315123ab.c.d.1.36...LOLP.2722.12.28[1]1231231234567291234562.3MarketClearingPriceMCP[2]2.3.1MCPBack-PropagationBP[3]3BP[4]BPANNBPBP101.MCPRadialBasisFunctionRBFBPBPRBFRBF2.12.1Fig2.1Radialbasisfunctionneuralnetworknmp12[,,,]TnXxxx=L12[,,,]TpXyyy=Li2()exp()2iiixcRxσ−=−,1,2,im=L(2.1)xnicixiσ211imixc−ixc−xic()iRxicixc−()iRxnxR∈x()iRxky11()mmkkiiikiiiywRxwz====∑∑,1,2,kp=L(2.2)kykikwik1234RBF1RBFiciσikw2RBFRBFa(1)(2)(3)kohonennk(4)k-meanskRBFic()12icRBFb(1)RBF2dmσ=(2)iijccσ=−ji(3)iijccσα=−α1.01.53RBFk-meansMCP2ISO2424[5]12(,,,)nxxxx=LJx∗12min(,,,)nJJxxx∗=L(2.3)2.MCPMCP[6]2132.22.2Fig2.2Multi-resolutiondecompositionalgorithm2.3wheep2.3Fig2.3Waveletreconstructionalgorithm[7]1MCP1(1,)pdt−2(2,)pdt−1(7,)pdt−1h(,1)pdt−2h(,2)pdt−2(,)ldt114(1,)ldt−2(2,)ldt−1(7,)ldt−1h(,1)ldt−[(1,),(2,),(7,),(,1),(,2),(,),(1,),(2,),(7,),(,1)]ijxpdtpdtpdtpdtpdtldtldtldtldtldt=−−−−−−−−−1,2,,iN=L1,2,,jM=L(2.4)ijxij[8]ijxNminmaxminijjijjjxxxxx∧−=−1,2,,jM=L(2.5)min12min{,,,}jjjNjxxxx=Lmax12max{,,,}jjjNjxxxx=Lmaxminmin()ijijjjjxxxxx∧=−+g(2.6)4.MCPRBF[9]1MAEδ,,11NMAEsimuktragkkppNδ==−∑(2.7),simukp,tragkpN2RMAEδ2,,11()NRMAEsimuktragkkppNδ==−∑(2.8)3MAPEδ215,,1,111NsimuktragkkMAPENtragkkppNpNδ==−=∑∑(2.9)2.3.2[10]123412,,,pxxxL(1)ppy011ppybbxbxε=++++L2~(0,)Nεσ(2.10)201,,,,pbbbσL12,,,pxxxLε4MCPMCP162.3.3NewEngland2420012002MATLABC1.MATLABwavedecwaverecwavedec[C,L]=wavedec(X,N,‘wname’)(2.11)[C,L]=wavedec(X,N,Lo_D,Hi_D)(2.12)XNwaverec[C,L]X=waverec(C,L,‘wname’)(2.13)X=waverec(C,L,Lo_R,Hi_R)(2.14)wavedecdb3db3db353db3DaubechiesDaubechiesdb3φψ2.42.4db3φψFig2.4φ(x)andψ(x)ofdb3wavelet217db32.52.5db3Fig2.5Filtersassociatedwithdb3wavelet2002102.63A1D2D3D2.6200210100Fig2.6Highfrequencyandlowfrequencysignalofpriceinthetenthperiodoftimein2002(100days)183A1D2D3D1D11wxD=21323wxxDADD=−=++2wx1wx2wxRBF[c,l]=wavedec(x,3,‘db3’);a3=appcoef(c,l,‘db3’,3);d3=detcoef(c,l,3);d2=detcoef(c,l,2);d1=detcoef(c,l,1);aa3=zeros(1,length(a3));dd3=zeros(1,length(d3));dd2=zeros(1,length(d2));dd1=zeros(1,length(d1));c1=[a3d3d2dd1];c2=[aa3dd3dd2d1];1wx=waverec(c1,l,‘db3’);2wx=waverec(c2,l,‘db3’);2002102.72.7200210100Fig2.7Reconstructeddatainthetenthpe