Lionel第1應(yīng)是加快主義小康社會的基本保證，是實現(xiàn)國民經(jīng)濟發(fā)展現(xiàn)代化的戰(zhàn)國的裝機容量已達10.7億千瓦，位居世界第二；年發(fā)電總4.7萬億千瓦時，超過，位居世界第一。電力系統(tǒng)規(guī)劃的會給國家建設(shè)帶來不可彌補的損失，反之，一個合理的電2.1投資在滿足以下約束條件下，在NT規(guī)劃年中合理安排Ngn個待建電廠（包括NghNgf個火電廠）的機組投產(chǎn)計劃，使得等年值最小。等年值表Nt Nt NtminB(I)(atiCRFcti)Xti(btjCRFdtj)YtjCRFt1 t1j t1k 中的Cplex優(yōu)化工具包進行求解Ntt1k進行將火電機組和水電機組的投產(chǎn)情況進行排列即為決策變量，其行數(shù)為100因為采用cplexbilp0、1優(yōu)化求解算法，所以約束條件12能夠自動滿足。約束3需要考慮總裝機臺數(shù)的約束也很容易實現(xiàn)。約束4考慮發(fā)電廠最個等式約束。約束5和約束6考慮電力電量平衡的約束中需要用到水電廠的單機預(yù)想出力，數(shù)據(jù)中給出的是每個月的預(yù)想出力，所以需要考慮的情況，將水電廠單機每個月的最小的出力作為每年的平均出力來考慮。這樣的出將第tRDt和電量備用系數(shù)REt作為輸入變量。進行3X1,7 1000000000Y1,7

X2,7 110000000 2,7 X3,7

1101

0010

0000

03,7Y0 Y 5,7

011100000*Y5,7X6,7

0001110000

6,7

0000111000

X7,7 7,78,7 0000011100Y8,7 0000001110 9,7 9,7 0000000111 10,7 10,7以代建電廠的容量備用系數(shù)RDt和電量備用系數(shù)REt0.2為算例，測試X0000000000010000000001100000000101000000101100100010010011000110000100101100010010001001000000100000522222421200000000005222224222010000000052222242220210000000522222422203110000005222224222222224222232300210052222242222433002200522222422234440023005222224222444410240052222242224444202400概算LOLP等量。這就是檢修規(guī)劃在整體規(guī)劃中發(fā)揮的作用的機組過某值，本作業(yè)要求電廠最多不能對超過2臺機組進行檢修。限制兩機組檢修間隔時間，本作業(yè)設(shè)定機組檢修開始時間要至少間隔4個數(shù)學(xué)模型的本項機組檢修規(guī)劃作業(yè)建立了兩個規(guī)劃數(shù)學(xué)模型，分別是：（1）變量定義量，若值為“1it月開始檢修，否則為“02412月，今年計劃1begin_2year1,121begin_2year1,131。month_loadt：表示t月的最大負(fù)荷；month_reservet：表示t月的備用率；installed_numberm：表示電廠m的可用數(shù)學(xué)模型的建立目標(biāo)函數(shù)month_

month_capacitytmonth_month_t約束條件檢修約束以等式約束形式出現(xiàn)，根據(jù)決策變量beginit和已知量durationi，可以線性求得機組狀態(tài)conditionit，滿足檢修連續(xù)性。t=1t>1

installed_

conditionitbegin_2x

(x8,機組檢修規(guī)劃的Cplex優(yōu)化規(guī)模過大，優(yōu)化時間過長，不適合在主程序中進行反復(fù)調(diào)用，而且檢修規(guī)劃并不要求對于優(yōu)化的目標(biāo)函數(shù)取極值，因啟發(fā)式算法步驟350876012個月的最大負(fù)圖3.112 911141052671718204圖3.2線，根據(jù)每個等效負(fù)荷曲線分別求取機組添加在12個月分別的各月的等效備用率；程序?qū)崿F(xiàn)及結(jié)果的極小值。如下式，程序采用yalmipcplex求解。 1114105267171820開開 電廠在n月是否間隔4是否是否是安排機組在n（n從1開始否圖3.3算例installed_number=[5222224102000000000=05000670003=0500067000368050000000009100000000000000000000000000000000000000000000000000000maintain_plan 455545maintain_plan 4555455535222222211222121222222222112222222222212212222222222221122222443443434344111101111111000000000000122222221222000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000000圖3.4概隨機生產(chǎn)模擬的計算方法包括標(biāo)準(zhǔn)卷激發(fā)、級數(shù)法、半不變量法、半不變量法Mrpi(xiiiK2MK3K K5M510M KM15MM10M 量的各階半不變量之和；2）隨量的a倍的r（r為非負(fù)整數(shù)）階半不變量等于該隨量的r階半不變量的ar倍過Gram-r級數(shù)展開式，由電力系統(tǒng)總的ELDC的各階半不變量求得實際的隨量分布的函數(shù)值。 r g f(x)N(x)dx+N(x)3H2(x) 4H3(x) 5H4(x) 6H5(x)…… 3!

H0H(x) H2(x)x2HH

(x)x3 gK/vK/Kr/2(r 程序否是是否否 是是根據(jù)ELDC半不變量應(yīng)用 r級數(shù)計算等效持續(xù)負(fù)荷曲線得到最終的ELDC計算新的ELDC計算新的ELDC計算新的ELDC按照水電出力排序?qū)ふ宜姷某隽ξ恢茫ǚ謩e計算EA，EL，使得根據(jù)ELDC半不變量應(yīng)用 r級數(shù)計算等效持續(xù)負(fù)荷曲計算得到圖4.1模擬結(jié)果0、0Rdt=0.06，Ret=0.06的情境下迭代求解。得到的ELDC圖4.2半不變量法隨機生產(chǎn)模擬LOLD求得的LOLP0.0367，EENS2850.8?；痣姀S發(fā)電量分別為：21204，由錯誤!未找到源?？梢钥闯觯氩蛔兞糠ㄓ捎诓捎肎ram-r級數(shù)近似求解，且級數(shù)僅為8，具有一定的近似性。得到的隨機生產(chǎn)模擬仿真結(jié)果ProbabilisticproductionsimulationisusedtoTheexpectedLOLPandEENSreliabilityindicesofthesystemduringthesimulationperiodTheexpecteddeploymentofeachindividualcoalgeneratorforoptimumeconomicTheexpectedoptimaldeploymentforhydroplantstoachieve umpeakshavingandeconomicsavingTheconvolutionmethodofprobabilisticproductionsimulationiscomputationallydemandingbutextremelyaccurate,andwasusedtoachieveourmostaccuratepredictionsforreliabilitiesindices,andindividualgeneratorenergyoutputs.PrinciplesofConvolutionProbabilisticproductionsimulationbeginsbyderivinganinitialloaddistributioncurve(LDC),??(0),fromgivenloaddata.IndividualgeneratorenergyoutputsandreliabilityindicesLOLPandEENScanbecomputedbyderivingafurthernloaddistributioncurvesbytheconvolutionmethod,wherenisthetotalnumberofgeneratorsinthesystem,andn=nhydro+ncoal.EachsuccessiveLDC??(1,2,…,??,…??)takesintoaccounttheprobabilityofoneadditionalgenerator’sforcedoutagerate,i.e.??(??)isthecurvethathastakenintoaccountthelikelihoodsofforcedoutagesfromgenerators1,2,…,??.Plantshaveforcedoutragerate(FOR),orunplannedoutagerate,????.Theprobabilitythataplantisworkingcorrectlyis????=1?Thei-thLDC,??(??),canbecalculatedbyconvolutionfromtheprevious??(???1).Usingtheconvolutionformulabelow,allnLDCscanbesuccessivelycomputedfrom??(0),ortherawloaddata:??(??)(??)=??????(???1)(??)+??????(???1)(???where????isthepowercapacityofthei-thExpectedGeneratorEnergyOutputTheplannedpowergenerationofeachcoalplant????1,…,??????????????canbecalculatedfrom??(??)(??)(someadditionalcomputationtodeterminethegeneratorsdeploymentisalsorequiredandwillbeshown??????=??????∫FigureWhere??=??????????isthenumberofhoursintheperiodofsimulation(inthiscaseea onth),and????=∑?? ????FigureMeanwhile,hydroplantsshouldalwaysberunattheirfullcapacity,??????????????+1,…,??????aresimplythegivenmonthlyenergyoutputsofthegenerators(????1,…,?????? )multipliedby????.eenceafter??(??????????)hasbeencalculated,theremainingcurves,i.e.??(??????????+1),…,??(??)arerelativelyquicklycomputedbytheconvolutionformulagivenabove.ReliabilityIndicesLOLPandAfterthefullsetofLDCshavebeenderived,thereliabilityindicesarederivedas????????=??(??)(???? ????????= ??(??)(??)????=∫Where??=∑????,and ??1 ????????is umloaduponthesystemduringFigureperiodofFigurebelowwasgeneratedusingdataintheassignmenttodemonstratetheprinciplesofLOLPandEENS.Threecurvesareshown:theinitialLDC,??(0),thepenultimatecoalgenerator’sLDC,??(???????????1)=??(11),andthefinalhydrogenerator’sLDC,??(??)=??(22).LOLPandEENSarealsoshown.Asmentionedinthefigure,arestrictednumberofgeneratorswereusedinthissimulationinordertocreateasignificantLOLPandEENS.Also,Figure4laterinthissectionshowshowhydroplantswerepositionontocurves??(0)and??(12)andtheLDCsthenamendedtomostaccuraypredictthepoweroutputsofgenerators1and12.Figure1:Illustrationofhigherorderloaddistributioncurves,andLOLPandEENSAlgorithmDesignThealgorithmwasdesignedaccordingtothefollowingInordertoachieveumpeakshavingrunhydrogeneratorsattheirfullpowercapacityasmuchaspossible.Alwaysrunhydrogeneratorsattheirfullmonthlyenergyoutput,becausetherearenofuelcosts.DeploycoalgeneratorsintheorderoftheireconomicForeaonththealgorithmgoesthroughthefollowingstages:Computationofthencoalcoalplants’loaddistributionEachLDCmustbeamendedtotakeintoaccountthestrategicdeploymentofhydroplantssothat??????canbecalculatedaccura Computationofthenhydrohydroplants’loaddistributioncurvesSavingofreliabilityindiceswhicharesentbacktothemainprogram.2overpagedepictsthesestagesofthealgorithm’sprocessinAlgorithmOptimalHydroPlantDeploymentforPeakPlacingofhydroplantsupontheloadcurvemustbecompletedforeachiterationof??(??).eydroplantsaregroupedintoaninitialnumberofequivalentgeneratingunits(EGUs),equaltothenumberofuniqueoperatingtimes(??????).Plantsthatsharethesame??????thismonthwillbegroupedintoasingleinitialEGU.Apositionupontheloadcurveisthenfoundwiththeobjective:minΔEi=min(???????where??????istheenergyundertheloadcurvesegmentofwidth??????,and??????isenergyoutputs.Inthesimulations,verysmallΔEivaluesoftheorderofafewkWhwereachievablebycalculatingeachloadcurvewithanaccuracyof0.1MW.WhensubsequentEGUsareplacedaloadcurveitispossiblethattheywilloverlap,whichisnotallowed.InthiscasetheyshouldbeiterativelymergedasillustratedinFigXX.AllEGUsareplacediteratively,mergingwheneverFigure3：Iterativeprocessforplacinghydro-plantequivalentgeneratingunitsontotheloadcurveumpeakBecauseofthesteepnessoftheloadcurvethatisproducedfromtheproblemdata,itturnedoutthatineverymonthofeveryyearallEGUswouldoverlap,andalgorithmwoulditerativelymergethemtogethertoproduceasingleEGUwith??????

??????,

??????

∑

??????andoperatingtime??????=??????/??????CoalGeneratorEnergyOutputFigAfterthecorrectpositionofthehydrogeneratorson??(???1)hasbeenfound,itthenpossibletoaccurayestimatetheoutputofcoal ??byFig??(???1) belowshowshowtheEGUpositioningwascompletedwhenalgorithmwasappliedtouponthe??(0)and??(11)curveswhichwereshownpreviouslyinFig1,andhowthecurveswereamendedsothat????1and????12wereInFigure4theredrectanglerepresentstheEGUforall10hydroplants,showingit’sequivalentoperatingtimeandcapacity.Ascanbeseen,thecorrectpositionoftheEGUshiftstotherightwithincreasingLDCiterations.(IftheEGUwasnotshifted,thentherewouldbeanincreasinglysubstatialinaccuracyinthepredictedcoalplantsoutputs,??????).12345678910Figure5:ActualsimulationresultsforDecemberYear10,showingLOLP,EENSandall50loaddistributioncurvesfor33coalgeneratorsand16hydrogeneratorsDiscussionandAsFigure5shows,actualresultswerefarbetterthanthedemonstrationexamplesshowninFigure1andFigure4.Thisisbecauseinouractualresults????rangedaround30%largerto50%largerthan????????,owingtoeffectivemaintenanceschedulingandinvestmentplanning.Itshouldbenotedthanas????movesfurtherawayfrom????????,therateofLOLP’sdecreaseisfasterthanexponential.TherangeofvariationmentionedaboveproducedLOLPresultsintherangeof2.5×10?4to1×10?9.Computationaltimevariedincloserelationtothefinenessusedfortheloaddistributioncurves.A1MWresolutionwasfoundtobetooslowforcomputingalltenyears,thewholesimulationrequiringaroundtenminutestocomplete.WhenLDC_Reswaschangedto10MW,thewholesimula