Page 34 - 2023年第54卷第2期
P. 34

[18] FENGZK,NIUW J,LIUS,etal.Multiplehydropowerreservoirsoperationoptimizationbyadaptivemutation
                       sinecosinealgorithmbasedonneighborhoodsearchandsimplexsearchstrategies [J].JournalofHydrology,2020,
                      590:125223.
                [19] 朱迪,梅亚东,许新发,刘章君.复杂防洪系统优化调度的三层并行逐步优化算法[J].水利学报,2020,
                      51(10):1199 - 1211
                [20] ZHANGW,LIUP,CHENX,etal.Optimaloperationofmulti - reservoirsystemsconsideringtime - lagsofflood
                       routing [J].WaterResourcesManagement,2016,30(2):523 - 540.
                [21] 冯仲恺,牛文静,程春田,等.梯 级 水 库 群 优 化 调 度 精 英 集 聚 蛛 群 优 化 方 法 [J].水 利 学 报,2016,47
                       (6):826 - 833.
                [22] 邹强,王学敏,李安强,等.基于并行混沌量子粒子群算法的梯级水库群防洪优化调度研究[J].水利学
                       报,2016,47(8):967 - 976.
                [23] SEETHARAM K V.Threelevelrulecurveforoptimum operation ofamultipurposereservoirusinggenetic
                       algorithms [J].WaterResourcesManagement,2021,35(1):353 - 368.
                [24] RAGSDELLKM,PHILLIPSDT.Optimaldesignofaclassofweldedstructuresusinggeometricprogramming[J].
                       JournalofEngineeringforIndustry,1976,98(3):1021 - 1025.
                [25] WOLPERTDH,MACREADYW G.Nofreelunchtheoremsforoptimization[J].IEEETransactionsonEvolu
                       tionaryComputation.1997,1(1):67 - 82.
                [26] LIANGJ,RUNARSSONT,MEZURA - MONTESE,etal.ProblemdefinitionsandevaluationcriteriafortheCEC
                      2006specialsessiononconstrainedreal - parameteroptimization[R].NangyangTechnologicalUniversity,Singa
                       pore ,2006.
                [27] CHENH T,WANGW C,CHAUKW,etal.Floodcontroloperationofreservoirgroupusingyin - yangfirefly
                       algorithm[J].WaterResourcesManagement,2021,35(15):5325 - 5345.




                            Mε - OIDEalgorithmforsolvingconstrainedoptimizationproblemsandits
                                     applicationinfloodcontroloperationofreservoirgroup
                                                                 2
                                                                                3
                                                        1
                                          1
                                                                                              1
                           WANGWenchuan,TIANWeican,XULei,LIUChangjun,XUDongmei
                     (1.CollegeofWaterResources,NorthChinaUniversityofWaterResourcesandElectricPower,Zhengzhou 450046,China;
                               2.CollegeofHydrologyandWaterResources,HohaiUniversity,Nanjing 210024,China;
                      3.ResearchCenteronFloodandDroughtDisasterReduction,ChinaInstituteofWaterResourcesandHydropowerResearch,
                                                    Beijing 100038,China)
                  Abstract:Constrainthandlingmethodsandarepresentativeinitialpopulationhaveasignificantimpactontheper
                  formanceofconstrainedoptimizationalgorithms.Aimingattheproblemsofthe εconstrainthandlingmethodinsol
                  vingconstrainedoptimizationproblemssuchasunstablecapabilityanddifficultiesinselectingempiricalparameter
                  values ,thispaperstartsfromthecurrenttwodifferent εconstraintprocessingmethods,throughtheanalysisoftheir
                  advantagesanddisadvantages ,theZ - εconstraintprocessingmethodadditionallyperforms δ relaxedoperationsone
                  qualityconstraintsaresupplementedtotheoverallframeworkoftheTS - εprocessingmethod,andaddsauser - de
                  finedparametertodealwithvariousconstraintconditions ,soastoproposeamodifiedεconstrainthandlingmethod.
                  Basedontheprimarydifferentialevolutionalgorithm ,alightweightconstrainedoptimizationalgorithm namedM ε -
                  OIDEwasproposedbycouplingitwiththeaforementionedmodified εconstrainthandlingmethodandclassicaloppo
                  sition - basedlearningpopulationinitializationstrategy.ThetestresultsontheCEC2006benchmarkfunctionsetverify
                  theeffectivenessofthecouplingstrategy,indicatingthattheproposedM ε - OIDEalgorithm hashighaccuracyand
                  strongrobustness.Inaddition ,theoptimizationofreservoirgroupfloodcontroloperationfurtherprovethattheM ε -
                  OIDEalgorithmisfeasibleandefficientindealingwithpracticalconstrainedoptimizationproblems.
                  Keywords:constrainedoptimization;εconstrainthandlingmethod;differentialevolutionalgorithm;opposition -
                  basedlearning;reservoirgroup;optimaloperationoffloodcontrol

                                                                                    (责任编辑:耿庆斋)

                —  1 5  —
                     8
   29   30   31   32   33   34   35   36   37   38   39