基于進化算法的數(shù)值優(yōu)化問題研究及其應用.pdf

1、中國科學院自動化研究所博士學位論文基于進化算法的數(shù)值優(yōu)化問題研究及其應用姓名：商允偉申請學位級別：博士專業(yè)：控制理論與控制工程指導教師：裘聿皇20040501ResearchandApplicationsofNumericalOptimizationBasedonEvolutionaryAlgorithmsAbstractEvolutionaryAlgorithms(EAs)suchasGeneticAlgorithms(GAs)，Ev

2、olutionStrategies(zs’s)，EvolutionaryProgramming(EP)，areaclassofstochasticsearchandoptimizationmethodsgleanedfromthemodeloforganicevolutionThedifferencesbetweentheEAsandthetraditionalsearchalgorithmsaremainlyintwoaspects：

3、populationbasedsearchandinformationexchangebetweentheindividualsTheapplicationsoftheEAstothenumericaloptimizationproblemsarestudiedThemaincontributionsofthedissertationaresummarizedasfollows1Thefrequentlyusedcrossoverand

4、mutationoperatorsofthereal—codedGAsarediscussedThesearchabilitiesoftheseoperatorsarecomparedTheanalysisandtheexperimentsshowthatthesearchabilityofanoperatormayvarygreatlyatdifferentevolutionarystagesAndthesearchabilityof

5、anoperatormayvarygreatlywhenitisusedtosolvedifferentproblemsIflittleisknownabouttheproblem，hybridizationofmanyoperatorsisafeasiblewaytoextendtheapplicationrangeandimprovetheperformanceoftheEAs2TheGAscanbeusedtosolvemulti

6、modalfunctionoptimizationproblemsbecauseofthepopulationbasedsearchThevariationoftheselectionprobabilityisstudiedwhenfitnesssharingschemeisaddedtothegeneticalgorithmtooptimizethemultimodalfunctionsTheanalysisandsimulation

7、resultsshowthatthegeneticalgorithmwithfitnesssharingCallmaintainpopulationdiversitytoacertainextentandtherangeofthefitnessfunctioncangreatlyinfluencetheperformanceofthealgorithmAmultilevelGeneticAlgorithm&TabuSearchalgor

8、ithm(MGA—TSlisproposedtooptimizemultimodalfunctionThemainGAperformsglobalsearchwhilethemicrogeneticalgorithm(MGA)exploitstheneighborhoodofthecurrentsolutionprovidedbythemainGATheTSisintroducedtopreventthecyclingsearchofs