结构重分析算法加速遗传算法求解结构频率优化问题

《结构重分析算法加速遗传算法求解结构频率优化问题》由会员上传分享，免费在线阅读，更多相关内容在教育资源-天天文库。

1、Author'spersonalcopyStructMultidiscOptim(2011)43:799–810DOI10.1007/s00158-010-0610-yRESEARCHPAPERFaststructuraloptimizationwithfrequencyconstraintsbygeneticalgorithmusingadaptiveeigenvaluereanalysismethodsWenjieZuo·TaoXu·HaoZhang·TianshuangXuReceived:21April2009/Revised:1Nove

2、mber2010/Accepted:6December2010/Publishedonline:13January2011cSpringer-Verlag2011AbstractStructuraloptimizationwithfrequencycon-KeywordsFaststructuraloptimization·straintsishighlynonlineardynamicoptimizationproblems.Adaptiveeigenvaluereanalysis·Geneticalgorithm·Geneticalgori

3、thm(GA)hasgreateradvantageinglobalFrequencyconstraintsoptimizationfornonlinearproblemthanoptimalitycrite-riaandmathematicalprogrammingmethods,butitneedsmorecomputationaltimeandnumerouseigenvaluereanal-ysis.Tospeedupthedesignprocess,anadaptiveeigenvalue1Introductionreanalysism

4、ethodforGA-basedstructuraloptimizationispresented.ThisreanalysistechniqueisderivedprimarilyTheproblemofstructuraloptimizationwithfrequencycon-ontheKirsch’scombinedapproximationsmethod,whichstraintshasattractedsubstantialattentionrecently(Dedeisalsohighlyaccurateforcaseofrepea

5、tedeigenvaluesandHulbert2009;Grandhi1993;Meskeetal.2006;Niuproblem.Therequirednumberofbasisvectorsateverygen-etal.2009).Frequencyconstraintsarehighlynonlinear,erationisadaptivelydeterminedandtherulesforselectingnon-convexandimplicitwithrespecttothedesignvari-initialnumberofba

6、sisvectorsaregiven.Numericalexam-ables(Gholizadehetal.2008;Torkzadehetal.2008).Soplesoftrussdesignarepresentedtovalidatethereanalysis-far,thecommonandefficientstrategiesfortheproblemsbasedfrequencyoptimization.Theresultsdemonstratethataretheoptimalitycriteria(OC)andmathematic

7、alprogram-theadaptiveeigenvaluereanalysisaffectsveryslightlytheming(MP)methods.Nevertheless,forOC(Rozvany1989),accuracyoftheoptimalsolutionsandsignificantlyreducesspecializedoptimalityformulaneedbededucedforeverythecomputationaltimeinvolvedinthedesignprocessofkindsofimplicitc

8、onstraints,soitisnotreadilytopro-large-scalestructures.gramuniformpr