并行進(jìn)化算法的研究與應(yīng)用.pdf

1、江南大學(xué)碩士學(xué)位論文并行進(jìn)化算法的研究與應(yīng)用姓名：馬艷申請(qǐng)學(xué)位級(jí)別：碩士專業(yè)：計(jì)算機(jī)應(yīng)用技術(shù)指導(dǎo)教師：須文波20070601江南人學(xué)碩f。學(xué)位論義AbstractInthispaperanovelclassofparallelevolutionarycomputationtechniqueSwarmIntelligenceAlgorithmandapplicationisdiscussed，amongwhichtheQuantumbeh

2、avedParticleSwarmOptimization(QPSO)algorithmisarecentlyproposedapproachandisavariantoforiginalparticleSwarmOptimization(PSO)QPSOisglobalconvergentandwillbeapromisingsolverforcomplexoptimizationproblem，whichisshownbysomep

3、reviousworkThus，theresearchofthispaperwillbeofsomewhatsignificanceinevolutionarycomputationareaandparallelareaOnthebaseofgeneticalgorithm，particle$warnlalgorithm，quantum—behavedparticleSWalTnalgorithmandsomeotherparallel

4、algorithm，weputforwardanewparallelmethodonparticleSWaHrlalgorithmandquantumbehavedparticleswarmalgorithmbytheinfluenceoftheimplementofgeneticalgorithmThemainideaisintroducingtheconceptsofislandmodelandchan百ngarithmeticop

5、eratorsIntheclusterenvironment，theoptimalalgorithmincreasingglobalsearchingabilitybymaintainingdiversityusingchangingarithmeticoperatorstocommunicatewitheachotherDuetothelongtimeforcommunication，bottleneckwillappearonpar

6、alleltestsWeamendcommunicationbyreducingthecommunicationcycleaccordingtodescendingnumeralarrayWesetexamplesbysomebasictestingcasetodescribethedesignideaandtheimplementprocessandweprovidemanytestingresultsandcomparisonwit

7、hotherserialandparallelresultsTheresultsindicatethatbothsearchingabilityandrunningtimearebetteranditsausefulmeansforsolvingcomplexoptimizationproblemsInthispaperwepresentadecisionmakingprocessthatincorporatesParallelPart

8、icleSwarmOptimization(PPSO)AlgorithmintomultistageportfoliooptimizationsystemTheobjectivefunctionistomaximizeone’seconomicutilityorendofperiodwealthExperimentsarcconductedtocompareperformanceoftheportfoliosoptimizedbydif