摘要
Inthispaper,aconstrainedgeneticalgorithm(CGA)isproposedtosolvethesinglemachinetotalweightedtardinessproblem.TheproposedCGAincorporatesdominancerulesfortheproblemunderconsiderationintotheGAoperators.ThisincorporationshouldenabletheproposedCGAtoobtainclosetooptimalsolutionswithmuchlessdeviationandmuchlesscomputationaleffortthantheconventionalGA(UGA).SeveralexperimentswereperformedtocomparethequalityofsolutionsobtainedbythethreeversionsofboththeCGAandtheUGAwiththeresultsobtainedbyadynamicprogrammingapproach.ThecomputationalresultsshowedthattheCGAwasbetterthantheUGAinbothqualityofsolutionsobtainedandtheCPUtimeneededtoobtaintheclosetooptimalsolutions.ThethreeversionsoftheCGAreducedthepercentagedeviationby15.6%,61.95%,and25%respectivelyandobtainedclosetooptimalsolutionswith59%lowerCPUtimethanwhatthethreeversionsoftheUGAdemanded.TheCGAperformedbetterthantheUGAintermsofqualityofsolutionsandcomputationaleffortwhenthepopulationsizeandthenumberofgenerationsaresmaller.
出版日期
2005年02月12日(中国Betway体育网页登陆平台首次上网日期,不代表论文的发表时间)