An improvedmulti-objectiveevolutionaryalgorithmforthevehiclerouting problem withtimewindows

The vehicleroutingproblemwithtimewindowsisacomplexcombinatorialproblemwithmanyreal- world applicationsintransportationanddistributionlogistics.Itsmainobjectiveistofindthelowest distance setofroutestodelivergoods,usingafleetofidenticalvehicleswithrestrictedcapacity,to customerswithservicetimewindows.However,thereareotherobjectives,andhavingarangeof solutionsrepresentingthetrade-offsbetweenobjectivesiscrucialformanyapplications.Although previousresearchhasusedevolutionarymethodsforsolvingthisproblem,ithasrarelyconcentratedon the optimizationofmorethanoneobjective,andhardlyeverexplicitlyconsideredthediversityof solutions.Thispaperproposesandanalyzesanovelmulti-objectiveevolutionaryalgorithm,which incorporatesmethodsformeasuringthesimilarityofsolutions,tosolvethemulti-objectiveproblem. The algorithmisappliedtoastandardbenchmarkproblemset,showingthatwhenthesimilarity measureisusedappropriately,thediversityandqualityofsolutionsishigherthanwhenitisnotused, and thealgorithmachieveshighlycompetitiveresultscomparedwithpreviouslypublishedstudiesand those fromapopularevolutionarymulti-objectiveoptimizer.