摘要
Aself-adaptivelargeneighborhoodsearchmethodforschedulingnjobsonmnon-identicalparallelmachineswithmultipletimewindowsispresented.Theproblems'anotherfeatureliesinoversubscription,namelynotalljobscanbescheduledwithinspecifiedschedulinghorizonsduetothelimitedmachinecapacity.Theobjectiveisthustomaximizetheoverallprofitsofprocessedjobswhilerespectingmachineconstraints.Afirst-infirst-outheuristicisappliedtofindaninitialsolution,andthenalargeneighborhoodsearchprocedureisemployedtorelaxandreoptimizecumbersomesolutions.Amachinelearningmechanismisalsointroducedtoconvergeonthemostefficientneighborhoodsfortheproblem.Extensivecomputationalresultsarepresentedbasedondatafromanapplicationinvolvingthedailyobservationschedulingofafleetofearthobservingsatellites.Themethodrapidlysolvesmostprobleminstancestooptimalornearoptimalandshowsarobustperformanceinsensitiveanalysis.
出版日期
2012年02月12日(中国期刊网平台首次上网日期,不代表论文的发表时间)