|
|
自动化学报 2005
一种基于强化学习的作业车间动态调度方法, PP. 765-771 Keywords: Reinforcementlearning,compositerules,meantardiness,dynamicjob-shopscheduling Abstract: ？Productionschedulingiscriticaltomanufacturingsystem.Dispatchingrulesareusuallyapplieddynamicallytoschedulethejobinadynamicjob-shop.Existingschedulingapproachessel-domaddressmachineselectionintheschedulingprocess.Compositerules,consideringbothmachineselectionandjobselection,areproposedinthispaper.Thedynamicsystemistrainedtoenhanceitslearningandadaptivecapabilitybyareinforcementlearning(RL)algorithm.Wedefinetheconceptionofpressuretodescribethesystemfeature.Designingarewardfunctionshouldbeguidedbytheschedulinggoaltoaccuratelyrecordthelearningprogress.CompetitiveresultswiththeRL-basedapproachshowthatitcanbeusedasreal-timeschedulingtechnology.
|
