资源描述:
《种求解多目标柔性作业车间调度改进粒子群算法》由会员上传分享,免费在线阅读,更多相关内容在教育资源-天天文库。
1、一种求解多目标柔性作业车间调度的改进粒子群算法白俊杰王宁生唐敦兵(南京航空航天大学CMS工程研究中心江苏南京210016)摘要:针对具有高纬搜索空间的多目标柔性作业车间调度问题,提出了一种基于偏好的多目标粒子群优化算法(PMOPSO)。算法引入了决策者的偏好信息,用以指导算法的搜索过程,使算法在决策者感兴趣的区域进行搜索,不但缩小了算法的搜索空间,提高了算法的效率,而且一次运算只求得偏好区域内若干个折中解,避免了决策者要在众多非劣解中做出困难的选择。在算法中,采用了新的偏好信息给定方法,即采用目标间重要关系、目标数值或目标
2、权重大致取值范围来表示偏好信息。采用该方法,不但便于决策者给定偏好信息,而且还可以根据决策者的需求,对搜索区域的范围进行适当的调整。针对偏好信息的特点,提出了一种模拟人类社会组织“投票选举”的偏好信息处理方法,该方法直观简便并易于实现。最后,通过实例仿真,对算法性能进行比较分析和评价,结果表明了算法的有效性和可行性。关键词:柔性作业车间调度;粒子群优化算法;多目标优化;偏好信息中图分类号:TH16;TP278文献标识码:AImprovedPSOAlgorithmfortheMulti-ObjectiveOptimizati
3、onFlexibleJobShopSchedulingProblemsBaiJun-jieWangNing-shengTangDun-bing(CMSResearchcentre,NanjingUniv.OfAeronauticsandAstronautics,Nanjing210016,Jiangsu,China)Abstract:Tosolvethemulti-objectiveflexiblejobshopschedulingproblemwithlargedimensionalsearchingspace,apre
4、ferencebasedmulti-objectiveparticleswarmoptimizationalgorithm(PMOPSO)wasproposed.Thepreferenceinformationofdecisionsmakerisincorporatedintothealgorithmtoleadthesearchingdirection.Sothat,notonlythesearchingspaceiscompressedandtheefficiencyofthealgorithmisimproved,b
5、utalsojustafewtrade-offsolutionslocatedinpreferredareaareobtainedinasinglerun,andthehardworkofchoosingasatisfyingsolutionfromnumerousnon-inferiorsolutionsiseliminated.Inthealgorithm,anewexpressionmethodofpreferenceinformationbasedonimportancerelationshipamongobjec
6、tivesandthevaluerangeofobjectivesorobjectiveweightswasproposed.Withthismethod,notonlythepreferenceofdecisionsmakercanbeeasilyspecified,butalsotherangeofsearchingareacanbeadjustedproperlyaccordingtotherequirementsofdecisionsmaker.Inviewofthecharacteristicsofprefere
7、nceinformation,anewpreferenceinformationhandlingmethod,whichsimulatesthe“vote”ofhumansociety,wasproposed.Themethodisintuitive,simpleandeasytouse.Finally,theperformanceofthealgorithmwasevaluatedthroughsimulations,andtheresultsdemonstratethefeasibilityandefficiencyo
8、fproposedalgorithm.Keywords:flexiblejobshopscheduling;particleswarmoptimizationalgorithm;multi-objectiveoptimization;preferenceinformation10/100.引言近年来,多