制造系統(tǒng)成品需求短期智能預(yù)測系統(tǒng).pdf

1、東南大學(xué)碩士學(xué)位論文制造系統(tǒng)成品需求短期智能預(yù)測系統(tǒng)姓名：王方順申請學(xué)位級別：碩士專業(yè)：控制理論與控制工程指導(dǎo)教師：嚴洪森20120329AbstractSHORT—TERM烈TELLIGENTFORECASTⅢGSYSTEMFORFⅢISHEDPRODUCTDEMAND烈MANU下ACTURⅢGSYSTEMMasterCandidate：WANGFangshunSupervisedbyProfYANHong—senSoUTHEASTU

2、NlVERSITYAbstractManufacturingindustryiSthemostimportantindustryofChina’Snationaleconomyanddirectlyreflectsthelevelofacountry’SproductionThecompany’Sproductswhichareproducedbythemanufacturingsystem，arefinishedproductsofm

3、anufacturingsystemProductdemandforecastingplaysanimportantroleinmakingproductionplan，determiningoperatingpoliciesandmakingmanagementdecisionineverydepartmentoftheenterpriseInmisthesisthetheoriesandmethodsaboutthedemandfo

4、recastingarestudiedthoroughlyanddeeplyThenavariableweightscombinationmethodandtheconceptofintelligentforecastingareproposedhere，andanimprovedcombinedforecastingmodelwithvariableweightsisproposedThen，anintelligentselectin

5、gmethodofthebestforecastingmodelbasedontheadaptivetheoryisproposedFinallytheshorttermintelligentforecastingsystemforfinishedproductdemandinmanufacturingsystemisdevelopedbyusingthecomputerinformationtechniqueThewholeproce

6、dureoftheresearchanddevelopmentoftheforecastingsystemisalsodescribedFirstlythebackgroundandsignificanceofthedevelopmentoftheshorttermintelligentforecastingsystemforfinishedproductdemandinmanufacturingsystemiSanalyzedThis

7、thesisalSOresearchesintothedevelopmentstatusoftheforecastingtechniqueandcombinationforecastingtechniquebothathomeandabroadThen，thetimeseriesforecastingtechnique，artificialneuralnetforecastingtechniqueandsupportingvectorm

8、achineforecastingtechniqueareintroducedindetail“17‘hetheoryandimplementationstepsoftheneuralnetforecastingmodeloptimizedbyparticleswarmalgorithmandanidentificationmethodofthenetworkparametersinsupportingvectormachinearea

9、lsostudiedandappliedsuccessfullytoapracticalsystemSecondlyinordertoreducethepredictionerrorofasinglemodelandimprovepredictionaccuracythecombinationforecastingmethodsarestudiedinthisthesisAvariableweightscombinationmethod

10、iSproposedonthebasisofimprovement，anditsimplementationiSdescribedindetailBesidesthisthesisanalyzesandstudiesthefilteringofredundantcombinationforecastingmodelsThefeasibilityandeffectivenessofthosemethodshavebeenprovedthr

11、oughinstancesThirdlyinordertoselectthebestpredictivemodelinlinewiththecharacteristiesofthesamplefromdozensofforecastingmethods，theconceptofintelligentforecastingisproposedhereThenanintelligentselectingmethodofthebestfore