摘要
Anewarchitectureofwaveletneuralnetworkwithmulti-input-layerisproposedandimplementedformodelingaclassoflarge-scaleindustrialprocesses.Becausetheprocessesareverycomplicatedandthenumberoftechnologicalparameters,whichdeterminethefinalproductquality,isquitelarge,andtheseparametersdonotmakeactionsatthesametimebutworkindifferentprocedures,theconventionalfeed-forwardneuralnetworkscannotmodelthissetofproblemsefficiently.Thenetworkpresentedinthispaperhasseveralinput-layersaccordingtothesequenceofworkprocedureinlarge-scaleindustrialproductionprocesses.Theperformanceofsuchnetworksisanalyzedandthenetworkisappliedtomodelthesteelplatequalityofcontinuouscastingfurnaceandhotrollingmill.Simulationresultsindicatethatthedevelopedmethodologyiscompetentandhaswellprospectstothissetofproblems.
出版日期
2004年02月12日(中国Betway体育网页登陆平台首次上网日期,不代表论文的发表时间)