利用神经网络控制连续浇铸过程中的热传导

？Incontinuouscasting,thecooling-solidificationprocessmustbebasedontheadaptationofheattransfer,whichisdirectlyconnectedtocastingconditionssuchascastingspeed,castingtemperature,andcoolingparameters.Mostcontrolschemesarebasedonthestaticrelationbetweencastingspeedandwaterflowrateineachcoolingzone;thisconstitutesanopenloopthatdoesnotconsiderthedynamicsurfacetemperature,whichisanimportantparameterforthefinalslabquality.Insteelmaking,thecasting-speedchangesaffecttheglobalheattransfer.Anoptimaloperationrequiresanadjustmentoftheprocesscontrolvariables,i.e.,globalheattransfer.Alearningneuralnetwork(NN)allowstheidentificationandthecontrolofanonlinearheattransfermodelinthecontinuouscastingprocess.Aheattransfermodelwasdevelopedusingthedynamicheatbalance.Acomparisonbetweentheexperimentalopenloopresultsandthoseofthemodelsimulationisconsidered.Followingadaptation,themodelisusedforcontrollingtheslabsurfacetemperatureinclosedloop,usingNNtechnologyandPIDcontrollers.TheNNidentificationandcontrolstrategygivesastabletemperatureclosedloopcontrolcomparativelytotheconventionalPID.