The Research On The Shape And Thickness Integrated Control Method Based On Neural Network Decoupling

Steel is an important material base of developing national economy and strengtheningcomprehensive national power.Strip steel is substantial law materials in national economy.In recent years,with the constant improvement of industrial automation level,therequirement of the quality of strip products becomes higher and higher.Strip formed by a series of deformation process of metal under roll effects.The wholerolling process could be affected by the own characteristics of metal and rolling conditions.Flatness quality and gauge precious are two important primary indexes of product qualityin strip mills.Thus the research on the automatic flatness control-automatic gaugecontrol(AFC-AGC) system is already at the front line of current study on strip rollingtechnology.But,the AFC-AGC is a complex system with non-linear,strong coupling,largetime delay,multivariable and real-time,so the conventional control methods can not getsatisfying results.As a result modem control technology associated with the intelligencetechnology has been the development tendency of AFC-AGC.This paper will regard flatness control and gauge control as a whole to consider.Thereestablishes an AFC-AGC integrated control model and applies the invariance principle andPID control algorithm to the decoupling control of AFC-AGC system.The simulationresults show that the decoupling control effect is good and effectively reduce the couplingimpact between the flatness control and gauge control;But,when the model parameters arechanged in the system operation process,decoupling control performance will beworse.And it is hard to establish precise mathematical model of AFC-AGC system,thatalso can affect decoupling effect.As a result, this paper puts forward the RBF neuralnetwork decoupling control method based on the neural network and the decouplingcontrol theory and designs the PID controller that based on RBF neural network to set theparameters to solve those problems.And they are applied to the AFC-AGC integratedcontrol system.This mathod effectively solve the problem that the uncertain factors of thesystem parameters impact control performance in the decoupling control process. Finally,the paper conducts the Matlab simulation on the AFC-AGC integrated control system that bases on the neural network decoupling,the simulation results indicate that this controlstrategy has improved the decoupling control precision and it is better than that oftraditional decoupling control method.