Optimum Design And Simulation Of Flatness Defect Recognition Model

With the development of social economy,iron and steel industry plays an increasingly important role,and many industries have put forward higher requirements for the production of modern cold rolled strip.Flatness is one of the important indexes to check the quality of strip material.It is the key to determine the quality of strip product.How to control the flatness quickly,stably and accurately has become an important subject for domestic researchers.Among them,flatness recognition is the foundation and important part of the whole closed-loop control system,and its recognition results are very important for the subsequent flatness control.In recent years,with the rise of artificial intelligence,more and more scholars apply artificial intelligence to flatness defect recognition,and greatly improve the accuracy of flatness recognition,which is conducive to future flatness control.Based on this,the main work of this paper are as follows:Aiming at the problem of low accuracy of traditional flatness defect recognition method,a hybrid optimization RBF-BP combined neural network flatness defect recognition method is proposed.Firstly,self-organizing map network(SOM)is used to cluster samples,and the network topology structure after clustering is used to determine the center and width of RBF,which overcomes the problem of unstable clustering results caused by random selection of centers in traditional clustering algorithm;then genetic algorithm(GA)is used to optimize the weight of the whole network.RBF-BP combined neural network is composed of a RBF sub-network and a BP sub-network in series.It has the ability of BP neural network to predict unknown samples and the advantage of fast approximation speed of RBF neural network.The simulation results show that the method of flatness defect recognition based on hybrid optimization RBF-BP neural network can identify common flatness defects and the accuracy of flatness defect recognition is improved by 48.02%.Considering that flatness control is a complex,non-linear and strong coupling controlsystem,and the contradiction between learning accuracy and learning speed inevitably exists in neural network.It is very difficult to establish an accurate mathematical model of shape control.In this paper,RBF-ARX model is applied to flatness defect recognition.RBF-ARX model has both the accuracy of ARX model for linear system modeling and the approximation ability of RBF neural network.It is very suitable for complex system modeling and control.And the application of this method in flatness defect recognition lays a theoretical foundation for flatness control in the future;meanwhile,the problem of getting into the local optimal solution for the traditional Structured Nonlinear Parameter Optimization Algorithm(SNPOM)in identifying the parameters of the RBF-ARX model.Combining cloud genetic algorithm with SNPOM algorithm,a hybrid optimization algorithm,CGA-SNPOM,is proposed.The simulation results show that the flatness recognition model optimized by CGA-SNPOM overcomes the shortcoming that SNPOM is easy to fall into local extremum,and the accuracy of recognition is increased by75.03%.