Recognition Method Of Shape Flatness Based On Improved Neural Network

With the rapid economic development and the continuous progress of science and technology,the community of various industries on the demand for strip steel is also growing.Therefore,the quality requirements of the strip steel is also getting higher and higher.Plate shape as an important indicator of the quality of the strip test plate,how to solve the problem of the shape of the card has also become an important topic of domestic and foreign experts.In this paper,artificial neural network theory as the basis for the study,the choice of cold-rolled strip steel plate shape defects as a research topic.In this paper,a plate-shaped recognition method based on Elman neural network is proposed by analyzing the flatness defects of cold-rolled strip steel plate.Firstly,the advantages and disadvantages of Elman neural network are analyzed in detail.For the neural network,there are some shortcomings of training speed and precocious phenomenon.It is proposed to optimize the network by genetic algorithm.In this paper,the performance characteristics of genetic algorithm are analyzed,and it is found that genetic algorithm is easy to fall into local extremum and premature convergence when solving multimodal nonlinear problems.In this paper,the artificial immune thinking,adaptive mechanism and chaos optimization are introduced into the evolution process of the algorithm.The immune selection,adaptive crossover and chaos mutation of the individual population are enhanced,so as to enhance the global search ability of the algorithm and improve the search precision.The genetic algorithm is improved,and then the chaos immune genetic optimization algorithm is proposed.Then,this algorithm is used to optimize the initial weights and threshold parameters of Elman neural network.A model of plate-shaped flatness defect recognition based on improved Elman neural network is established and the model method is simulated by several groups of test samples.By comparing the model and the simulation results of BP,Elman and GA-Elman networks,it is proved that the chaotic immune genetic algorithm is used to identify the flatness defects of Elman network,which solves the problem that the network convergence is slow and easy to premature.Compared with the simple optimization of the genetic algorithm using the network model,it has a higher recognition accuracy,faster identification.Therefore,the method is also significant for real-time control of plate flatness.