Studying On Prediction Methods Of Deformation Resistance Of QP1180

The metal plastic deformation resistance is a basic quantity in the process of steel processing,in order to make a reasonable rolling process,we need to accurately determine the deformation resistance value of the metal under any different conditions.Most researchers now use the combination of computer and mathematical models to deal with experimental data with multiple nonlinear regression.However,such a method is too complicated,which is not conducive to the field control,and there is a large error.With the emergence of artificial neural network and swarm intelligence optimization algorithm,they show the superiority of the powerful adaptive learning ability and the highly complex nonlinear problem in the processing engineering.Many scientists are trying to use these new artificial intelligence algorithms to predict deformation resistance.In this paper,the particle swarm optimization is used to optimize t the BP neural network model,and a PSO-BP model can be successfully established to predict deformation resistance of metal.According to the basic theory of BP neural network,we have successfully verified the fitting capacity of its nonlinear function.This paper puts forward a kind of inertia weight increases with the number of iterations and the change in the form of nonlinear adaptive adjustment of the improved particle swarm optimization algorithm and take advantage of the improved PSO algorithm to optimize the BP neural network weights and threshold.In this way,the PSO-BP prediction model was successfully established.Using the established PSO-BP model,the complex nonlinear mapping relationship between temperature,tensile rate and strain and deformation resistance of QP1180 steel is established by training directly from the data obtained through experiments.Through the verification of test samples,the predicted test data predicted and actual numerical error evaluation is 6.06%,which can basically satisfy our theoretical research.Based on the PSO-BP prediction model,the training data of two groups of experimental samples were predicted respectively,and then the two groups of experimental data were fitted by the function relations in this paper.It can be found that the prediction effect of the neural network is consistent with that obtained by fitting,and the variation law of deformation resistance can be revealed.Again some Non-sample data prediction based on this model,get under 420 ℃ and 440 ℃,the stretching rate were 0.2 and 0.4 predicted results,Through cartographic research,it is found that the trend of these four curves and the complex relationship between deformation resistance and influencing factors can be reflected correctly.The PSO-BP prediction model can provide a very reliable theoretical research assistant for us to explore the deformation resistance of steel materials under high temperature deformation.