Research And Prediction Of Servo Tensile Property Of Medium-Mn Steel Based On Neural Network

The third-generation automobile high-strength steel,medium-manganese steel is one of the third-generation automobile steel development routes with the advantages of high strength and plasticity and lower cost.The properties of flexibility,slide-adustability and controllablity of the servo press enables the metal sheet to achieve higher quality during the forming process.At present,for the research of medium manganese steel,more attention is paid to the regulation of the structure and composition of medium manganese steel.For the tensile speed sensitive materials,changes in forming speed will cause the change of fomring quality.However,the influence of servo tensile parameters on the elongation of medium-manganese steel is rarely studied.In this paper,different servo tensile parameters are investigated,and the neural network is used to predict the elongation.The variation rule of the elongation of medium manganese steel under servo tensile is studied.The material used in this paper is the third-generation automotive high-strength steel,medium-manganese steel with a thickness of 1.6mm,the steel has beend heat treated.Research and prediction of the elongation changing rule is carried out by tensile test under servo control.The BP and RBF neural networks were used to predict the elongation.The prediction results were compared and analysed.Finally,the RBF neural network was used to predict the elongation trend of 3,600 samples.Following conclusions were obtained:(1)At different tensile speeds(3mm/min,15mm/min,75mm/min,150mm/min,300mm/min),the elongation of the tensile specimen of medium manganese steel exhibits tensile speed sensitivity.At tensile speeds from 3 mm/min to 150 mm/min,the elongation of medium manganese steel gradually decreases.However,the yield strength and tensile strength of medium manganese steel did not change significantly.(2)Through the servo control,the different tensile speed positions were tested,and the influence of the position on the elongation of the medium-manganese steel was studied,and the positions were 3 mm,4.5 mm,and 6 mm,respectively,all in the uniform deformation range.In the deformation interval,three tensile speeds were set and the tensile speed combination experiments were performed.The rates were 3 mm/min,15 mm/min and 75 mm/min,respectively.It is found that the change of tensile speed before the 3mm and 4.5mm positions hardly affects the elongation of medium manganese steel,and the change of tensile speed after 4.5mm will reduce the elongation of medium manganese steel.The degree of elongation reduction is related to the combination of tensile speeds.(3)Prediction of the elongation of medium manganese steel by BP neural network and RBF neural network.Both have achieved good results.Through the comparison of residual analysis and the comparison of the minimum average relative error,it is concluded that the RBF neural network has higher accuracy and stability.(4)Trend prediction of experimental results by RBF neural network is carried out.Tensile speed combination predictions were made at 4 mm and 5 mm locations,respectively.The elongations of 3,600 data were predicted,and the trend of the elongation distribution showed good agreement with the previous experimental results.Finally,different optimal ranges of servo tensile parameters are given.