Study On Neural Network Prediction Technique Of Variable Blank-holder Force In Deep Drawing Forming

With the rapid development of sheet metal forming technology, The combination of artificial intelligence technology and plastic forming technology is paid more and more attention. As a new science of intelligent technology, the artificial neural network can solve many complex fit of problems by the highly self- responsiveness and the ability of self-learning, it can provide a good solution for the optimization and prediction of nonlinear model.As a typical processing technology in the field of plastic processing, The blank-holder force(BHF) has a significant influence of the deep drawing forming effect of sheet. It is necessary to product high quality drawing part that can ideally realize the real-time prediction and necessarily control of the BHF. However, in actual production process, the BHF related to many factors, the influence degree of each factor for it is different. In the process of sheet metal drawing, and how to accurately predict change rule of the variable blank-holder force(VBHF) becomes a complex nonlinear problem.First, in this paper, a model for the effect of BHF in rectangular box deep drawing was established, based on the plastic forming theory and principle of sheet metal drawing, analyzed the characteristics of rectangular box deep drawing forming and several typical failure modes, and then put forward the method and measure criteria of judging failure; the safety limit diagram of VBHF in sheet metal drawing is determined, detailed derivation the critical BHF formula of flange area and corner wrinkling and rupture in rectangular box drawing forming, and analyzed the factors affecting the critical BHF.Furthermore, in the process of rectangular box drawing, the several typical ways of BHF loading is analyzed by DYNAFORM simulation, and the simulation results are summarized and compared, and the ideal loading mode of VBHF is obtained, the data samples were selected on the basis of this.Then, according to the basic theory, basic formula and basic method of artificial neural network, the neural network prediction model is established. In view of the research condition and the characteristics of the samples, the sample data is acquired by “fixed gap method”. The more optimal loading curve of VBHF in rectangular box deep drawing forming was successfully predicted. And the simulation results of the more excellent positive trapezoid curve of VBHF and the simulation results of the fitting curve of VBHF obtained by network prediction are compared, verify that the latter is more in line with the production requirements.Eventually, the intelligent prediction system software of VBHF is developed with the help of mathematical analysis software MATLAB7.8 and programming software VB(6.0), it is easy to operate, easy to learn, high security, high accuracy and high reliability.