Springback Prediction of Sheet Metal Formed by Rubber Bladder Based on Wavelet Neural Network

The sheet metal parts formed by rubber bladder generally have the problem of large springback. The forming and springback process of rubber bladder forming parts were simulated by using the finite element simulation software Pam-stamp2G. The simulation results were compared with the experimental results to verify the feasibility of numerical simulation instead of actual forming. A typical aircraft wing rib sheet metal part was studied. Taking forming pressure, punch fillet radius,sheet thickness and flanging height as input layers and springback as output layer, the 4-6-1 three-layer wavelet neural network model was established. Based on the orthogonal test design and springback results of numerical simulation, the sample data were obtained and the network model was trained and tested. Based on the two groups of data parameters, the verification experiment for the net work model was carried out. The results show that the errors between the prediction value of the two groups of wavelet neural network and the corresponding test value were only 4.57% and 4.53%, respectively, which meets the requirements of industrial production and verifies the reliability of the wavelet neural network prediction model.