PSO-BP Neural Network-Based Optimization of Automobile Rear Longitudinal Beam Stamping Process Parameters

This paper uses a sheet forming optimization method based on numerical simulation and PSO-BP neural network model, and optimizes the forming process parameters of a vehicle’s rear longitudinal beam as an example with the maximum thinning rate. Combined with DYNAFORM software, the process parameters with high correlation with the maximum thinning rate were the crimping force and friction factor based on grey system theory (GS theory). Based on these two highly correlated process parameters, 400 sample points of Latin supercube sampling were used to train and verify the PSO-BP neural network model, and then the particle swarm algorithm was used to optimize the process parameters of the model. The result predicted that the pressure edge force was 422.5279 kN. When the friction coefficient is 0.10464, the maximum thinning rate is 22.8133%, while the maximum thinning rate obtained by numerical simulation of the predicted process parameters using DYNAFORM is 23.724%, so the difference between the predicted value and the simulated value is only 0.9107%, and the error rate of the two is only 3.99%. The neural network model established is accurate, and combining numerical simulation with the PSO-BP neural network model can be used as a new method to optimize the process parameters during stamping.