| Springback is one of the main quality defects in the stamping process of automobile panels, which determines the final shape of parts formed. Considering that springback has great influence on product shape and dimensional precision , and which brings difficulties to subsequent welding and assembling procedure, how to predict and control springback in stamping becomes more necessary to improve the components' quality and ensure the assembling efficiency.In this paper, auto crossbeam is regarded as the research object, and process control method and FEM is adopted to study springback of automobile panels. The application of intelligent optimization technology in springback control is discussed mainly. The equivalent drawbead restraining force model is optimized firstly, and the reasonable arrangement of force is obtained according to the combination of uniform design, response surface methodology and multi-objective genetic algorithm. Based on this, the mathematical model between key process parameters and the mean springback of crossbeam is built by means of neural network technology and response surface methodology. The ideal springback prediction model is gained through comparing all of modeling methods. With using genetic algorithm, immune algorithm, swarm intelligence theory and simulated annealing algorithm, six process parameters which have important effect on springback are optimized aiming at minimizing springback of crossbeam. From which, a feasible scheme for stamping is obtained by analyzing and comparing performance of each algorithm. The optimization result shows that the springback of crossbeam is controlled effectively.The study solves the problem that the optimization method for springback control is single at present with use of several intelligent optimization technologies, and is significant to die design and technical preparation in practice. |
