【摘要】 以某新开发MPV高强度钢纵梁为研究对象,将压边力、摩擦系数和凹模圆角作为为试验因子,以选定截面线沿y方向的最大位移为评价目标,建立了有限元仿真均匀试验设计方案,利用试验结果建立基于人工神经网络的回弹预测模型,并通过随机建立的仿真数据验证了其准确性。结果表明,可以将神经网络、有限元仿真和均匀试验结合起来用于板料冲压工艺参数的优化,并且可以明显缩短工艺参数的优化时间,提高了工艺设计效率。更多还原

【Abstract】 The test makes the high strength steel automotive longeron as reasearch object,takes the blank holder pressure,frictional coefficient and die fillet as test factors and gets the springback value along Y-axis as evaluation objectives.Uniform design experiment of FE simulation model is set up basing on what we’ve selected.Then a network prediction model of high strength steel automotive rail’ s springback is created,being based on the analysis results.At last,better combination of parameters is found by using the network in the vicinity of the optimal combination of uniform design experiment.The results have been proved to be correct by using FEM simulation.The result shows that the neural network,numerical simulation and uniform pilot program can be combined to optimize the stamping process parameters.The time of parameters’optimization will be reduced obviously.At the same time the design-efficiency could be greatly increased.更多还原