基于灰色关联的铝合金板拉深成形扭曲回弹工艺参数优化

Optimization on parameters of twist springback process for aluminum alloy sheet deep drawing based on grey correlation

针对铝合金板成形工艺参数如何选取的难点,利用灰色关联准则对铝合金板成形质量进行分析,通过因子关联度的方差分析,获得铝合金板成形工艺中的主要影响因子。为了减少板料成形工艺参数优化时间,以铝合金板成形中主要影响因子为设计变量,以板料成形后扭曲回弹、增厚、减薄为成形目标,使用拉丁超立方抽取样本,通过Dynaform软件进行数值模拟获得训练样本,利用人工免疫算法训练RBF神经网络,建立主要影响因子与成形目标之间的RBF神经网络近似模型,最后采用人工免疫算法对该模型进行优化,获得最优工艺参数。以Numisheet'96 S梁为研究对象,利用本文所提出的方法进行拉深成形研究,通过对比分析优化前后的成形结果,证明了该方法能极大地提高铝合金板的成形质量。

For the difficulty in selecting the aluminum alloy sheet forming process parameter, the aluminum alloy sheet forming quality was analyzed by the grey correlation criterion. The main influence factors of aluminum alloy sheet forming were obtained by the analysis on cor- relation factor variance. To reduce the time of parameter optimized in the sheet forming, the main influence factors were chosen as the de- sign variables, and the twist springback, thickening and thinning after sheet forming were regarded as forming targets. Based on the latin hypercuhe method, the training samples were obtained by Dynaform software. RBF neural network was trained by the artificial immune al- gorithm, and the RBF neural network approximation model between the main influence factors and the forming quality was established. Finally, the model was optimized, and the optimal process parameters were obtained. The deep drawing forming for S-rail of Numisheet＇96 was researched by the above method. The resuhs show that the quality of aluminum alloy sheet forming can be improved greatly when comparing the forming results before and after optimization.