基于随机聚焦搜索算法的冲压成形工艺优化  

Optimization of stamping process based on stochastic focusing search algorithm

作　　者：龙玲[1,2] 殷国富[1] 邹云[2] 肖兵[1] 

机构地区：[1]四川大学制造科学与工程学院,四川成都610065 [2]成都纺织高等专科学校机械工程与自动化系,四川成都611731

出　　处：《计算机集成制造系统》2012年第2期314-320,共7页Computer Integrated Manufacturing Systems

基　　金：国家自然科学基金资助项目(51075287);国家科技重大专项资助项目(2010ZX04015-011)~~

摘　　要：针对板料冲压成形工艺优化问题,研究了一种群集智能算法。该方法通过正交实验与数字化仿真技术相结合获取神经网络的学习样本,利用反向传播神经网络构建随机聚焦搜索算法的目标函数模型。在此模型基础上,应用随机聚焦搜索算法对板料冲压成形的工艺参数进行优化。以深盒形件为例,将优化后的工艺参数输入eta/DYNAFORM仿真模型进行验证,结果表明该算法可获得较好的成形质量。为了进一步验证随机聚焦搜索算法在执行效率及寻优的全局搜索方面的优越性,与遗传算法的优化结果进行对比分析,说明随机聚焦搜索在板料冲压成形工艺参数优化方面是一种较好的优化算法。Aiming at the problem of sheet metal stamping process optomization, a kind of swarm intelligence algorithm was studied. The training samples for neural network were obtained by combining numerical simulation tech nology with orthogonal experiments, and the objective function model of Stochastic Focusing Search （SFS） algorithm was constructed by using back propagation neural network. On the basis of this function model, SFS algo rithm was used to optimize the sheet stamping process parameters. By taking a deep box-shaped workpiece for instance, the optimized process parameters were input into eta/DYNAFORM simulation model to validate this algorithm, and the result showed that good forming quality was obtained. To further verily the advantages of SFS algorithm in efficiency and optimization of global searching, the genetic algorithm＇s optimal result was compared with SFS. The result showed that SFS was a better optimization algorithm in sheet stamping process parameter optimization respect.

关 键 词：板料成形 随机聚焦搜索算法 工艺参数优化 数值模拟 正交试验 反向传播神经网络训练 

分 类 号：TG386[金属学及工艺—金属压力加工]