基于多目标优化的钢管拉拔成形过程设计  

Forming Process Design of Pipe Drawing Based on Multi-objective Optimization

作　　者：胡龙飞[1] 刘全坤[1] 王成勇[1] 胡成亮[1] 冯秋红[1] 

机构地区：[1]合肥工业大学材料科学与工程学院

出　　处：《农业机械学报》2007年第10期161-164,共4页Transactions of the Chinese Society for Agricultural Machinery

摘　　要：针对钢管在拉拔成形中出现的拉拔力过大、成形后钢管的残余应力大的问题,提出了基于"FEM-ANN-MOGA"方法的拉拔过程优化方案。结合正交设计、有限元模拟技术和BP神经网络,建立了拉拔应力、残余应力与成形参数之间多目标优化的非线性映射模型。采用基于向量评价的多目标遗传算法和小生境技术,求得了均匀分布的Pareto最优解。通过定义满意度函数,选出了符合要求的满意解,并对其进行了仿真。仿真结果与优化结果基本吻合,验证了该优化方法的正确性与可行性。Considering the problems of oversize drawing force in pipe forming and the excessive stresses in pipe surface, a concept of multi-objective optimization was presented based on ＂FEM -ANN - MOGA＂ method. Orthogonal design, finite-element simulation and BP neural network were combined together to build the nonlinear mapping relations between drawing stress, residual stress and forming parameters. In the process of optimization, the multi-objective genetic algorithm based on vector evaluation technique and niche technique was adopted to obtain the evenly distributed Pareto-optimal solutions. By defining a satisfactory degree function, the satisfactory solution was selected, and the FEM simulation verification was carried out. The simulation results accorded with the optimized ones perfectly, which showed that the method is feasible and credible.

关 键 词：钢管 拉拔 多目标优化 BP神经网络 遗传算法 

分 类 号：TG356.5[金属学及工艺—金属压力加工]