基于遗传算法的挤压模具型腔形状优化设计

Optimizing of extrusion die profile based on genetic algorithm

以磨损为目标建立优化数学模型,采用B样条函数插值描述模具型腔轮廓形状,基于有限元和修正的Archard磨损模型计算结果训练BP神经网络,建立模具型腔控制点与目标函数之间的映射关系,计算遗传算法的适应度值,优化模具型腔.研究结果显示:采用本方法得到的模具型腔形状,与锥形模相比,沿其表面最大磨损深度降低了63.9%,磨损深度分布均匀,说明此设计方法可行.

Based on wear depth on die profile surface, the optimizing mathematical model was built. A method of B-spline function interpolation was used to describe extrusion die profile. The results of FEM simulation and modified Archard theory were applied to train BP neural network. The nonlinear mapping relations between reference points of die profile and wear depth were obtained, the fitness of genetic algorithm was calculated and die profile was optimized. The result shows that the maximum wear depth along the optimal profile, compared with the conical profile, decreases by 63.9% and the distribution of wear depth is more uniform. It shows that the design method is feasible.