挤压锻造模具形状的混合智能优化

数值仿真技术已成为挤压锻造模具设计中的重要评价手段，有助于设计人员更好的理解设计空间。然而，由于计算时问长，目标函数无法显式表述，搜索空间可能不连续等因素，很难直接基于数值仿真通过梯度法进行设计优化。非梯度的优化方法比如遗传算法、模拟退火法等由于需要更多的仿真次数，也很难实现。本文基于人工智能中的归纳学习方法，提出了一种针对挤压锻造工艺优化的基于数值仿真归纳知识的混合优化方法，并通过一个挤压锻造的实例验证了该方法的有效性。

Hybrid intelligent optimization for extrusion and forging die shape design

Numerical simulation technique has become the important verifying tools in extrusion and forging die shape design. It helps designers to know the design parameters space better. However, the vast computation cost, the implicity of object function and discontinuousness of search space limit the application of traditional gradient-based optimization methods. Some nongradient based optimization techniques, such as generic algorithm and simulated annealing usually need more simulation runs and become infeasible in a large scale simulation. In this paper, a hybrid intelligent optimization method based simulation knowledge is developed aiming at extrusion and forging process. The result of an extrusion-forging die shape optimization shows that this method is valid and feasible.