基于Kriging模型的等速万向节滑套冷精整工艺的优化

Optimization on cold sizing process of CVJ spherical housings based on Kriging model

针对车用等速万向节滑套在实际冷精整后杯壁端部出现不平整的问题,基于Deform有限元分析软件,提出Kriging代理模型与粒子群算法相结合的方法,以杯壁高度差为优化目标,使用拉丁超立方抽样法对预成形件尺寸参数进行抽样、完成实验设计,通过数值模拟得到仿真结果,并拟合成相应的Kriging模型。采用MATLAB将Kriging模型与粒子群算法相结合,对因素进行全局寻优,得到最小杯壁高度差的预成形件尺寸。产品试制后,杯壁高度差由原来工艺的12 mm缩小为3 mm,验证了此优化方法的实用性和正确性,提高了锻件的材料利用率。

For the problem of uneven cup wall end after cold sizing for CVJ spherical housing, setting the height difference of cup wall as the optimization objective, a combination method of Kriging surrogate model with particle swarm optimization was applied based on finite element analysis software Deform. Then, the dimension parameters of perform part was sampled by the Latin hypercube sampling method, and the experimental design was completed. Next, the simulation result was obtained by FEM and fitted to the corresponding Kriging model. Furthermore, the Kriging surrogate model was combined with the particle swarm optimization by software MATLAB, the minimum height difference of perform part cup wall was obtained by the global optimization on factors. At last, the height difference of the industrial trial production is reduced from 12 mm to 3 mm, which proves the practicability and validity of the optimization method, and the material utilization rate is improved as well.