基于信赖域改进的动态Kriging模型的冲压成形工艺参数优化

Optimization of Stamping Process Parameters Based on Improved Dynamic Kriging Model with Trust Region

在工程优化问题中使用基于代理模型的优化方法能有效提高优化效率,减少优化时间。为了提高基于期望最大加点准则(EI)的动态Kriging建模效率与精度,利用信赖域方法与最小响应面加点准则(MSP)对期望最大加点准则进行改进,提出了改进的混合加点准则,发展了基于信赖域改进的动态Kriging优化策略。以NUMISHEET2002标准考题中翼子板为研究对象,将5条拉延筋阻力作为设计变量,板料减薄率作为响应值,基于改进的混合加点准则,建立翼子板成形工艺参数与成形质量之间的动态Kriging模型,进一步利用粒子群算法获得最优工艺参数。结果表明:优化后的工艺参数能有效消除板料成形过程中的拉裂缺陷,提高翼子板成形质量。

The use of surrogate model-based optimization methods in engineering optimization problems can effectively improve optimization efficiency and reduce optimization time.In order to improve the modeling efficiency and accuracy of the Kriging model based on the expected maximum sampling criterion(EI),this paper used the trust region and the minimum response surface sampling criterion(MSP)to improve the expected maximum sampling criterion,proposed an improved hybrid sampling criterion,and developed an improved dynamic Kriging optimization strategy based on trust region.Taking the fender in the NUMISHEET2002 as the research object,using the resistance of 5 drawbeads as the design variable,the sheet metal thinning rate as the response value,and using the improved hybrid sampling criterion to establish the dynamic Kriging model between the forming process parameters and quality of the fender,the particle swarm algorithm was used to obtain the optimal process parameters.The result shows that the optimized process parameters can effectively eliminate the cracking defects in the sheet metal forming process and improve the forming quality of the fender.