融合全局逼近与局部精化的协同优化方法

Collaborative optimization method via global approaching and local precision

针对标准协同优化方法存在对初始点选取敏感、优化结果易收敛于局部最优解等问题,通过对松弛因子法特点的分析,给出了一种综合应用固定松弛因子和动态松弛因子方法,融合全局逼近与局部精化的改进协同优化方法,并将Kriging近似模型引入到改进的协同优化方法中,减小了迭代计算量,提高了优化效率,可快速精确地收敛到全局最优解处.通过汽车盘式制动器的优化设计,验证了本文方法的有效性与稳定性.

Due that the standard collaborative optimization method is sensitive to the initial point selection and easily converges to the local optimum, an improved collaborative optimization method is proposed based on the feature analysis via relaxation factor method.By combining global approaching with local precision via the fixed and dynamic relaxation factors, the Kriging approximation model is introduced to reduce iterative computation and improve optimization efficiency for global optimum.Therein, the disc brake is used as a case to illustrate the feasibility and stability of the proposed method.