协同优化计算方法的分析与改进

针对协同优化算法对初始点选取敏感的问题,从保证系统级优化可行有解的角度,以一典型算例为研究对象,对已有的改进措施进行了分析与完善.当初始点在可行域内时,采用固定因子松弛法进行求解,并给出了动态松弛法失效的原因.当初始点在可行域外时,利用动态松弛因子和固定松弛法因子相结合的策略进行求解,并对动态松弛法优化结果精度不高,以及其值在可行域边界徘徊的情况,分别从系统级和学科级给出了相应的解释.最后,采用减速器优化算例,对所提出的优化策略进行了验证,得到了较为满意的结果.

Analysis and Improvement on Algorithm of Collaborative Optimization

In order to ensure the solvability of system level optimization,a typical example is taken as the object and the exiting improved algorithms are analyzed and perfected based on the formulation of system level optimization.When the initial points are chosen in the feasible region,the fixed relaxation method is implemented and reasons that cause the failure of dynamic relaxation method are presented.When the initial points are chosen out of the feasible region,a combination strategy of fixed and dynamic relaxation method is implemented for the optimal computation.Meanwhile,the optimization results obtained from dynamic relaxation method are not accurate and wander at the feasible region boundary.The explanation for these phenomena is presented in the view of both system level and subsystem level.In the end,the speed reducer example proves that the proposed method has good performance.