论文部分内容阅读
为了求解结构和控制器多目标集成优化设计问题,获得用H2或H∞范数定义的最优系统性能和控制代价的所有非劣解,采用了基于动态加权计划的进化算法.该算法采用的权值随着进化代数的变化而变化,而不是固定值,通过结合线性矩阵不等式(LMI)或Riccati控制器设计方法,一次运行就能够得到均匀分布的非劣解.与加权的单目标遗传算法相比,该方法可以大大减少求解集成优化设计问题的计算强度.通过汽车悬架的集成设计表明了该方法的有效性.
In order to solve the problem of multi-objective integration optimization of structure and controller design and obtain all the non-inferior solutions of optimal system performance and control cost defined by H2 or H∞ norm, a dynamic algorithm based on dynamic weighting scheme is adopted. The weights change with evolutionary algebra rather than fixed values, and a uniform distribution of non-inferior solutions can be obtained in a single run by combining linear matrix inequalities (LMIs) or Riccati controller design methods.With the weighted single-objective genetic algorithm This method can greatly reduce the computational intensity of solving integrated optimization design problems.The design of the car suspension shows the effectiveness of this method.