人机结合多目标协同进化算法求解卫星设备布局

发布时间：2018-06-17 05:05

本文选题：多层系统 + 多目标优化　；参考：《计算机集成制造系统》2017年11期

【摘要】：针对多层系统多目标优化的多舱段卫星设备布局设计的工程满意Pareto最优解集求解困难问题,基于协同进化框架,给出一种人机结合的多目标协同差异进化算法。算法将人机结合方法引入基于协同进化框架的多目标协同进化算法,构建含人工和算法的精英解集合的档案,提出用差异进化算法随机自动实现由算法子个体和人工子个体两两组合生成个体的3种方式,用于子系统的人机结合。在迭代前期采用子系统人机结合增加算法的多样性和分散性,在迭代后期采用系统人机结合增加算法的收敛性和分布性,从而提高基于协同进化框架的多目标协同进化算法求解这类多层系统的多目标优化的计算性能,获得Pareto最优解集。最后,通过案例验证了该算法具有较好的计算性能。
[Abstract]:In order to solve the problem of engineering satisfactory Pareto optimal solution set for multi-level system multi-objective optimization satellite equipment layout design, a multi-objective co-differential evolution algorithm based on co-evolution framework is proposed. The man-machine combination method is introduced into the multi-objective co-evolution algorithm based on co-evolution framework, and the file of elite solution set with artificial and algorithm is constructed. In this paper, three methods of generating individuals by the combination of algorithm sub-individuals and artificial sub-individuals are proposed, which are used in man-machine combination of subsystems. In the early stage of iteration, the diversity and dispersion of the algorithm are increased by using the man-machine combination of subsystems, and the convergence and distribution of the algorithm are increased by the combination of system and man-machine in the later stage of iteration. Therefore, the multi-objective co-evolution algorithm based on co-evolution framework can improve the computational performance of solving multi-objective optimization for this kind of multi-layered system, and obtain the Pareto optimal solution set. Finally, a case study shows that the algorithm has good computational performance.
【作者单位】：大连理工大学机械工程学院;大连理工大学计算机科学与技术学院;
【基金】：国家自然科学基金资助项目(61472062)~~
【分类号】：TP18;V423

【相似文献】