求解含等式约束优化问题的遗传算法  

Genetic Algorithm to Solve Optimization Problem with Equality Constrains

作　　者：胡宽[1,2] 常新龙[1] 宋笔锋[2] 张琳 龙兵[1] 余堰峰[1] 

机构地区：[1]第二炮兵工程学院二系,西安710025 [2]西北工业大学航空学院,西安710072 [3]航天科技集团公司第43研究所,西安710025

出　　处：《上海交通大学学报》2011年第7期966-969,974,共5页Journal of Shanghai Jiaotong University

基　　金：国家自然科学基金资助项目(10402035)

摘　　要：针对遗传算法较难处理含等式约束的优化问题,在设计变量独立性分析的基础上对等式约束采用了降维处理方法,不仅使等式约束在优化时始终严格满足,而且经降维处理后优化问题仅包含不等式约束;然后,借鉴多目标优化思想,提出了从个体违反约束程度和违反次数2方面同时对种群进行排序,使算法对个体的排序和选择更符合实际.实例验证了该算法的有效性和可行性.Aiming at difficultly solving optimization with equality constraint in genetic algorithm,based on independence analysis of design variables,a descending dimension method was used to deal with equality constraint.In this way,not only equality constraints can be strictly satisfied during optimization,but also there are only inequality constrains in optimization.Furthermore,referring to the multi-objective idea,individuals was ranked through violation degree and violation times simultaneously,which is more in accord with practice.Finally,three numerical examples were used to examine the proposed method,satisfying results were achieved,which indicates the method is valid and feasible.

关 键 词：遗传算法 等式约束 降维 排序 优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]