上海交通大学学报

上海交通大学学报（自然版） ›› 2011, Vol. 45 ›› Issue (07): 966-969.

• 自动化技术、计算机技术 • 上一篇    下一篇

求解含等式约束优化问题的遗传算法

胡宽1，2，常新龙1，宋笔锋2，张琳3，龙兵1，余堰峰1   

  1. （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）
  • 收稿日期:2010-07-09 出版日期:2011-07-29 发布日期:2011-07-29
  • 基金资助:

    国家自然科学基金资助项目(10402035)

Genetic Algorithm to Solve Optimization Problem with Equality Constrains

 HU  Kuan-1, 2 , CHANG  Xin-Long-1, SONG  Bi-Feng-2, ZHANG  Lin-3, LONG  Bing-1, YU  Yan-Feng-1   

  • Received:2010-07-09 Online:2011-07-29 Published:2011-07-29

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.

关键词: 遗传算法, 等式约束, 降维, 排序, 优化

Abstract: 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.

Key words: genetic algorithm, equality constraint, descending dimension, rank, optimization

中图分类号: