摘要
为解决复杂系统可靠性设计的多目标优化问题,采用NSGA Ⅱ进化算法,在一次性获得大量均匀分布的非劣解后,根据定义的满意度函数对每个非劣解进行模糊评价。运用群体排序、小生境、比较操作和精英保留等关键技术,将整个非劣解集分为若干个具有不同性能特征的类,决策者可根据各自的偏好从中选择最终的满意解。分析结果表明,进化算法具有较高的优化效率。
The NSGA-Ⅱ evolutionary algorithm was adopted to solve multi-objective reliability optimization in this paper. After many Pareto-optimal solutions obtained in a single run, the fuzzy evaluation of each Pareto-optimal solution was made based on sub-objective satisfactory degree, The total Pareto-set could be divided into several subsets with different characteristics through the technologies such as group sequencing, small environment, comparison and reservation. The satisfactory solution was selected according to the decision maker's fuzzy preference for each sub-objective. It was showed that the evolutionary algorithm was high efficiency in solving a multi-objective reliability optimization problem.
出处
《上海航天》
2004年第2期7-10,49,共5页
Aerospace Shanghai
基金
国家自然科学基金资助项目(69931040)
关键词
进化算法
系统可靠性
多目标优化
模糊评价
满意解
Reliability
Multi-objective optimization
Evolutionary algorithm
Pareto-optimal solution
Fuzzy evaluation
Satisfactory solution
