一种混合交叉策略的多目标演化算法及其性能分析  

Multi-objective Evolutionary Algorithm Based on Double Crossover and Its Performances Analysis

作　　者：汪祖柱[1] 程家兴[1] 

机构地区：[1]安徽大学计算智能与信号处理教育部重点实验室

出　　处：《系统仿真学报》2005年第10期2383-2387,共5页Journal of System Simulation

基　　金：国家自然科学资金(60175018);安徽省青年教师资助项目(2004jq108);安徽大学人才队伍建设经费资助

摘　　要：演化算法(EA)是求解多目标优化问题(MOP)重要而有效的方法,而应用演化策略、技巧是改善解性能的重要途径。作者叙述了多目标优化问题的有关概念,结合已有算法中的方法,设计了基于两种交叉操作相互结合的多目标演化算法(MOEADC),并且分析相关性能。该算法不仅具有较高的计算效率,而且具有较好的收敛性能,并且运用了有关方法维护了解集的分布性能。算例结果表明该算法的良好性能。Evolutionary algorithm is a main and effective method solving a multi-objective optimization problem （MOP）. It is a significant approach, in which solutions＇ performance of MOP is improved, by using all kinds of evolutionary strategies and techniques. Some concepts about a multi-objective optimization problem were described, and with some operators in several noted algorithms, a multi-objective evolutionary algorithm based on double crossover was designed, and relevant performances were analyzed. The proposed algorithm is not only computationally efficient, but also has good convergence performance, and some method is also applied to maintain solutions＇ diversity. The experiment shows it performs well.

关 键 词：演化算法 多目标演化算法 性能分析 演化策略 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]