协同进化蚁群算法及其在多目标优化中的应用  

Co-Evolutionary Ant Colony Algorithm and Its Application to Multi-Objective Optimization Problems

作　　者：陶振武[1] 肖人彬[1] 

机构地区：[1]华中科技大学CAD中心,武汉430074

出　　处：《模式识别与人工智能》2005年第5期588-595,共8页Pattern Recognition and Artificial Intelligence

基　　金：国家自然科学基金(No.60474077);教育部优秀青年教师计划(2003)资助项目

摘　　要：针对蚁群算法ACS的控制参数难以确定和早熟停滞等缺陷,提出了进化蚁群系统算法模型EACS.EACS通过引入选择、交叉和变异等操作,实现算法参数的自适应调整。标准测试实例的计算结果表明,EACS算法能够克服上述缺陷,便于工程应用,根据协同进化的思想进一步提出了多目标协同进化蚁群算法CACSM.CACSM中的多个群体协同进化,每个群体对应一个目标,并对其它群体的搜索产生影响,CACSM实现了仅通过算法一次运行便求得若干Pareto最优解,提供了更大的决策空间。最后通过一个多目标组合优化问题——岩石钻孔机路径选择问题的求解,验证说明了CACSM的有效性和适用性。In view of the drawbacks of Ant Colony System （ACS） such as the difficulty in choosing the parameters, premature and stagnation, the Evolutionary Ant Colony System （EACS） algorithmic model is proposed, EACS realizes the adaptive adjustment of the parameters by introducing the genetic operations of selection, crossover and mutation. Some experimental results on the TSP benchmarks show that the EACS can overcome the drawbacks above and it is convenient for engineering application. Furthermore, a Co-evolutionary Ant Colony System for Multi-Objective Optimization Problem （CACSM） based on EACS is proposed according to the ideas of co-evolution. In CACSM, the objectives are optimized with the co-evolution of multiple ant colonies. Each colony is corresponding to an objective, and it will also influence the activities of other colonies. CACSM can find the Pareto solution set of the MOOP in one run and helps to extend the decision space of the decision-maker. When CACSM is used to solve the problem of the path planning of rock-drilling machine, a typical MOOP in discrete space, the results validate the effectiveness and engineering applicability of it.

关 键 词：蚁群算法 旅行商问题 协同进化 多目标优化问题 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]