用于求解约束优化问题的自适应佳点集进化算法  

Novel Constrained Optimization Evolutionary Algorthm Based on Self-adaptive Good Points Set

作　　者：唐素勤[1,2] 蔡自兴[1] 江中央[1] 肖赤心[1] 

机构地区：[1]中南大学信息科学与工程学院,湖南长沙410083 [2]广西师范大学计算机科学与信息工程学院,广西桂林541004

出　　处：《小型微型计算机系统》2009年第11期2248-2253,共6页Journal of Chinese Computer Systems

基　　金：国家基础研究项目(A1420060159)资助;国家自然科学基金项目(60404021)资助;湖南省院士基金项目(06IJY3035)资助

摘　　要：提出一种用于求解约束优化问题的自适应佳点集进化算法.新算法利用佳点集原理设计多点交叉算子,该交叉算子能够根据父代个体的相似度自适应调整交叉点的位置和子代个体的数目,产生具有代表性的子代个体.在约束处理技术上,改进了Deb的三条比较准则,提出一种新的适应度函数用于比较个体优、劣的比较准则.通过对13个标准测试函数的试验比较验证了新算法的有效性和稳健性.Novel constrained optimization evolutionary algorithm based on self-adaptive good points set（COEAAGP） is proposed to tackle constrained optimization problems （ COPs ） in this paper. A many points crossover operater is redesigned by using the good points set principle in number theory to adjust the crossover points locality and the amount of the offspring freely by the similarity degree of the population , so that it can produce a small but representative set of points as the potential offsping. A new fitness function, where feasible solutions and infeasible solutions are compared based on their constraint violation is proposed. The new approach is compared against other evolutionary optimization techniques in thirteen benchmark functions. The results obtained show that the new approach is a general, effective and robust method.

关 键 词：进化算法 佳点集 约束优化 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP301[自动化与计算机技术—控制科学与工程]