有界区域上多峰函数全局优化问题的改进演化算法  

An Improved Evolutionary Algorithm for Solving Multimodal Function Global Optimization Problem on a Bounded Area

作　　者：黄樟灿[1] 李炜[1] 

机构地区：[1]武汉理工大学理学院,湖北武汉430070

出　　处：《武汉大学学报（理学版）》2007年第1期55-58,共4页Journal of Wuhan University:Natural Science Edition

基　　金：国家自然基金资助项目(60133010)

摘　　要：针对演化算法求解有界区域上的多峰函数全局优化问题中,保持种群多样性和搜索效率的矛盾,提出了一种结合了多样性维持机制和加速算子的改进演化算法并对算法作了收敛性分析.实验结果表明,在低维情况下,算法容易收敛于全局最优解,优于基本遗传算法;在高维情况下,虽不满足收敛条件,但算法仍能收敛到本文意义下的ε-满意解,优于基本粒子群算法.In this paper, we mechanism and accelerating population bounded ar propose an improved evolutionary algorithm co operators, which focuses on the contradiction b diversity and search efficiency in solving ea. The convergence analysis shows our muhimodal function global algorithm can converge to mbining diversity maintaietween the maintenance of optimization problem on a global optimization under some circumstance. When dealing with low dimensional problems, our algorithm is prone to converge to global optimization and outperforms canonical genetic algorithm, while with high dimensional problems, the converging condition cannot be satisfied, our algorithm can converge to the so-called e-satisfied result and outperforms canonical particle swarm optimization algorithm.

关 键 词：演化算法 多样性维持机制 加速算子 收敛性分析 

分 类 号：O242[理学—计算数学] Q332[理学—数学]