一种提高遗传算法全局收敛性的方法  

Improving the Global Convergence of the Genetic Algorithm

作　　者：何大阔[1] 王福利[1] 

机构地区：[1]东北大学信息科学与工程学院,辽宁沈阳110004

出　　处：《东北大学学报（自然科学版）》2003年第6期511-514,共4页Journal of Northeastern University(Natural Science)

基　　金：教育部高等学校骨干教师资助计划;辽宁省自然科学基金资助项目(002013)

摘　　要：通过对遗传算法过早收敛原因的分析,认为遗传算法出现过早收敛主要与问题解的分布状况、种群个体的分布情况及遗传算子的应用有关,提高算法全局收敛性能的核心就是如何使算法科学地处理种群多样性及识别个体对全局收敛性能的作用·提出几类与遗传算法全局收敛性能关系较大的个体,并结合小生境进化共享函数思想,形成一种旨在提高遗传算法全局收敛性、求解全局最优解的遗传算法,仿真结果验证了这种算法良好的全局收敛性能·Based on the analysis of the reason of the early convergence of the genetic algo rithm, it was shown that the early convergence of the genetic algorithm is main ly concerned with the distribution of the solutions to the problem, the distribu tion of the individuals of the population and the application of the genetic ope rators. The key to improve the performance of the global convergence of the gen etic algorithms was believed to apply the algorithms to deal with the diversity of population scientifically and recognize the contribution of the individuals t o the global convergence. Several kinds of the individuals, which have much to d o with the performance of the global convergence, have been proposed and combine d with the niche evolution sharing function. A genetic algorithm is suggested to improve the global convergence of the genetic algorithms and to obtain the glob al optimal solution. Simulation results show that the new algorithm has good per formance on the global convergence.

关 键 词：遗传算法 收敛性 多样性 遗传算子 全局最优 共享函数 

分 类 号：TP13[自动化与计算机技术—控制理论与控制工程]