一种基于能量熵的快速遗传算法研究  

A Fast Genetic Algorithm Based on Energy-Entropy

作　　者：张毅[1] 杨秀霞[1] 

机构地区：[1]海军航空工程学院自动控制系,山东烟台264001

出　　处：《系统工程理论与实践》2005年第2期123-128,共6页Systems Engineering-Theory & Practice

基　　金：国家"十五"预研重大项目

摘　　要：在分析标准遗传算法的优越性与存在不足的基础上,提出了对遗传算法的改进方法.将能量熵的选择加入到遗传算法的退火选择中,以充分地探索解空间,保持种群的多样性.将伪梯度搜索应用于对个体的邻域搜索,利用当前种群的有效信息及系统信息,提高寻优速度.对典型的TSP问题及一实际电力网络故障恢复的仿真研究表明,改进算法全局优化性能优于启发式遗传算法及标准、退火遗传算法,同时使收敛速度有了较大的提高.Based on analysis of the properties of genetic algorithm(GA), a novel improved method is provided. The energy-entropy selection is used in GA annealing selection, which can explore the solution space sufficiently and keep the population diversity. Pseudo-gradient is used to neighborhood search, which can exploit the effective information in the current population and the system information. The results of simulation tests on typical traveling salesman problem (TSP) and a real power network fault restoration show that the improved algorithm is better than GA, heuristic GA and simulated annealing GA in global optimization, which increases the convergence speed significantly.

关 键 词：遗传算法 能量熵 伪梯度 旅行商问题 电力网络故障恢复 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]