多变异策略的差分演化算法研究  

Research on differential evolution algorithm with multiple mutation strategies

作　　者：黄华 刘罡[2] 李会珍 HUANG Hua;LIU Gang;LI Huizhen(School of Information Engineering of Wuhan College,Wuhan 430212,China;Computer School of Hubei University of Technology,Wuhan 430086,China;School of Mathematical and Physical Sciences of Wuhan Textile University,Wuhan 430200,china)

机构地区：[1]武汉学院信息工程学院,湖北武汉430212 [2]湖北工业大学计算机学院,湖北武汉430068 [3]武汉纺织大学数理科学学院,湖北武汉430200

出　　处：《长江信息通信》2023年第4期51-54,共4页Changjiang Information & Communications

基　　金：武汉学院校级科研项目(X2022030);湖北省教育厅科学技术研究计划指导性项目(B2021365);湖北省自然科学基金指导性项目(2022CFC065);湖北省高校优秀中青年科技创新团队(T2022055)。

摘　　要：针对经典差分演化算法易出现全局搜索能力低、收敛速度慢的特点,提出了一种多变异策略的差分演化算法。首先,该算法在经典差分演化算法的基础上,引入混沌映射机制进行初始化,可以增强算法的全局搜索能力。然后,选择多种变异策略进行优化:(1)利用Hilbert变异矩阵有选择方向的进行变异,减少变异的随机性;(2)利用正交表来选择种群变异的分量,增强全局搜索能力;(3)利用K-means聚类算法,对种群进行聚类选择,从而加快算法的收敛速度。对多个经典测试函数进行试验研究,研究结果表明:该方法具有快速的收敛能力、良好的稳定性,其优化性能显著提升。Aiming at the characteristics of the classical differential evolution algorithm,such as low global search ability and slow convergence speed,a differential evolution algorithm with multiple mutation strategies is proposed.First,based on the standard differential evolution algorithm,the algorithm introduces chaotic mapping mechanism,which can enhance the global search ability of the algorithm;Then,we select a variety of mutation strategies to optimization,(1)use the Hilbert mutation matrix to select the direction of mutation to reduce the randomness of mutation,(2)use the orthogonal table to select the components of population mutation and enhance the global search ability,(3)use the K-means clustering algorithm to cluster the population,so as to speed up the convergence speed of the algorithm.The experimental results of several classical test functions show that the method has fast convergence ability,good stability and its optimization performance is significantly improved.

关 键 词：差分演化算法 混沌映射 正交设计 Hilbert变异矩阵 K-MEANS聚类 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]