摘要
为解决差分进化算法(DE)易陷入局部最优、收敛速度慢等问题,提出一种基于多样变异随机搜索的差分进化算法(DMSDE),并证明算法依概率收敛.DMSDE算法在保留DE算法变异操作的同时采用变异比例因子自适应调整策略提高种群进化效率;然后利用改进的交叉算子加快算法收敛速度;此外,构造了一个新颖的多样变异算子来增强算法局部搜索能力并确保种群多样性.通过8个常用标准测试函数上的实验表明,所提出的算法在收敛精度、稳定性、收敛速度方面都优于其他5种算法,具有较高的优化性能.
To solve the problems of falling into local optimum easily and converging to global optima slowly of differential evolution(DE),a new DE algorithm based on random search with diversity mutation is proposed,called DMSDE,and is proved that it converges in probability.The DMSDE retains the mutation operator of DE algorithm and uses mutation scaling factor self-adaptive strategy to improve the efficiency of population evolution.Then the improved crossover operator is used to speed up the convergence speed.In addition,a novel diversity mutation operator is constructed to enhance the local search ability and ensure the diversity of population.Experimental results of 8 unconstrained benchmark functions show that the proposed algorithm is outperform to the other 5 algorithms in terms of convergence accuracy,stability and convergence rate,and has higher optimization performance.
作者
曾辰子
余旌胡
邹桢苹
ZENG Chenzi;YU Jinghu;ZOU Zhenping(School of Science,Wuhan University of Technology, Wuhan 430070, Hubei, China;School of Economics and Management, Wuhan University, Wuhan 430072, Hubei, China)
出处
《武汉大学学报(理学版)》
CAS
CSCD
北大核心
2018年第3期211-216,共6页
Journal of Wuhan University:Natural Science Edition
基金
国家自然科学基金青年基金(11601400)资助项目
关键词
差分进化
交叉
多样变异
全局优化
differential evolution (DE)
crossover
diversity mutation
global optimization