摘要
针对差分进化算法种群的多样性和收敛速度的矛盾问题,提出一种增强全局搜索能力的差分进化算法(简称MP-WDE)。首先,在种群初始化阶段将种群划分为多个子种群,分别采用不同的变异策略来提高种群的多样性。同时,利用动态递增的交叉概率因子来提高进化过程种群多样性。最后,通过引入策略加权机制,将变异策略"DE/current-to-best/2"改进为加权变异策略"DE/current-to-best/2 or DE/current-to-rbest/2",以提高算法的收敛速度。通过与其他经典的改进差分进化算法在CEC2005的对比实验,该算法具有良好的寻优效果。无论在解的精度还是收敛速度方面都有所提高,因此MPWDE算法可以执行。
For the contradiction between the population diversity and the convergence rate exists in the differential evolution algorithm, proposes differential evolution with Enhance Global Search Capability（MPWDE）. Firstly, the population is divided into multiple subpopulations with different mutation strategies, which improves the diversity of the population at the initial stage of the population. Meanwhile, the dynamic crossover probability factor is adopted to improve the population diversity in evolutionary process. Finally, by introducing weighted strategies mechanism, the mutation strategy DE/current-to-pbest/2 is replaced by new mutation strategy DE/current-to-pbest/2 or DE/currentto-rbest/2, in order to improve the convergence speed of the algorithm. This paper algorithm makes a great effect in optimization, which compares with the other classical algorithms. Both the solution accuracy and the convergence speed are improved, so the MPWDE algorithm can be implemented.
作者
范宇凌
FAN Yu-ling(College of Engineering,Huaqiao University,Quanzhou 362021)
出处
《现代计算机》
2018年第10期18-23,共6页
Modern Computer
基金
华侨大学研究生研究创新能力培养计划资助项目(No.1611422002)
关键词
差分进化
多种群
动态递增
加权变异
Differential Evolution
Multiple Subpopulations
Dynamic Increment
Weighted Strategy
