基于熵模型的动态粒子群优化算法

受多种群并行寻优机制的启发,提出了一种基于熵模型的动态粒子群优化算法(entropy dynamic multiPSO,EDM-PSO)用于处理动态优化问题.将解空间划分为多个子空间,在每个子空间中利用熵模型增加种群多样性,多种群并行搜索,利用多点环境检测机制检测环境变化.对动态多峰benchmark优化问题进行了数值实验,并与其他几种动态优化算法进行了比较,结果表明:EDM-PSO算法对于处理动态优化问题具有优势.

Entropy-based Dynamic Particle Swarm Optimization Algorithm

Inspired by the multi-population parallel optimization mechanism, this paper proposes an Entropy-based Dynamic Multi-population Particle Swarm Optimization ( EDM-PSO) algorithm which can be utilized to deal with dynamic optimization problems. The solution space was divided into multiple sub-spaces, in which the entropy models were utilized in each sub-space to increase the diversity of populations. Additionally, the multi-population parallel searching mechanism and multi-point detection mechanism were also implemented to seek the optimal solution and to detect ambient environmental changes respectively. Finally, a comparison between EDM-PSO and several other dynamical optimization algorithms in terms of the errors ( standard deviation ) when addressing a moving peaks function benchmark problem was made, resulting in that the EDM-PSO algorithm can be more beneficial to solving dynamic problems.