基于变异策略的粒子群算法

在研究粒子群算法的特点之后,将变异因子融入到粒子群算法之中,提出了一种带有变异策略的粒子群算法(MPSO).该变异因子可以提高算法对解空间的开发能力,从而降低了粒子群算法陷入局部最优的可能性.实验结果表明,经过对4个无约束问题、1个高维线性约束问题以及1个实际应用问题的测试,带有变异策略的粒子群算法可以成功地解决高维无约束问题和带有线性约束的高维问题.实验结果也表明,MPSO算法具有很强的收敛性和稳定性,是一种很有前途的优化算法.

Particle Swarm Optimization Based on Mutation Strategy

After studying the features of PSO (particle swarm optimization), the mutation factor was integrated into the algorithm to form its new version, i.e., the mutational particle swarm optimization (MPSO) in which a mutation strategy was incorporated. With the mutation factor the algorithm can improve the developability of solution space so as to decrease the possibility that the PSO falls into local optima. The tested results of four well-known unconstrained benchmark optimization problems, a constrained problem and an actual application problem revealed that MPSO can successfully tackle both the high-dimensional unconstrained problems and the high-dimensional problems with linear constraints. Moreover, MPSO as a promising optimization algorithm has strong convergence and high stability.