一种带变异算子的PSO算法

A PSO Algorithm with Mutation Operator

论文在基本PSO算法基础上,引入了遗传算法的变异算子。通过变异算子的控制函数,将PSO算法的训练过程分为前期和后期。在算法训练的前期,变异率取较大的值,以选择较多的粒子进行变异操作,目的是增强种群内部粒子的多样性,使得PSO算法能够在解空间的较大范围内进行搜索,以避免算法过早陷入局部最优解;在训练的后期,变异率取较小的值,以选择较少的粒子进行变异操作,目的是减弱种群内部粒子的多样性,使得PSO算法能够在解空间的较小范围内进行搜索,以提高算法的收敛精度。仿真研究表明,论文的算法具有较高的收敛精度,并且解决局部极小问题也相当成功。

In this paper, the mutation operator, which is used in Genetic Algorithm, is introduced into the basical PSOalgorithm. By mutation operator control function, the training process of the PSO algorithm is divided into early phase andlate phase. In the early phase of the algorithm train, the mutation rate is larger, so the more particles are chosen to take themutation operation, in order to enhance the diversity of the population’s particles, and the PSO algorithm can hunt in a largesolution space to avoid being trapped in the local optimal solution too early. In the late phase of the algorithm train, the mutationrate is smaller, so the less particles are chosen to take the mutation operation, in order to attenuate the diversity of thepopulation’s particles, therefore the PSO algorithm can hunt in a smaller solution space to improve the convergence accuracyof the algorithm. Simulation results show that, the convergence precision of this algorithm is higher, and solution of the localminimum problem is quite well.