基于随机惯性权重的简化粒子群优化算法

Simplified particle swarm optimization algorithm based on stochastic inertia weight

针对标准粒子群优化算法易出现早熟收敛、搜索速度慢及寻优精度低等缺陷,提出一种基于随机惯性权重的简化粒子群优化算法。算法采用去除速度项的粒子群简化结构,通过随机分布的方式获取惯性权重提高新算法的局部搜索和全局搜索能力,并且学习因子采用异步变化的策略来改善粒子的学习能力。考虑到个体之间的相互影响关系,每个粒子的个体极值用所有粒子个体极值的平均值代替。通过几个典型测试函数仿真及F-检验结果表明,提出的算法在搜索速度、收敛精度、鲁棒性方面较已有改进算法有了显著提高,并且具有摆脱陷入局部最优解的能力。

Abstract： This paper proposed a new particle swarm optimization （PSO） algorithm based on two aspects of improvement in standard PSO to avoid the problems about premature convergence and low precision. It applied the iteration formula of PSO based on the simple PSO which removes the velocity parameter. As two important factors in PSO, it determined inertia weight using stochastic variable, and learning factor was using asynchronous change strategy, to enhance the balance of global and lo- cal search of algorithm. Taking into account the interactive relationship among all particles, it replaced the personal best value of each by the mean value of them. Through several typical test functions simulation and F-test results show that the proposed algorithm not only has great advantages of convergence property over standard PSO and some other modified PSO algorithms, but also effectively avoids being trapped in local optimal solution.