一种动态改变惯性权重的自适应粒子群算法

Adaptive particle swarm optimization algorithm with dynamically changing inertia weight

针对标准粒子群算法在进化过程中种群多样性降低而早熟的问题,提出一种动态改变惯性权重的自适应粒子群算法.采用种群中平均粒子相似程度作为种群多样性的测度,并用于平衡算法的全局探索和局部开发.基于对惯性权重随种群多样性测度变化的动态分析,建立了惯性权重随种群多样性测度的变化关系,并将其引入该算法中.最后对6个经典测试函数进行仿真,结果表明该算法在平均最优值和成功率上都有所提高,特别是对多峰函数效果更明显.

To overcome the premature caused by standard particle swarm optimization （PSO） algorithm searching for the large lost in population diversity, an adaptive PSO with dynamically changing inertia weight is proposed. The average of similarity of particles in the population as the measure of population diversity is introduced into proposed algorithm to balance the trade-off between exploration and exploitation. A function relationship between inertia weight and the measure of population diversity is established by analyzing the dynamically relationship between them, which is embedded into the algorithm. The simulation results show that the algorithm has better probability of finding global optimum and mean best value, especially for multimodal function.