基于滑动平均极值的粒子群优化算法

Particle swarm optimization based on moving-weighted-average best position

针对标准粒子群优化算法(SPSO)易陷入局部最优,进化后期收敛速度慢的缺点,提出一种基于滑动平均极值的粒子群优化算法(MWAPSO)。改进了标准粒子群算法中的速度更新方程,使得粒子在进化过程中追随个体极值、全局极值和滑动平均极值。将该算法应用于4个典型的测试函数,实验结果表明,与标准粒子群算法相比,该算法在运行初期具有更强的探索能力,能够有效地避免粒子群体陷入早熟收敛。有更好的收敛性和更快的收敛速度。

Concerning the disadvantage of the standard particle swarm optimization （SPSO） that is easily trapped in the local optimization and the convergence speed is slow in the evolution later, a particle swarm optimization based on moving-weighted-average best position （MWAPSO） is proposed. The change of the velocity equation in SPSO made particles tracking the local best position and the global best position and the moving-weighted-average best position in the process of the evolution. The experimental results on four classical bench- mark functions illustrate that, compared with the SPSO, the MWAPSO has better exploitation ability at the beginning and it can keep particles from getting into the premature convergence more effectively, and achieves better and faster convergence.