A Particle Swarm Optimization Algorithm with Distributed Adaptively Weighted Delays

Abstract

针对粒子群算法(PSO)容易陷入局部最优值、收敛精度低等缺陷，提出一种新的带分布式自适应时延的粒子群算法(PSO-DW)。改进的算法主要在RODDPSO算法的基础上考虑时延的时变性和种群的进化状态，以平衡算法的全局搜索和局部搜索能力，降低早熟收敛的可能性，提高算法的收敛速度和精度。主要思想：1) 在引入了分布式时延的速度更新公式中，每个时延项配以自适应权重，2) 引入通过当前状态和概率转移矩阵预测下一进化状态的预测机制，3) 分布式时延的强度因子由预测状态所确定。在九个基准函数上与四个算法作对比的实验结果表明，改进后的算法在寻优质量、稳定性、收敛速度等方面更具优越性。 A new particle swarm optimization algorithm (PSO) with distributed adaptively weighted delays (PSO-DW) has been proposed to overcome the defects of the PSO algorithm, such as falling into local optimal value, low convergence accuracy. Based on the RODDPSO algorithm, the improved algorithm further considers the time-varying delays and the evolutionary states of the population, so that it can balance the global search and local search ability of the algorithm, reduce the possibility of premature convergence, and improve the convergence speed and accuracy of the algorithm. The main ideas are: 1) each delay is equipped with adaptive weight in the velocity update formula; 2) prediction mechanism of the next evolutionary state has been introduced by the current state and probability transfer matrix; 3) intensity factor of the distributed delay is determined by the prediction state. The experimental results show that the improved algorithm has more advantages in optimizing quality, stability and convergence speed by comparing with four algorithms on nine benchmark functions.