An effective particle swarm optimization algorithm with disturbance

Abstract

Particle swarm optimization (PSO) has been widely studied because of its advantages of fast convergence and few control parameters, therefore many improvements have been proposed. However, some variants of PSO inevitably increase the parameters needed to tune and computation cost and lose the advantages of faster convergence and easier implementation of the PSO algorithm. To overcome these problems, this paper proposes an improved PSO, which is called EPSO. Compared with the standard particle swarm optimization algorithm, the acceleration coefficient is removed and the learning factor is replaced to reduce the algorithm's dependency on its parameters. Noise-based disturbance is introduced to escape local optima. Afterward, a large number of benchmark functions were used to detect the performance of EPSO, and experimental results showed that it had higher accuracy and faster convergence speed.