Particle Swarm Optimization Based on Punctuated-equilibrium Model

Abstract

This paper presents a modified particle swarm optimization, named as punctuated-equilibrium particle swarm optimization (PEPSO). This method refers to punctuated-equilibrium Model (PEM) which is a pattern of group development in organizational behavior. PEM uses the long-term equilibrium phase and the short-term abrupt phase to solve the problem of group stagnation. The present work mathematically modelled this alternating process. The efficiency of the proposed PEPSO was evaluated using CEC 2014 benchmark functions. The experiments showed that PEPSO could solve premature convergence and had more good convergence accuracy than PSO on some test functions. Furthermore, it was also confirmed that the swarm was divided into different groups and the individuals of every group almost acted in unison. This provides a good explanation between PSO and organizational behavior from a new experimental perspective.