Particle swarm optimization with a leader and followers

Abstract

Referring to the flight mechanism of wild goose flock, we propose a novel version of Particle Swarm Optimization (PSO) with a leader and followers. It is referred to as Goose Team Optimization (GTO). The basic features of goose team flight such as goose role division, parallel principle, aggregate principle and separate principle are implemented in the recommended algorithm. In GTO, a team is formed by the particles with a leader and some followers. The role of the leader is to determine the search direction. The followers decide their flying modes according to their distances to the leader individually. Thus, a wide area can be explored and the particle collision can be really avoided. When GTO is applied to four benchmark examples of complex nonlinear functions, it has a better computation performance than the standard PSO.