An Adaptive Particle Swarm Optimization for Global Optimization

The paper suggests a new modified approach to improve the performance of particle swarm optimization (PSO). Inspired by the intelligent behaviors of the natural biotic populations, the modified PSO is based on an adaptive strategy, the particle should stop the inertia movement to enhance the learning from its experiences and its neighbors when it is found to be in wrong searching direction, and stop the learning process to fly straight when it is found to be the nearest to the destination in the swarm. Furthermore, four different forms of the adaptive PSO model are presented. Comparison results with the standard PSO on the examination of some unconstrained and constrained global optimization functions show the effectiveness of the new modified approach.