Particle swarm optimization

Abstract

Particle swarm optimization (PSO), which is based on a social-psychological model of social influence and social learning, is one of the modern heuristic algorithms for optimization. It is applied to optimize non-linear as well as multi-modal functions by simulating social behaviours of fish schooling or bird flocking to search their food. In this project, the performance of PSO and chaotic particle swarm optimization was investigated on eleven benchmark test functions with their own distinctive nature. We replaced the use of random number to simulate social behaviours of particles. This was done by introducing five chaotic sequences with different characteristics as well as piecewise linear chaotic map. Those chaotic sequences were implemented in different parts of the algorithm to compare performance unambiguously. The data and observations were summarized to prove how CPSO improves the optimal value.