Chaotic Opposition-Learning Based Particle Swarm Optimization

Particle swarm optimization (termed as PSO) is a simple and efficient meta-heuristic searching technique which has been extensively employed to solve various practical issues. However, PSO has some inherent shortcomings, such as easily to fall into local optima and occur imbalance between the global search and local search. To overcome these issues, this work puts forward a novel algorithm called chaotic opposition-learning based particle swarm optimization (abbreviated as COLPSO). First, a chaotic opposition-learning approach is used to produce initial particles for the purpose of improving the quality of swarm. Subsequently, a state-based inertia weight strategy is leveraged to well balance the local exploitation and global exploration capabilities of particles. In addition, a stochastic-interacting position update strategy is exploited to be expected to get excellent convergence performance by the influence of other particles. At length, COLPSO is simulated based on six classical test functions, and the conducted experiments indicate that COLPSO has significantly improved the demerits of standard PSO method and obtained better robustness and stability.