Initializing PSO with probability distributions and low-discrepancy sequences: The comparative results

This paper investigates the effect of initiating the population with various probability distributions and low discrepancy sequences on the behavior of Particle Swarm Optimization (PSO). The probability distributions: Gaussian, Exponential, Beta and Gamma distribution and the low discrepancy sequences: Van der Corput and Sobol are considered in this study. Based on these probability distributions, six algorithms namely BTPSO, GAPSO, GPSO, EPSO, VCPSO and SOPSO are presented. The proposed algorithms are tested on standard benchmark problems and the results are compared with the basic version of PSO which follows the uniform distribution for initializing the swarm. The simulation results show that a significant improvement can be made in the performance of PSO, by simply changing the distribution of random numbers to other than uniform distribution as the proposed algorithms outperform the basic version by a noticeable percentage.