Effect of Particle Initialization on the Performance of Particle Swarm Niching Algorithms

The vector-based PSO (VBPSO) [3,4,5] was developed to locate multiple solutions to multi-modal optimization problems. Three versions of the VBPSO were published, and shown to be very efficient in locating mutliple optima. This is despite the fact that the VBPSO algorithms initialize particles using standard pseudo random number generators. The main objective of this article is to show that the perfomance of the VBPSO algorithms can be improved by initializing particles using Sobol sequences [1,2].