Stability Analysis and Parameter Selection of a Particle Swarm Optimizer in a Dynamic Environment

The paper addresses the issues of parameter selection of a particle swarm optimization algorithm by a thorough stability analysis of the swarm dynamics. The effectiveness of the work lies in considering the dynamic behavior of the local and the global best particle positions, which usually are treated as constant in the existing analysis. The behavior of an individual particle here is modeled as a closed loop control system, represented by a signal flow graph. The stability analysis of the closed loop system is undertaken using Jury's test and root locus technique of classical control theory, and the result obtained from the analysis offers a more stringent condition on parameter selection in comparison to the existing results on stability analysis. Computer simulation of particle swarm algorithm further confirms better performance of the algorithm, when parameters are selected following the results of stability analysis.