Inertia weight control strategies: Particle roaming behavior

Abstract

The performance of particle swarm optimization (PSO) algorithms have shown to be very sensitive to the main control parameters. Due to this sensitivity, various approaches have been developed to dynamically adjust the value of these control parameters in an attempt to remove the necessity of prior control parameter tuning. A large number of these adaptive approaches are different ways in which the inertia weight value can be adjusted during the search process. Recent studies have shown that these adaptive approaches are not very efficient. This paper supplements previous studies of the different inertia weight control strategies, focusing on the roaming behavior of particles under the different control strategies. It is shown that a number of these inertia weight control strategies exhibit excessive roaming behavior, resulting in wasted search effort.