A Novel Exponential Dynamic Inertia Weight for Particle Swarm Optimization

The traditional particle swarm optimization (PSO) algorithm suffers from shortcomings like easily falling into local optimum and inadequate sharing of information among particles, To ad-dress these limitations and enhance the search capacity of the particle swarm algorithm, we present a novel particle swarm optimization algorithm known as Exponential Dynamic Inertia Weight for Particle Swarm Optimization (ExDyPSO) in this paper. ExDyPSO is composed of two parts: firstly, by introducing dynamic inertia weight based on exponential distributions and acceleration factors that vary with the number of iterations, harmonizes the global and local search capabilities. Secondly, a stochastic particle-based jump-out strategy is proposed to surmount the case of particles falling into stagnation during the search process, thus effectively addressing the issue that PSO is prone to falling into local optimum. To assess the performance of ExDyPSO, we carried out experiments on eight benchmark functions and compared its performance against four alternative PSO variations. The experimental findings demonstrate that ExDyPSO achieves quicker and more accurate convergence towards the global optimal solution, all the while sustaining population diversity.