Dynamic Repulsion-Attraction Particle Swarm Optimization Based on Adaptive Adjustment

In recent years, the traditional Particle Swarm Optimization (PSO) algorithm has drawn increasing attention due to its deficiencies such as reduced population diversity and insufficient information sharing among particles. To overcome these limitations and enhance the algorithm's search capability, this paper proposes a new PSO variant called Dynamic Repulsion-Attraction PSO based on Adaptive Adjustment (DRA-PSO). The DRA-PSO introduces an inertia weight that varies with the number of iterations and learning factors that adjust according to the inertia weight to improve the adaptability of the algorithm. Furthermore, a repulsion-attraction strategy based on population diversity is designed to balance global and local search capabilities. Finally, a terminal elimination strategy is incorporated to establish a self-purification mechanism for population quality and sustain the activation of the exploration capability. To evaluate the performance of DRA-PSO, comparative experiments were conducted with four other improved PSO algorithms using nine benchmark test functions. The results demonstrate that DRA-PSO not only maintains population diversity but also achieves faster convergence and higher global optimization accuracy.