A Distributed Architecture for Particle Swarm Optimization Meta-heuristic Methods Based on Multi-agent Systems

This chapter delves into the innovative fusion of Particle Swarm Optimization (PSO), Multi-Agent Systems (MAS), and parallelism to tackle intricate optimization problems. The study addresses the limitations of traditional PSO by introducing autonomous agents that enhance decision-making and convergence. By leveraging parallel computing, the proposed model accelerates the optimization process and improves solution robustness. The chapter provides a comprehensive background, detailing previous works that laid the groundwork for this integrated approach. It introduces a novel Parallel Particle Swarm Optimization (PPSO) model that employs geographical neighborhoods to enhance exploration and prevent premature convergence. The experimental section evaluates the PPSO model using a diverse set of benchmark functions, demonstrating superior performance in terms of success rate and computation time compared to sequential PSO. Furthermore, the chapter explores the integration of MAS with PSO, highlighting the benefits of autonomy, learning, and cooperative behaviors in achieving improved optimization outcomes. The proposed Multi-Agent Parallel PSO (MAPPSO) architecture is detailed, showcasing the interaction between Environment Agents and Particle Agents. Experimental results confirm that MAPPSO excels in solution quality and reduces computation time, making it a promising approach for addressing large-scale optimization challenges. The chapter concludes with a discussion on future work, emphasizing the potential of MAPPSO in real-world applications and the need for further performance analyses.