Balanced Global and Local Properties Multi-objective Particle Swarm Optimization (GL-MOPSO): A microgrid scheduling optimization algorithm

Many multi-objective optimization problems exist in electric power systems. Traditional multi-objective particle swarm optimization (MOPSO) adopts the crowding distance method to find the global best position; consequently, the algorithm presents satisfactory local property but poor global property. In this paper, a fuzzy similarity matrix method, which can enhance global search capability, is introduced into MOPSO to replace the crowding distance method. However, this method sacrifices the local property of the algorithm. To harmonize the global and local search capabilities, an improved MOPSO called Balanced Global and Local Properties MOPSO (GL-MOPSO) is proposed in this paper. In each generation of the process, the selection between the crowding distance method and the fuzzy similarity matrix method is guided by monitoring the convergence and diversity of the swarm. Furthermore, the strategy for updating the local best position is improved.