An Improved Multi-Objective Particle Swarm Optimization Algorithm Based on Competitive Mechanism and Fitness Ranking

Most MOPSO algorithms only use a single search strategy and update method to guide particle iteration, which can lead to difficulties in solving some difficult MOPs. Especially when solving problems with many objective irregular Pareto front, there may also be an imbalance between convergence and diversity. To tackle this problem, this paper proposes a multi strategy particle swarm optimization algorithm (CF-MOPSO), which uses competition mechanism and fitness ranking method. Using a competitive mechanism to select particles, CF-MOPSO introduces an enhanced fitness allocation strategy to keep the solution widely and evenly distributed. And in the selection strategy, the recently proposed θ- dominance are used to sort solutions, which increases the pressure to choose the optimal solution while also improving convergence speed. The proposed CF-MOPSO algorithm was applied to several scalable benchmark multi-objective problems for testing, and compared with five recently proposed MOPSO algorithms. The experimental results plots that the CF-MOPSO algorithm achieves significant improvements in optimization quality and convergence speed while balancing convergence and diversity. This result demonstrates its significant advantage in obtaining Pareto optimal sets with good performance.