A fuzzy logic constrained particle swarm optimization algorithm for industrial design problems

Most of the industrial design problems have non-linear constraints, high computational cost, non-convex, complicated, and large number of solution spaces. This poses a challenge for algorithms to effectively handle constraints and improve solution accuracy. To address these challenges, a fuzzy logic particle swarm optimization algorithm incorporating a correlation-based constraint handling method (FILPSO-SCA$\varepsilon$) is proposed. In FILPSO-SCA$\varepsilon$, an adaptive $\varepsilon$ constraint handling method with correlation analysis is introduced to dynamically adjust the utilization of constraints and the objective function information. The particle swarm optimization algorithm is employed as the searcher, and to augment its search capability, a set of fuzzy logic rules integrating individual feasibility is designed. These rules dynamically generate parameters in learning strategies by considering fitness and the distance between individuals. To mitigate premature convergence problems, we introduce an individual learning mechanism utilizing stagnation detection. 28 constrained optimization problems and 2 industrial design problems are utilized for comparison with 16 well-known constrained evolutionary algorithms. The proposed algorithm ranks first among the 16 comparative algorithms, with a success rate of 100% in solving industrial design problems.