A dynamic tri-population multi-objective evolutionary algorithm for constrained multi-objective optimization problems

Balancing between the convergence, feasibility, and diversity of the population is the key to solving constrained multi-objective optimization problems (CMOPs). However, the existing constrained multi-objective optimization evolutionary algorithm (CMOEAs) face challenges in converging to the constrained pareto front (CPF) with well-distributed feasible solutions. To address this issue, this paper proposes a dynamic tri-population multi-objective evolutionary algorithm, called TDPSCMO. In the initial phase, the first and second populations evolve dynamically to solve the original CMOP and the unconstrained Multi-Objective Optimization Problem (MOP). This dynamic offspring generation emphasizes convergence and feasibility. In the later stages, the allocation of computing resources to the first population is adjusted based on the variation in offspring numbers. Simultaneously, a third population is introduced to enhance population diversity by treating Constraint Violation degrees as an additional objective function, thus capturing more valuable information. The performance of TDPSCMO is further tested through 57 benchmark test problems and 21 real-world applications using several state-of-the-art algorithms. The results show the competitiveness of the proposed algorithm when addressing CMOPs.