Towards a Penalty Annealing Approach in MOEA/D for Constrained Multi-Objective Optimization
Miguel A. Jiménez-Domínguez, Néstor A. García-Rojas, Saul Zapotecas Martinez, R. Díaz Hernández, Leopoldo Altamirano-Robles, Bilel Derbel · Proceedings of the Genetic and Evolutionary Computation Conference Companion · 2025
Over the years, MOEA/D has proven to be a highly effective and efficient algorithm for solving complex problems. Consequently, several studies have focused on extending MOEA/D to address constrained multi-objective optimization problems. In this article, we introduce a dynamic penalty function into the MOEA/D framework to handle these constraints. This penalty function is based on the method of annealing penalties, modified to exhibit linear behavior when interacting with feasible and infeasible solutions during the search process. The proposed approach enables MOEA/D to manage constraints effectively. To assess the performance of our method, we evaluated it on the well-known complex problems from the CEC'2009 benchmark set, while comparing it gainst several state-of-the-art algorithms. Our empirical results show that the proposed method delivers competitive solutions and, in some cases, outperforms the MOEAs included in the comparative study.