Constrained Multi-Objective Optimization with dMOPSO: An Adaptive Penalty Approach

In this paper, we propose a Multi-objective Particle Swarm Optimization (MOPSO) approach based on a decomposition framework with constraint-handling mechanisms to solve Constrained Multi-objective Optimization Problems (CMOPs). Specifically, we use a penalty function that is dynamically adjusted during the search process. As a result, the proposed approach extends a decomposition-based Multi-Objective Particle Swarm Optimizer (dMOPSO) to handle constraints. The resulting algorithm, namely Constrained dMOPSO (CdMOPSO), is evaluated using the well-known CF test problems taken from the CEC‘2009 test suite. Using convergence and spread indicators, we compare the solutions produced by our proposed approach with those produced by other multi-objective evolutionary algorithms. The results show that the proposed CdMOPSO is highly competitive and, in some cases, outperforms four state-of-the-art algorithms that were adopted in our comparative study.