A Multiobjective Electromagnetism-Like Algorithm with Improved Local Search

The Multiobjective Electromagnetism-like Mechanism (MOEM) is a relatively new technique for solving continuous multiobjective optimization problems. In this work, an enhanced MOEM algorithm (EMOEM) with a modified local search phase is presented. This algorithm derives from the modification of some key components of MOEM including a novel local search strategy, which are relevant for improving its performance. To assess the new EMOEM algorithm, a comparison with an original MOEM algorithm and other three multiobjective optimization state-of-the-art approaches, OMOPSO (a multiobjective particle swarm optimization algorithm), MOSADE (a multiobjective differential evolution algorithm) and NSGA-II (a multiobjective evolutionary algorithm), is presented. Our aim is to assess the ability of these algorithms to solve continuous problems including benchmark problems and an inventory control problem. Experiments show that EMOEM performs better in terms of convergence and diversity when compared with the original MOEM algorithm. EMOEM is also competitive in comparison with the other state-of-art algorithms.