HyMOPOG: A novel hybrid nature-inspired algorithm to approximate Pareto fronts for multi-objective problems

This study proposes a novel approach using a hybridized evolutionary intelligence paradigm to generate Pareto solutions for multi-objective problems. The solver, called HyMOPOG, is proposed to solve multi-objective (MO) problems based on the hybridization (Hy) of the particle-based optimization (PO) theorem and natural selection in the genetic (G) algorithm. To assess the performance of HyMOPOG, a set of benchmark test functions (UF1–10, CF1–10, and DTLZ1–8), and three engineering optimization problems were used based on five evaluation metrics (HV, GD, IGD, distance, and spread) and Friedman and Wilcoxon tests. NSGAII, MOPSO, PESAII, and a recently proposed algorithm named CCMO were used to evaluate the performance of the proposed algorithm. The HyMOPOG benefits from the operators of multi-objective particle swarm optimization (MOPSO) and non-dominated sorting genetic algorithm-2 (NSGAII) and an effective updating population scheme, which uses global search ability of MOPSO and local search ability of NSGAII. Therefore, it eliminates the weaknesses of each algorithm, has a balance between local and global search, and has an increased chance of escaping from being trapped in local optima. Results indicate the efficiency and applicability of the proposed approach.