Comparing and Ranking Metaheuristic Algorithms on Common Multi-Objective Test Problems

This paper presents a comparative study of five well-known metaheuristic algorithms — NSGA-II, MOPSO, ACO, WOA, and GWO — tested on standard multi-objective benchmark problems such as ZDT1, ZDT2, ZDT3, DTLZ1, and DTLZ2. The experiments were carried out using MATLAB R2021 with a consistent setup across all algorithms. Performance was evaluated using four important metrics: Inverted Generational Distance (IGD), Hyper volume (HV), Spread, and Convergence. Each algorithm was assigned a final score based on these metrics to prepare an overall ranking. The results show that NSGA-II performed the best in terms of both convergence and diversity, while MOPSO was a strong alternative choice. GWO gave stable but moderate results, whereas ACO and WOA showed weaker performance on complex problems. This study helps provide a baseline comparison and will serve as a reference point for future development of improved hybrid optimization techniques.