Many-Objective Multi-Verse Optimizer (MaOMVO): A Novel Algorithm for Solving Complex Many-Objective Engineering Problems

Research has demonstrated the effectiveness of metaheuristic algorithms in addressing optimization challenges involving multi-objectives. These algorithms excel in generating a range of solutions that not only approach the Pareto front but are also evenly distributed. Multi-objective algorithms are effective for problems with two or three objectives. However, their efficiency diminishes in scenarios with many objectives, as the diversity selection and convergence pressure become less effective. Balancing convergence and diversity in multi-objective optimization pose a significant challenge. In response, this study introduces a novel many-objective multi-verse optimizer algorithm named MaOMVO for addressing many-objective problems. It integrates reference point and niche preserve to improve convergence and diversity and employs an innovative information feedback mechanism technique for population renewal. The superiority of the MaOMVO algorithm is evident in tests with MaF problems having 5, 8 and 15 objectives, as well as five real-world problems (RWMaOP1—RWMaOP5). It outperforms four leading algorithms Many-objective moth flame optimization, many-objective particle swarm optimization, non-dominated sorting genetic algorithm-III and reference vector guided evolutionary algorithm in terms of generational distance by 70%, inverted generational distance by 52%, spacing by 46.66%, spread by 55.55%, hypervolume by 52% and running time by 52% with concave, convex and mixed pareto fronts, confirming its robustness in diverse optimization scenarios.