Abstract
Multi-objective optimization problems (MOPs) demand algorithms that effectively balance convergence, diversity, and computational efficiency. To address this challenge, a novel Multi-Objective Human Evolutionary Optimization Algorithm (MOHEOA) is proposed, inspired by the dynamics of human societal evolution. MOHEOA structures the search process into two adaptive phases: human exploration and human development, integrating a fixed-size dynamic archive to maintain and utilize non-dominated Pareto solutions. The algorithm begins with a logistic chaos mapping for population initialization, ensuring robust diversity. During the development phase, individuals are classified into leaders, explorers, followers, and losers, each employing specialized strategies tailored for multi-objective search. A roulette-wheel selection mechanism dynamically selects leaders from the archive, optimizing the trade-off between exploration and exploitation. To validate MOHEOA’s performance, extensive experiments on twenty-three benchmark test functions and four real-world engineering design problems are conducted. Comparative evaluations against state-of-the-art multi-objective algorithms demonstrate that MOHEOA consistently outperforms competitors in convergence speed, solution diversity, and Pareto optimality. The algorithm’s robustness and adaptability make it a compelling choice for complex optimization tasks. For reproducibility and further research, the MATLAB implementation of MOHEOA is publicly available at: https://github.com/swatzash/MOHEOA.