A Large-Scale Constrained Multi-Objective Evolutionary Algorithm with Promising Region Detection and Diversity Enhancement

Most existing constrained multi-objective evolutionary algorithms (CMOEAs) suffer from slow convergence and may even fail to find feasible solutions when dealing with problems that involve large-scale decision spaces and complex constraints. To this end, this paper introduces a promising region-guided CMOEA. In the proposed algorithm, a promising region detection and diversity enhancement strategy is designed to avoid entrapment in local optima. Specifically, this strategy first utilizes well-converged feasible non-dominated solutions among all solutions examined so far to detect the promising region where the constrained Pareto front may exist. Then, three non-dominated sorting procedures based on extended Pareto dominance, extended reverse Pareto dominance, and classical Pareto dominance, are sequentially executed to guide the population to evenly search the promising region and approach the constrained Pareto front from diverse search directions. In addition, an accelerated evolutionary search strategy is devised to improve reproduction quality in large-scale search spaces. Comprehensive experiments conducted on four benchmark test suites and thirty coal mine integrated energy system dispatch problems demonstrate the superiority of the proposed algorithm over seven state-of-the-art algorithms.