A strategy cooperative algorithm based on state-awareness for large-scale multi-objective optimization

Large-scale multi-objective optimization problems (LSMOPs) usually involve hundreds to thousands of decision variables. When dealing with unconstrained 2-3-objective LSMOPs, multi-objective evolutionary algorithms (MOEAs) are likely to get trapped in local optima, making it difficult to ensure the diversity and convergence of solutions within limited computational resources. To tackle this challenge, we propose a strategy-cooperative algorithm based on state-awareness for large-scale multi-objective optimization, abbreviated as LMOEA-SC. In LMOEA-SC, we have designed a state-aware mechanism that can monitor the evolutionary state of the population in real-time. Based on the real-time information, LMOEA-SC flexibly switches and collaborates between the proposed learning strategy based on diversity protection competitive swarm optimization (DPCSO) and the escape strategy based on global exploration sampling (GES), thus effectively coping with different evolutionary states and challenges. The obtained statistical results, with a 73% improvement, clearly show that compared with six state-of-the-art MOEAs, LMOEA-SC has significant competitiveness in numerous large-scale multi-objective test instances with up to 2,000 decision variables.