Dynamic-opposite learning enhanced meta-heuristic approach for solving multiple industrial optimization problems

Advanced industrial manufacturing involves a number of complex operation and decision NP-hard problems, such as industrial robot trajectory planning and flexible job shop scheduling, which calls for effective optimization approaches. In this work, we propose an enhanced version of the moth-flame optimization (MFO) algorithm, referred to as dynamic-opposite learning differential evolution MFO (DOLDEMFO), tailored for addressing continuous and integrated industrial optimization challenges. A dynamic opposite learning (DOL) strategy is embedded to asymmetrically refine the search region, enhancing both exploration and exploitation potential. Additionally, differential evolution (DE) is incorporated into the MFO framework, leveraging its structural simplicity, convergence characteristics, and robustness, to further refine search efficiency. To evaluate the proposed approach, extensive experiments on CEC series benchmarks and two industrial cases were performed. Comparative analysis with conventional metaheuristics verifies that DOLDEMFO consistently delivers competitive results and consistent performance across diverse industrial optimization tasks.