A temporal-segmentation-and-fusion-based large-scale constrained multi-objective evolutionary algorithm for coal mine integrated energy systems dispatch

Abstract

Under the dual-carbon policy, the optimization of coal mine integrated energy systems (CMIESs) has garnered increasing attention from researchers. However, the characteristics of multiple optimization objectives, strong coupling constraints, and high-dimensional decision variables pose significant challenges for optimization methods. Existing constrained multi-objective optimization algorithms struggle to effectively solve such problems with multiple complex characteristics, and they are prone to getting stuck in local optima. To address this issue, this paper proposes a temporal-segmentation-and-fusion-based large-scale constrained multi-objective evolutionary algorithm, which divides the evolutionary process into two stages. In the subspace optimization stage, the large-scale space is divided into multiple smaller subspaces according to the problem’s temporal divisibility characteristics, then these subspaces are progressively integrated and optimized by using a portion of the computational resources. Once all subspaces are integrated, the original space optimization stage is activated, then the remaining computational resources are employed to search the original large-scale space. In addition, a differential evolution mutation strategy based on dynamic neighborhoods is studied, which effectively balances global exploration and local exploitation, guiding the population toward promising regions in the search space. Finally, the experimental results with several advanced evolutionary algorithms on an actual CMIES dispatch case demonstrate the efficiency of the proposed algorithm.