Multiform differential evolution with feasibility-oriented knowledge transfer for interval-constrained scheduling of coal-mine integrated energy systems

The scheduling of coal mine integrated energy systems confronts great challenges, such as high dimensionality, strong coupling constraints, multi-objective optimization and source-load uncertainties. Existing optimization methods struggle to simultaneously address the aforementioned challenges. We here propose a novel two-phase multiform differential evolution algorithm with feasibility-oriented knowledge transfer for scheduling coal mine integrated energy systems with source-load uncertainties, characterized by interval values. Within the proposed multiform optimization framework, the intricate energy coupling relationships are decomposed to formulate two alternative tasks: the electric-thermal task with uncertainty and electriccooling task with certainty. In the electric-thermal task, the objectives, constraints, and some optimization variables related to derived energy are represented as intervals. A feasibility-oriented knowledge transfer mechanism by exchanging the coupled electrical variables of the alternative tasks is developed to facilitate the knowledge sharing for effectively performing multiform evolutionary optimization. The proposed two-phase differential evolution incorporates different neighborhood dual-mutation mechanism. In the first phase, a diversity-driven mutation is employed to evolve the two alternative tasks and possibly explore the intersecting feasible regions. In the second phase, a convergence-driven mutation is implemented to refine the original task within the feasible space and obtain a set of well-distributed feasible Pareto solutions. The efficacy of the proposed algorithm is demonstrated through a case study conducted at a coal mine in Shanxi by comparing with six state-of-the-art constrained multi-objective evolutionary algorithms and an interval constrained multi-objective evolutionary algorithm. The results demonstrate the superior performance of the propose method in feasibility, convergence, diversity, and uncertainty handling.

© 2026 IEEE. All rights reserved, including rights for text and data mining and training of artificial intelligence and similar technologies. Personal use is permitted, but republication/redistribution requires IEEE permission.