Co-Optimization of Resource Allocation and Operation in Large-Scale Energy Systems---Part II: Methods and Case Studies

This two-part paper proposes a parallel-distributed equivalent projection—Monte Carlo tree search (EP-MCTS) method for allocation and operation co-optimization in large-scale energy systems. Building upon the theoretical foundation established in Part I, Part II validates the proposed method through applications in both non-hierarchical energy systems (e.g., integrated energy systems, IESs) and hierarchical systems (e.g., transmission-distribution systems, TDSs), demonstrating its efficacy in overcoming computational and privacy barriers. To address the cyclic EP problem arising in hierarchical TDS applications, a novel cycle-resolving strategy extends EP’s applicability beyond tree-like EP problems to cyclic EP problems while preserving solution equivalence. Case studies demonstrate our method’s intrinsic privacy protection by compressing sensitive data into non-reconstructable numerical proxies, reducing shared parameters to merely 0.548% of benchmark requirements in IES applications. The proposed EP-MCTS also enables computational tractability for previously unsolvable large-scale problems (11,772-node and 150-scenario cases solved within 120 hours), accelerating initial feasible solutions by 8.09× with higher accuracy and achieving 34.49× faster convergence to satisfactory solutions through variable reduction (up to 98.78% fewer decision variables) and parallelized MCTS processing.