•  
  •  
 

Journal of System Simulation

Abstract

Abstract: To address the energy management and privacy preservation problems faced by the coordinated optimization of distributed integrated energy systems, a distributed coordinated optimization strategy based on the multi-agent proximal policy optimization algorithm was proposed. An energy management model was established under the MDP framework; the electrical and thermal heterogeneous energy characteristics were considered; a multi-region two-layer interaction mechanism was constructed. Under the framework of centralized training and decentralized execution, homomorphic encryption was utilized to avoid privacy leakage during the coordination process, while accurately quantifying individual contributions to mitigate the problem of variance explosion in multi-agent policy evaluation. In the hourly scheduling of the system, the minimum daily cost was taken as the objective function to search for the optimal strategy. The simulation results show that the proposed algorithm can perform adaptive training based on a large amount of historical data and complete the derivation of an optimal strategy, which can simultaneously reduce the operating cost of each region while satisfying the engineering constraints.

First Page

476

Last Page

487

CLC

TP391

Recommended Citation

Tao Caixia, Chen Naikun, Gao Fengyang, et al. Distributed Optimization for Integrated Energy Based on Multi-agent Reinforcement Learning[J]. Journal of System Simulation, 2026, 38(2): 476-487.

Corresponding Author

Chen Naikun

DOI

10.16182/j.issn1004731x.joss.25-0728

Download

Share

COinS