Consumer-Centric Decentralized Federated Reinforcement Learning for Energy Scheduling in Community Integrated Energy System
IEEE Transactions on Consumer Electronics · 2026-05 · 72 (2) · 5164-5176
DOI: 10.1109/tce.2025.3643134 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Consumer Electronics, 2026-05. Verfasst von Hao Yang und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 35 Literaturangaben. Gefördert von: National Natural Science Foundation of China, Zhejiang Provincial Natural Science Foundation of China.
Abstract (Originalsprache)
This paper presents a consumer-centric decentralized federated reinforcement learning (DFRL)-based energy scheduling strategy for community integrated energy systems (CIES). In this strategy, each region assigns one agent that can independently manage the CIES by training its own local network. Different from both the multi-agent deep deterministic policy gradient algorithm, which relies on a shared experience pool, and the traditional federated learning reinforcement learning method, where a central server (CS) is required for parameter aggregation, our proposed DFRL algorithm allows the local agent only to share and update network parameters in a round-robin sequence without the need for a CS. This leads to simple algorithm design and easy implementation, and privacy is also preserved, as no agent has access to the local data of other agents, while still enabling cooperative optimization among the CIESs. In addition, as no CS is required, the single-point failure risk can be avoided. Experimental evaluations on a 4-region 5-equipment CIES show that RR-DFRL achieves a total operating cost of ${\$}201,250.39$ , which is within 8.46% of the Gurobi optimum ( ${\$}185,556.65)$ , and requires 0.636 s online computing time versus 6.37 s for Gurobi. The total communication overhead in training is 42.50 MiB, lower than CFRL (85.00 MiB) and far below MADDPG (2404.74 MiB), all while maintaining data privacy.
Förderung
- National Natural Science Foundation of China
- Zhejiang Provincial Natural Science Foundation of China
Zitieren
APA
Yang, H., Guo, F., Zhu, X., Wang, L., & Wu, Z. (2026). Consumer-Centric Decentralized Federated Reinforcement Learning for Energy Scheduling in Community Integrated Energy System. IEEE Transactions on Consumer Electronics, 72(2), 5164-5176. https://doi.org/10.1109/tce.2025.3643134
Vancouver
Yang H, Guo F, Zhu X, Wang L, Wu Z. Consumer-Centric Decentralized Federated Reinforcement Learning for Energy Scheduling in Community Integrated Energy System. IEEE Transactions on Consumer Electronics. 2026;72(2):5164-5176. doi:10.1109/tce.2025.3643134
BibTeX
@article{Yang2026,
author = {Yang, Hao and Guo, Fanghong and Zhu, Xinwen and Wang, Lei and Wu, Zhengguang},
title = {Consumer-Centric Decentralized Federated Reinforcement Learning for Energy Scheduling in Community Integrated Energy System},
journal = {IEEE Transactions on Consumer Electronics},
year = {2026},
volume = {72},
number = {2},
pages = {5164--5176},
doi = {10.1109/tce.2025.3643134}
}
EUVIMED