Optimization and control of integrated energy systems based on multiagent deep reinforcement learning
This study presents a multi-agent deep reinforcement learning framework utilizing the MATD3 algorithm to enhance the control scheduling of energy systems. The proposed approach effectively integrates the benefits of energy power storage stations while minimizing the overall costs associated with energy production and distribution. By employing a collaborative multi-agent strategy, this method mitigates biases towards any individual stakeholder, ensuring a fair and efficient scheduling process. The results demonstrate the potential of MATD3 in achieving optimized performance in complex energy systems, contributing to smarter energy management strategies and fostering sustainable energy practices.
Authors
- Yingjun Ruan (ORCID: https://orcid.org/0000-0001-8275-5965)
- Meng Zhang (ORCID: https://orcid.org/0000-0002-1743-0686)
- Meng Hua (ORCID: https://orcid.org/0000-0002-3121-6344)
- Fanyue Qian (ORCID: https://orcid.org/0000-0002-9133-7850)
- Yuting Yao (ORCID: https://orcid.org/0000-0001-7281-2065)
- Hao Shen (ORCID: https://orcid.org/0000-0002-5003-4992)
- Qiang Zhang (ORCID: https://orcid.org/0000-0001-6765-930X)
- Tingting Xu (ORCID: https://orcid.org/0000-0002-9589-1349)
- Qingrong Liu (ORCID: https://orcid.org/0000-0003-1200-8029)
- Xiaoyu Liu (ORCID: https://orcid.org/0000-0003-2159-519X)
Institutions
- Tongji University (CN)
- Southern Company (United States) (US)
- Shanghai University of Electric Power (CN)
- Xinjiang New Energy Research Institute (China) (CN)
Publication Details
- Published
- 2025-06-30
- DOI
- https://doi.org/10.1117/12.3066781
- Primary Topic
- Power Systems and Renewable Energy
- Type
- article
- Field-Weighted Citation Impact
- 0.00