Skip to main content

ORIGINAL RESEARCH article

Front. Energy Res.
Sec. Smart Grids
Volume 12 - 2024 | doi: 10.3389/fenrg.2024.1448046
This article is part of the Research Topic AI-based Energy Storage Systems View all 10 articles

Active power balance control of wind-photovoltaic-storage power system based on transfer learning double deep Q-network approach

Provisionally accepted
Jun Xiao Jun Xiao Wen Zhao Wen Zhao Wei Li Wei Li Yankai Zhao Yankai Zhao Yongzhi Li Yongzhi Li *Xudong Ma Xudong Ma Yuchao Liu Yuchao Liu
  • State Grid Shanxi Power Company, Lvliang, China

The final, formatted version of the article will be published soon.

    Aiming at the active power (AP) balance control problem of wind-photovoltaic-storage (WPS) power systems in the region with a high proportion of renewable energy (RE) units, this study proposes the transfer learning double deep Q-network (TLDDQN) method for controlling the energy storage device to balance the AP of WPS power systems. The TLDDQN method combines the advantages of transfer learning methods that can rapidly adapt to new environments to improve the double deep Qnetwork algorithm's training speed. In addition, this study proposes a method to combine the adaptive entropy mechanism with the DDQN algorithm and improve the adaptive entropy mechanism to enhancing the training capability of agents. Compared with the AP balance control method based on particle swarm optimization, the AP balance control method based on the proposed TLDDQN can more accurately balance the AP of the WPS power system to reduce the output of thermal power generators. The proposed TLDDQN algorithm is applied to a regional WPS power system for the experimental simulation of AP balance control. Experimental results demonstrate that compared to the DDQN algorithm, the TLDDQN algorithm proposed in this study trains agents faster and the AP balance control method of the TLDDQN-based WPS power system can more accurately control the storage device to reduce the output of thermal power generators, compared with the particle swarm optimization. Overall, the TLDDQN algorithm proposed in this study can provide some insights and theoretical references for research in related fields, especially those requiring decision making.

    Keywords: wind-photovoltaic-storage power system, Renewable Energy, active power balance control, Double Deep Q-Network, Transfer Learning

    Received: 12 Jun 2024; Accepted: 30 Aug 2024.

    Copyright: © 2024 Xiao, Zhao, Li, Zhao, Li, Ma and Liu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

    * Correspondence: Yongzhi Li, State Grid Shanxi Power Company, Lvliang, China

    Disclaimer: All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.