Scholarlink

<p>To address the voltage limit violation problems caused by the large-scale integration of renewable energy into distribution networks, a multi-agent cluster control strategy for voltage regulation in wind–photovoltaic–storage distribution networks is proposed. Firstly, based on a composite index comprising electrical distance, active power regulation capability and reactive power regulation capability, an improved clustering algorithm is employed to divide the modified IEEE 33-bus distribution network system which integrates wind turbine, photovoltaic and storage into clusters. Each cluster is then treated as an individual agent, with photovoltaic inverters, wind turbine, and distributed energy storage within the cluster serving as control devices. The Double Deep Q-Network algorithm within a multi-agent deep reinforcement learning framework is then utilized for cluster voltage control, with the objective of minimizing voltage deviation. Finally, the voltage control results of the modified IEEE 33-bus distribution network system are analyzed under different scenarios. Simulation results demonstrate that the proposed method improves the optimization performance by 5.89% and 16.13% compared to the DQN algorithm and the particle swarm optimization algorithm, respectively.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Multi-agent cluster control of voltage in wind–photovoltaic–storage integrated distribution networks

  • 期刊论文

  • 发表日期：

    2025年4月29日

查看全文
  • Enyu Jiang,
  • Chenfan Fang,
  • Yuqiang Wang,
  • Shunfu Lin,
  • Yang Mi,
  • Dongdong Li

摘要

To address the voltage limit violation problems caused by the large-scale integration of renewable energy into distribution networks, a multi-agent cluster control strategy for voltage regulation in wind–photovoltaic–storage distribution networks is proposed. Firstly, based on a composite index comprising electrical distance, active power regulation capability and reactive power regulation capability, an improved clustering algorithm is employed to divide the modified IEEE 33-bus distribution network system which integrates wind turbine, photovoltaic and storage into clusters. Each cluster is then treated as an individual agent, with photovoltaic inverters, wind turbine, and distributed energy storage within the cluster serving as control devices. The Double Deep Q-Network algorithm within a multi-agent deep reinforcement learning framework is then utilized for cluster voltage control, with the objective of minimizing voltage deviation. Finally, the voltage control results of the modified IEEE 33-bus distribution network system are analyzed under different scenarios. Simulation results demonstrate that the proposed method improves the optimization performance by 5.89% and 16.13% compared to the DQN algorithm and the particle swarm optimization algorithm, respectively.