Scholarlink

<p>The increasing penetration of renewable energy sources and sensitive electronic loads has intensified power quality challenges in modern transmission systems. Among FACTS devices, the Unified Power Flow Controller (UPFC) offers comprehensive control over voltage and power flow; however, its performance is highly dependent on the employed control strategy. Conventional PI-based and static AI-based controllers often exhibit limited adaptability under dynamic operating conditions. This paper proposes a learning-assisted Model Reference Adaptive Control (MRAC) framework integrated with a Long Short-Term Memory (LSTM) network for UPFC-based power quality enhancement. Unlike conventional AI controllers, the LSTM is employed to dynamically adapt reference signals, while the MRAC loop ensures stable tracking and robustness. The proposed controller is evaluated using detailed MATLAB/Simulink simulations under varying load and disturbance conditions. Results demonstrate improved voltage regulation, reduced power oscillations, and significantly lower total harmonic distortion compared with conventional control approaches. The offline training and online inference structure ensures computational feasibility for real-time implementation, making the proposed method suitable for practical transmission system applications.</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

Learning-Assisted Model Reference Adaptive Control of Unified Power Flow Controller for Enhanced Power Quality

  • 期刊论文

  • 发表日期：

    2026年4月13日

查看全文
  • Kunta Srikanth,
  • Srinivas Kunta,
  • Rahul Wilson Kotla,
  • Nimmakanti Anil

摘要

The increasing penetration of renewable energy sources and sensitive electronic loads has intensified power quality challenges in modern transmission systems. Among FACTS devices, the Unified Power Flow Controller (UPFC) offers comprehensive control over voltage and power flow; however, its performance is highly dependent on the employed control strategy. Conventional PI-based and static AI-based controllers often exhibit limited adaptability under dynamic operating conditions. This paper proposes a learning-assisted Model Reference Adaptive Control (MRAC) framework integrated with a Long Short-Term Memory (LSTM) network for UPFC-based power quality enhancement. Unlike conventional AI controllers, the LSTM is employed to dynamically adapt reference signals, while the MRAC loop ensures stable tracking and robustness. The proposed controller is evaluated using detailed MATLAB/Simulink simulations under varying load and disturbance conditions. Results demonstrate improved voltage regulation, reduced power oscillations, and significantly lower total harmonic distortion compared with conventional control approaches. The offline training and online inference structure ensures computational feasibility for real-time implementation, making the proposed method suitable for practical transmission system applications.