Scholarlink

<p>This paper presents the design and development of an adaptive online event-triggered load frequency control (LFC) framework for islanded microgrids using machine learning techniques. The proposed approach addresses the stability challenges and performance limitations of conventional LFC strategies arising from the dynamic and stochastic characteristics of modern microgrids with high renewable penetration. A Neural-PI controller, pre-trained to capture complex system dynamics, is employed to enable intelligent event-triggered switching and real-time adjustment of control gains, thereby enhancing frequency regulation and overall system stability. The framework is implemented in Python using a Jupyter Notebook environment and evaluated on the IEEE 14-bus test system, where results demonstrate its ability to dynamically adapt event thresholds under varying operating conditions, leading to improved stability and control efficiency. Accordingly, the Neural-PI controller markedly outperforms the conventional PID controller, achieving a 58% reduction in peak frequency deviations, a 33% improvement in settling time, and a 95% reduction in control actions through event-triggered switching, with particularly strong performance during periods of renewable intermittency. Its adaptive capability enabled 2.5 times more effective inertia mode changes, reduced oscillations and overshoot, and contributed to a 19% increase in renewable hosting capacity alongside a 37% reduction in load-frequency control operational costs, making it especially suitable for islanded and low-inertia grids.</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

Online event-triggered switching load frequency control of islanded microgrids through machine learning

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月5日

查看全文
  • Osita U. Omeje,
  • Luciano Olukanni,
  • Daniel E. Okojie,
  • Candidus U. Eya,
  • Abiola B. Aina

摘要

This paper presents the design and development of an adaptive online event-triggered load frequency control (LFC) framework for islanded microgrids using machine learning techniques. The proposed approach addresses the stability challenges and performance limitations of conventional LFC strategies arising from the dynamic and stochastic characteristics of modern microgrids with high renewable penetration. A Neural-PI controller, pre-trained to capture complex system dynamics, is employed to enable intelligent event-triggered switching and real-time adjustment of control gains, thereby enhancing frequency regulation and overall system stability. The framework is implemented in Python using a Jupyter Notebook environment and evaluated on the IEEE 14-bus test system, where results demonstrate its ability to dynamically adapt event thresholds under varying operating conditions, leading to improved stability and control efficiency. Accordingly, the Neural-PI controller markedly outperforms the conventional PID controller, achieving a 58% reduction in peak frequency deviations, a 33% improvement in settling time, and a 95% reduction in control actions through event-triggered switching, with particularly strong performance during periods of renewable intermittency. Its adaptive capability enabled 2.5 times more effective inertia mode changes, reduced oscillations and overshoot, and contributed to a 19% increase in renewable hosting capacity alongside a 37% reduction in load-frequency control operational costs, making it especially suitable for islanded and low-inertia grids.