Scholarlink

<p>The sustainable energy achievement is one of the crucial goals that must be accomplished. To attain this, the exploitation of renewable energy, such as wind is essential. Doubly fed induction generators (DFIGs) give great efficiency at different speeds when used in wind turbine systems (WTSs). The uncertainty in dynamic working conditions such as unexpected wind speeds may result in erratic behavior of the machine in terms of output power, rotor speed, and introduction of oscillations in the power system resulting in unwanted voltage profile. Traditional controllers are used to rectify these conditions but they lack in output efficacy when subjected to dynamic behavior of the system. Hybrid adaptive neuro-fuzzy inference system with proportional integral (ANFIS-PI) controllers are also limited by lengthy training and high computational demands. To overcome these limitations, this paper introduces a novel support vector regression (SVR)-based controller for DFIG-based WTS. The proposed engineering solution leverages the computational efficiency and adaptability of SVR to simultaneously regulate the rotor-side converter (RSC) and grid-side converter (GSC). Through comprehensive simulations, the SVR controller is shown to eliminate rotor speed overshoot, reduce settling time by approximately 70%, and minimize DC-link voltage ripple by nearly 95.45% compared to PI controllers. Furthermore, it achieves a nearly 60% faster settling time than ANFIS-PI controllers while operating with significantly lower complexity and minimal training data. The proposed SVR framework offers a lightweight, accurate, and real-time feasible control solution that balances computational simplicity, adaptability, and dynamic performance, thus strengthening the system’s overall stability and efficiency.</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

Machine Learning Approach Using SVR for Optimized Control of DFIG Wind Turbine Systems

  • 期刊论文

  • 发表日期：

    2025年10月15日

查看全文
  • Farhat Nasim,
  • Shahida Khatoon,
  • Ibraheem,
  • Mohammad Shahid

摘要

The sustainable energy achievement is one of the crucial goals that must be accomplished. To attain this, the exploitation of renewable energy, such as wind is essential. Doubly fed induction generators (DFIGs) give great efficiency at different speeds when used in wind turbine systems (WTSs). The uncertainty in dynamic working conditions such as unexpected wind speeds may result in erratic behavior of the machine in terms of output power, rotor speed, and introduction of oscillations in the power system resulting in unwanted voltage profile. Traditional controllers are used to rectify these conditions but they lack in output efficacy when subjected to dynamic behavior of the system. Hybrid adaptive neuro-fuzzy inference system with proportional integral (ANFIS-PI) controllers are also limited by lengthy training and high computational demands. To overcome these limitations, this paper introduces a novel support vector regression (SVR)-based controller for DFIG-based WTS. The proposed engineering solution leverages the computational efficiency and adaptability of SVR to simultaneously regulate the rotor-side converter (RSC) and grid-side converter (GSC). Through comprehensive simulations, the SVR controller is shown to eliminate rotor speed overshoot, reduce settling time by approximately 70%, and minimize DC-link voltage ripple by nearly 95.45% compared to PI controllers. Furthermore, it achieves a nearly 60% faster settling time than ANFIS-PI controllers while operating with significantly lower complexity and minimal training data. The proposed SVR framework offers a lightweight, accurate, and real-time feasible control solution that balances computational simplicity, adaptability, and dynamic performance, thus strengthening the system’s overall stability and efficiency.