Scholarlink

<p>This review presents an integrated framework for system identification of active magnetic bearing (AMB) systems, focusing on high-speed, frictionless rotor applications. We provide a comprehensive overview of the theoretical foundations, methodological evolution, engineering implementation, and validation strategies. The review traces the progression from classical imbalance-response and frequency/time-domain methods to more advanced nonlinear and data-driven approaches, with a particular emphasis on physics-informed neural networks (PINNs) and digital-twin-based methods that bridge physical interpretability with data-driven adaptability. Beyond algorithmic approaches, we highlight crucial engineering factors—such as real-time hardware constraints, excitation design, sensor and actuator nonidealities, data preprocessing, and multilevel validation—that influence the reliability of system identification. Finally, we propose a forward-looking roadmap focusing on lightweight modeling, explainable AI integration, and predictive health management, aiming to guide both researchers and practitioners in advancing adaptive, robust, and digital-twin-enabled AMB systems.</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

A review of system identification for active magnetic bearings: a closed-loop perspective from theory to engineering validation

  • 期刊论文

  • 发表日期：

    2026年1月22日

查看全文
  • Zhe Wang,
  • Han Wu,
  • Xinyan Song,
  • Yuwan Zou,
  • Xingwei Sa,
  • Zhenjun Zhao

摘要

This review presents an integrated framework for system identification of active magnetic bearing (AMB) systems, focusing on high-speed, frictionless rotor applications. We provide a comprehensive overview of the theoretical foundations, methodological evolution, engineering implementation, and validation strategies. The review traces the progression from classical imbalance-response and frequency/time-domain methods to more advanced nonlinear and data-driven approaches, with a particular emphasis on physics-informed neural networks (PINNs) and digital-twin-based methods that bridge physical interpretability with data-driven adaptability. Beyond algorithmic approaches, we highlight crucial engineering factors—such as real-time hardware constraints, excitation design, sensor and actuator nonidealities, data preprocessing, and multilevel validation—that influence the reliability of system identification. Finally, we propose a forward-looking roadmap focusing on lightweight modeling, explainable AI integration, and predictive health management, aiming to guide both researchers and practitioners in advancing adaptive, robust, and digital-twin-enabled AMB systems.