An Enhanced Active Disturbance Rejection Control Method for PMSM Speed Control Using Error-Corrected Cascaded ESO and Error-Transformed Adaptive RBF Neural Network

• Published in: IEEE transactions on power electronics pp. 1 - 14
• Main Authors: Kang, Yuxin, Deng, Yongting, Zhai, Chuanlong, Li, Wenjie, Cao, Haiyang, Liu, Xiufeng, Zhang, Zhimin, Xu, Wei
• Format: Journal Article
• Language: English
• Published: IEEE 2025
• Subjects: Active disturbance rejection control (ADRC); aperiodic and periodic disturbances; cascaded extended state observer (CESO); Couplings; Harmonic analysis; Mathematical models; Permanent magnet motors; permanent magnet synchronous motor (PMSM); Power harmonic filters; Radial basis function networks; radial basis function neural network (RBFNN); Regulation; Resonant frequency; Torque; Velocity control
• ISSN: 0885-8993; 1941-0107
• DOI: 10.1109/TPEL.2025.3618670

Under complex operating conditions, permanent magnet synchronous motor (PMSM) drive systems are subject to uncertain aperiodic and periodic disturbances, which degrade speed control performance. To address this issue, this paper proposes an enhanced active disturbance rejection control (ADRC) method. Firstly, the limitations of conventional linear ADRC and its existing improvements in disturbance rejection are analyzed. Based on this, an error corrected cascaded extended state observer (EC-CESO) is proposed. Compared with traditional CESO, EC-CESO provides greater flexibility in zero placement, which enhances its ability to suppress aperiodic disturbances while reducing sensitivity to high-frequency noise. Secondly, an error-transformed adaptive radial basis function neural network (ET-ARBFNN) is designed to compensate for the shortcomings of EC-CESO against periodic harmonics. Unlike internal model principle-based resonant control, ET-ARBFNN requires no prior harmonic order information while achieving a significantly enhanced dynamic response. Furthermore, the disturbance rejection performance and stability of the proposed enhanced ADRC combining EC-CESO and ET-ARBFNN are analyzed. Finally, the effectiveness and superiority of the proposed method are experimentally validated on a PMSM platform.