Adaptive robust speed control based on recurrent elman neural network for sensorless PMSM servo drives

• Published in: Neurocomputing (Amsterdam) Vol. 227; pp. 131 - 141
• Main Authors: Jon, Ryongho, Wang, Zhanshan, Luo, Chaomin, Jong, Myongguk
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.03.2017
• Subjects: Adaptive; Learning rate; Permanent magnet synchronous motor (PMSM); Reconstruction error; Recurrent Elman neural network (RENN); Robust; Recurrent Elman neural network (RENN); Adaptive; Reconstruction error; Robust; Learning rate; Permanent magnet synchronous motor (PMSM)
• ISSN: 0925-2312; 1872-8286
• DOI: 10.1016/j.neucom.2016.09.095

In this paper, an adaptive robust control scheme based on recurrent Elman neural network (RENN) is proposed to achieve high-performance speed tracking despite of the existence of system uncertainties for the sensorless permanent magnet synchronous motor (PMSM) servo drive. Firstly, the dynamics of sensorless PMSM operated with the system uncertainties are described in details. Secondly, an adaptive RENN speed controller (ARENNSC) composed of an RENN controller and a compensated controller is developed to achieve the adaptive robust speed control of PMSM drive. The RENN controller is designed to imitate an ideal speed control signal for sensorless PMSM, and the compensated controller is designed to compensate an error between ideal control signal and actual RENN signal, including an RENN reconstruction error. The adaptive laws are derived based on Lyapunov theorem to ensure the stability of ARENNSC. Then, a calculation method of ideal learning rate is also presented to improve the adaptive performance of ARENNSC. The simulation results demonstrate the feasibility, robustness and good dynamic performance of the proposed adaptive RENN speed control scheme.