Globally neural-adaptive simultaneous position and torque variable structure tracking control for permanent magnet synchronous motors. (January 2017)

• Record Type: Journal Article
• Authors: Hwang, Chih-Lyang; Fang, Wei-Li; Shih, Ching-Long
• Is Part Of: Journal of vibration and control. Volume 23:Number 1(2017)
• Page Start: 147
• Page End: 163
• Publication Date: 2017-01
• Subjects: Globally neural-adaptive control -- on-line neural network modeling -- PMSM -- position and torque control -- variable structure tracking control -- Lyapunov stability; Vibration -- Periodicals; Damping (Mechanics) -- Periodicals; 620.3
• DOI: 10.1177/1077546315575827
• Languages: English
• ISSNs: 1077-5463

Abstract

A globally neural-adaptive simultaneous position and torque variable structure tracking control (GNASPTVSTC) for the permanent magnet synchronous motors (PMSMs) subjected to excess uncertainties (e.g., time-varying system parameters, friction and load torques for different operating conditions, the avoidance of zero control gain, the similar convergence of position and torque) is developed. Based on Lyapunov stability criterion, the desired torque is first derived from the mechanical subsystem. A simultaneous position and torque variable structure tracking control (SPTVSTC) with the avoidance of zero control gain and the similar convergence of position and torque is first designed to obtain acceptable performance for the PMSM with mild uncertainties. To improve the PMSM in the presence of excess uncertainties, the integration of SPTVSTC and two on-line neural network models for uncertainties is employed to construct the proposed GNASPTVSTC. For approximating these non-autonomous uncertainties, they are assumed to be absolutely bounded for time variable and relatively bounded for other variables, respectively. It not only improves the steady state performance as compared with SPTVSTC, but also enhances the system stability in the face of excess uncertainties. The compared simulation results validate the global tracking ability outside of approximated set and the excess robustness for different amplitudes of uncertainties and saturated control input.