Observer-based deadbeat predictive speed controller for surface-mounted PM synchronous motor

초록

This paper investigates an observer-based deadbeat predictive speed controller (DBPSC) for a prototype surface-mounted permanent magnet synchronous motor (SPMSM) drive system. Unlike the conventional deadbeat controllers which are mainly used for deadbeat current control in the inner loop in a cascaded control structure, the proposed deadbeat controller is designed by incorporating a speed controller with a load torque disturbance observer. Specifically the proposed deadbeat speed control law consists of a deadbeat stabilizing feedback control term and a compensation control term. The stability of the proposed closed-loop system is assured based on the former term by forcing all poles of the dynamic error vector toward the origin within a short transient time, which guarantees a fast transient response of the control system by using the solution of an optimal control problem. The latter term is proposed based on the predicted q-axis current to compensate for the model nonlinearities and parameter uncertainties. Next, the deadbeat load torque observer (DBLTO) is proposed to provide a precisely estimated load torque with a high estimation rate for the predicted q-axis current in the latter term, which helps the proposed scheme in obtaining a small steady-state error (SSE), a low total harmonic distortion (THD), and robustness against the parameter variations. In addition, the proposed scheme possesses a simple structure, so it reduces the computational burden for digital implementation. The Lyapunov theory is utilized in the stability proof process of the proposed observer-based method. Finally, the comparative studies between the proposed DBPSC and a conventional deadbeat speed controller (DBSC) are performed under the sudden changes of speed and load torque through simulations on MATLAB/Simulink and experiments on a prototype SPMSM drive using a TI TMS32OF28335 DSP platform. (C) 2020 ISA. Published by Elsevier Ltd. All rights reserved.

키워드