HIGH-PERFORMANCE CONTROL OF IPMSM BY USING MODIFIED-DTC ALGORITHM AND ANN BASED SPEED CONTROLLER

Abstract

In this thesis, we present a performance enhancement on direct torque control (DTC) for interior permanent magnet synchronous motor (IPMSM) drive. To solve some key drawbacks of the conventional DTC and space vector modulation (SVM) based DTC control strategies, the predictive controller with super-twist sliding mode (STSM) based torque controller is applied to modify SVM based DTC and backpropagation algorithm based neural network is adopted to tune the parameters of speed controller. First, a novel STSM torque controller based predictive calculator for generating reference voltage vector is analysed and designed in Chapter 2. This modified DTC strategy is named as STSM-DTC in this thesis. In previous research [27], the classical proportional-integral (PI) torque controller is applied and named as PI-DTC method. In this thesis, both STSM-DTC and PI-DTC strategies combine SVM scheme applied to IPMSM drive. Generally, the torque-ripple and flux-ripple reduction performance has been improved considerably by adopting SVM. The simulation results demonstrate that the proposed STSM-DTC method can enhance the torque control performance furtherly more compared to PI-DTC with smaller torque ripple. Moreover, it has the merits of fast dynamic response, robustness against load torque disturbance and speed-ripple reduction, as verified in Chapter 4. In addition, the whole proposed STSM-DTC control strategy, for IPMSM drives can obtain good performance of stator current with very low THD (total harmonic distortion) level. Next, a novel PI speed controller applying backpropagation based neural network is designed and proposed in Chapter 3. This speed controller is named as NN-PI speed controller in this thesis. The proposed method can overcome the disadvantages of conventional PI controller with high dependence on controlled model and time-consuming parameters-tuning work. Compared to classical PI speed controller, the simulation results prove the effectiveness of the proposed NN-PI speed controller with faster dynamic response, good speed tracking, much smaller speed overshoot and better robustness against load torque variation and load inertia variation, as presented in Chapter 4.