Date of Graduation


Document Type


Degree Name

Doctor of Philosophy in Engineering (PhD)

Degree Level



Electrical Engineering


Yue Zhao

Committee Member

Juan C. Balda

Second Committee Member

Shengfan Zhang

Third Committee Member

Jingxian Wu


Control, DTC, Motor Drive, SiC, Three-level Inverter, Virtual Space Vector (VSV), Virtual Voltage Vector


Direct torque control (DTC) is an extensively used control method for motor drives due to its unique advantages, e.g., the fast dynamic response and the robustness against motor parameters variations, uncertainties, and external disturbances. Using higher switching frequency is generally required by DTC to reduce the torque ripples and decrease stator current total harmonic distortion (THD), which however can lower the drive efficiency. Through the use of the emerging silicon carbide (SiC) devices, which have lower switching losses compared to their silicon counterparts, it is feasible to achieve high efficiency and low torque ripple simultaneously for DTC drives.

To overcome the above challenges, a SiC T-type neutral point clamped (NPC) inverter is studied in this work to significantly reduce the torque and flux ripples which also effectively reduce the stator current ripples, while retaining the fast-dynamic response as the conventional DTC. The unbalanced DC-link is an intrinsic issue of the T-type inverter, which may also lead to higher torque ripple. To address this issue, a novel DTC algorithm, which only utilizes the real voltage space vectors and the virtual space vectors (VSVs) that do not contribute to the neutral point current, is proposed to achieve inherent dc-link capacitor voltage balancing without using any DC-link voltage controls or additional DC-link capacitor voltages and/or neutral point current sensors.

Both dynamic performance and efficiency are critical for the interior permanent-magnet (IPM) motor drives for transportation applications. It is critical to determine the optimal reference stator flux linkage to improve the efficiency further of DTC drives and maintain the stability of the drive system, which usually obtained by tuning offline and storing in a look-up table or calculated online using machine models and parameters. In this work, the relationship between the stator flux linkage and the magnitude of stator current is analyzed mathematically. Then, based on this relationship, a perturb and observe (P&O) method is proposed to determine the optimal flux for the motor which does not need any prior knowledge of the machine parameters and offline tuning. However, due to the fixed amplitude of the injected signal the P&O algorithm suffers from large oscillations at the steady state conditions. To mitigate the drawback of the P&O method, an adaptive high frequency signal injection based extremum seeking control (ESC) algorithm is proposed to determine the optimal reference flux in real-time, leading to a maximum torque per ampere (MTPA) like approach for DTC drives. The stability analysis and key parameters selection for the proposed ESC algorithm are studied. The proposed method can effectively reduce the motor copper loss and at the same time eliminate the time consuming offline tuning effort. Furthermore, since the ESC is a model-free approach, it is robust against motor parameters variations, which is desirable for IPM motors.