Contribution to the Improvement of Sensorless DTC-SVM for Three-Level NPC Inverter-fed Induction Motor Drive

Abstract

This paper deals with the high performance of multi-level direct torque control (DTC) for induction machines without speed and stator flux sensors. The estimation is performed using a sliding mode observer, which is known for its robustness in high and low-speed operations, the control is based on a backstopping speed controller. This control technique was introduced years ago to circumvent the problems of sensitivity to parametric variations, it presents a high dynamic but their major problems are the variable switching frequency, the size and complexity of the switching tables, and undulations of the torque. The proposed approach is to replace switching tables with constant switching frequency control using three-level spatial vector modulation (SVM). Theoretical elements and simulation results are presented and discussed. As a result, the flux and torque ripple of three-level DTC-SVM control is greatly reduced compared to the flux and torque ripple of DTC-classic control. The advantages of the training system have been validated by the simulation results.