Enhanced methodology for injection-based real-time parameter estimation to improve back-EMF self-sensing in induction machine deadbeat-direct torque and flux control drives

Abstract

Back-EMF self-sensing is commonly used in induction machine (IM) drive systems. Deadbeat-direct torque and flux control (DB-DTFC) has also shown to be a highly effective method for IM control. At low speeds, accurate machine parameters are important for both DB-DTFC and back-EMF self-sensing performance. In this paper, an injection-based real-time parameter estimation method in IM DB-DTFC drives for improving the back-EMF self-sensing performance is proposed. The methodology for choosing an appropriate injected signal frequency is introduced. By injecting the signal, the positive and negative sequence components can be obtained through signal processing procedures. Then, machine parameters can be derived from these components. To obtain more accurate estimates, the rotor bar skin effect is considered. With encoder velocity feedback, the proposed DB-DTFC injection method induces no additional torque ripple, compared to traditional d-axis pulsating voltage vector injection used in IFOC. The positive and negative sequence components can be obtained with an electrical position estimation block. However, in self-sensing mode, the injected flux linkage induces an erroneous speed ripple, which causes a real speed ripple. Two approaches are presented to eliminate the speed ripple. Finally, with injection-based parameter estimation in IM DB-DTFC, the improvements of back-EMF self-sensing performance are experimentally evaluated.