Stator Temperature Estimation of Direct-Torque-Controlled Induction Machines via Active Flux or Torque Injection

Abstract

This paper proposes two thermal monitoring methods for induction machines with direct torque control (DTC). The stator resistance $R_{s}$ is used as a direct indicator of average winding temperature and is estimated via a dc current offset in the stator winding. In a DTC drive system, the major challenge is how to excite the dc current offset via the only existing flux and torque control loops. The quantitative relationship between the desired dc current offset and the corresponding changes in flux linkage and torque is derived, and this shows that the dc current offset can be achieved either by superimposing flux linkage bias command $[\matrix{{\bf \Delta \Psi} _{{\bm s\alpha},{\rm ref}}\cr {\bf \Delta\Psi}_{{\bm s\beta},{\rm ref}}}]$ in the flux control loop (method 1) or torque ripple command ${\bf \Delta }T_{em,{\rm ref}} $ in the torque control loop (method 2), both shown in Fig. 1 . Simulation results confirm the analytical analysis of the two proposed methods; while the experimental data prove that both methods achieve accurate temperature estimation under various operating conditions. The proposed simple and efficient signal-injection-based temperature estimation technique is particularly advantageous since it eliminates the need for embedded temperature sensors, requires no hardware change to conventional DTC drive systems, and has a minimal impact on the induction machine's normal operation.