Considerations of stator resistance online-tuning method for MRAS-based speed sensorless induction motor drive

]Model reference adaptive system (MRAS) based estimator has become the dominant speed estimation approach for sensorless induction motor drives. However, stator resistance thermo variation always deteriorates speed estimation stability and accuracy at low speeds, especially in the generating-mode operation, which is a common challenge to MRAS-based speed estimators. To improve the drive's robustness, generally the stator resistance should be estimated simultaneously at low speeds. In this paper, the two estimators, conventionally regarded as a complex fourth-order MIMO system, are decoupled to be two simple second-order SISO systems based on the two-time-scale approach. Simple but general design principles rather than complicated numerical solutions are presented to stabilize the estimators. A novel error function with dynamic convergence is proposed to estimate stator resistance in acceleration, deceleration and impact-load situations. Stability of the proposed estimators is verified by extensive experiments. Analysis method of this paper can be extended to other MRAS schemes.