High-performance speed estimation of induction machines based on adaptive filtering

This paper presents a novel algorithm for sensorless detection of the rotor speed applied to induction machines, which relies on rotor slot harmonics (RSHs). The estimator structure is proposed, for extracting and tracking the RSH found in the stator line current, which offers improved results compared with previously published methods. The proposed method reduces delays, output jitter, and steady-state error. A stability analysis is provided and used in the design procedure to ensure system reliability. The capability of using the estimated signal as speed feedback in a sensorless drive is tested and confirmed. The system employs a bandpass filter with a centre frequency that is adapted according to the estimated speed, a high bandwidth phase-locked loop, and an observer. Tracking limits and parameter design are explained analytically; the method is verified in simulations and confirmed experimentally. Experimental verification is performed on a standard three-phase induction motor, using a three-phase pulse-width modulation inverter.