Sensorless field orientation of an induction motor drive using a time-varying observer

The study discusses the problem of state and rotor flux position estimation for the induction motor (IM) drive and presents a sensorless time-varying observer. The method treats the induction motor as a time-varying plant and uses non-linear techniques to develop a state observer based on the motor´s model in the stationary reference frame. Initially, the observer is developed assuming that the speed of the motor is available through measurement - the resulting observer is asymptotically stable and some of its gains are time varying. To obtain a sensorless observer, the speed measurement is eliminated by replacing the speed signal with a speed estimate (assumed to be inaccurate but reasonably close to the real speed). The observer with improper speed is analysed - this is not asymptotically stable; however, the estimated fluxes are in phase with the real fluxes. This property allows estimation of the rotor flux angle of the IM. This is because the accuracy of the direct field orientation method that is based on the tan^-1 function only depends on the phase of the flux components. The study shows that the observer yields fluxes with magnitude higher than real when the speed signal fed in the observer is smaller than real (and vice versa). The design of the observer under improper speed is analysed theoretically; it is then supported by simulations and is validated by experimental tests.