An MRAS-type estimator for the speed, flux magnitude and rotor flux angle of the induction motor using sliding mode

Abstract

The paper discusses the problem of estimating the speed, the flux magnitude and the rotor flux angle of the induction motor (IM) and presents an estimation method based on two Sliding Mode Observers (SMOs) and the Model Reference Adaptive System (MRAS) technique. The method is based on implementation of two SMOs that both yield the magnitude of the rotor flux; one observer is the reference model, the other is the adjustable model. The MRAS method is used to adapt the speed signal which is an input into both SMOs. The reference model is designed using the equations of the IM in the rotating reference frame. It is shown that its estimated flux magnitude is insensitive to the input speed. The adjustable model uses the IM equations in the stationary reference frame. Its output fluxes have magnitudes inverse proportional with the input speed; however, their phases are always accurate (this allows estimation of the flux angle). Using MRAS, the speed is corrected such that the flux magnitudes coming out of the two models match. Based on the structure developed, the paper also a speed estimation method. The simulations validate the theoretical development.

Rotor flux estimation method for cage induction generators used in wind power applications

“…The importance of the rotor flux estimation is well recognized in the literature and different rotor flux estimation methods are reported [6], [7], [10]- [15]. They differ with respect to the accuracy, robustness, and sensitivity against model parameter variations [6].…”

A New Formulation of Reactive-Power-Based Model Reference Adaptive System for Sensorless Induction Motor Drive

“…Several speed sensorless schemes have been proposed in literature [3]- [31]. These techniques are classified based on full-order and reduced-order observers [5]- [9], Kalman filter [20]- [24], sliding modes [27]- [31], artificial neural networks (ANN) [17], [18], predictive control [25], [26], saliency and signal injection [3], [4], and model reference adaptive systems [9]- [19], [25]- [27].…”

“…Sliding mode methods [27]- [31] have the advantage of finite time convergence unlike exponential convergence of observer based methods. They can also reject matched disturbances but have the serious problem of chattering.…”

“…Also, the injected signals may interfere with the normal operation of the drive causing oscillations in speed and torque. Model reference adaptive system (MRAS) based methods [9]- [19], [25]- [27] for rotor speed estimation has earned popularity due to its simplicity, performance and ease in implementation. Schauder proposed one such technique using rotor flux as the functional candidate [10].…”

“…In another attempt [17], an artificial neural network is used in the MRAS to make it stable in the regenerative mode but as already stated, use of neural network requires training and there is no standard method of choosing the neural networks. Apart from these, many authors tried to combine MRAS with other methods stated above like sliding modes [27], ANN [17], [18], predictive control [25], [26] etc., but only resulted in more complicated methods especially when the comparison is made in terms of computational burden.…”