Abstract - Model Reference Adaptive System (MRAS) based techniques are one of the best methods to estimate the

rotor speed due to its performance and straightforward stability approach. These techniques use two different models (the reference model and the adjustable model) which have made the speed estimation a reliable scheme especially when the motor parameters are poorly known or having large variations. The scheme uses the error vector from the comparison of both models as the feedback for speed estimation. Depending on the type of tuning signal driving the adaptation mechanism, there could be a number of schemes available such as rotor flux based MRAS, back e.m.f based MRAS, reactive power based MRAS and artificial neural network based MRAS. All these schemes have their own trends and tradeoffs. In this paper, the performance of the rotor flux based MRAS (RF MRAS) and back e.m.f based MRAS (BEMF MRAS) for estimating the rotor speed was studied. Both schemes use the stator equation and rotor equation as the reference model and the adjustable model respectively. The output error from both models is tuned using a PI controller yielding the estimated rotor speed. The dynamic response of the RF-MRAS and BEMF-MRAS sensorless speed estimation is examined in order to evaluate the performance of each scheme.
-

Index terms : BEMF-MRAS, MRAS, parameter variations, RFMRAS, sensorless speed, tracking capability. The study of speed sensorless control of the IM has undergone through maturing years when new techniques came into introduction to improve the previous techniques. The motivation is to find one method that can cater the entire problem related to speed sensorless IM. Among them, MRAS based techniques have been proven to be one of the best methods being proposed by the researchers due to its good high performance ability and straight-forward stability approach [1]. The method was first proposed in [2] followed by [3] which consists of a reference model (RM), an adjustable model (AM) and an adaptation mechanism. RM is independent of the rotor speed whereas AM requires the rotor speed information. Through Landau's idea of comparing the outputs of RM and of AM, the error between the two models can be minimized using the adaptation mechanism [4], as depicted in Fig. 1 Various structures of MRAS can be realized depending on the quantities that form the error vector. They are rotor-flux based MRAS, back e.m.f based MRAS, reactive power based MRAS and artificial intelligence based MRAS [5]. Each of these schemes has their own trends and tradeoffs.

I.

INTRODUCTION

Reference model

y ey

Adjustable model

Adaptation mechanism Tuning signal