Article Transactions of the Institute of Measurement and Control 1–15 Ó The Author(s) 2017 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0142331217734631 journals.sagepub.com/home/tim Novel hybrid estimator based on model reference adaptive system and extended Kalman filter for speed-sensorless induction motor control Ridvan Demir and Murat Barut Abstract This paper presents a novel hybrid estimator consisting of an extended Kalman filter (EKF) and an active power-based model reference adaptive system (AP-MRAS) in order to solve simultaneous estimation problems of the variations in stator resistance (R s ) and rotor resistance (R r ) for speed-sensorless induction motor control. The EKF simultaneously estimates the stator stationary axis components (i sa and i sb ) of stator currents, the stator stationary axis components (u sa and u sb ) of stator fluxes, rotor angular velocity (v m ), load torque (t L ) and R r , while the AP-MRAS provides the online R s estima- tion to the EKF. Both the AP-MRAS, whose adaptation mechanism is developed with the help of the least mean squares method in this paper, and the EKF only utilize the measured stator voltages and currents. Performances of the proposed hybrid estimator in this paper are tested by challenging sce- narios generated in simulations and real-time experiments. The obtained results demonstrate the effectiveness of the introduced hybrid estimator, together with a 22:87% reduction in the processing time and size of the estimation algorithm in terms of previous studies performing the same estima- tions of the states and parameters. From this point of view, it is the first such study in the literature, to our knowledge. Keywords Induction motors, sensorless control, extended Kalman filter, model reference adaptive system, parameter estimation Introduction Speed-sensorless control of induction motors requires the esti- mation of v m and the stator stationary axis components of stator or rotor fluxes. However, temperature and frequency- based variations of R s and R r , and changes in t L , cause dete- riorations in these estimations, and thus in the tracking per- formances of controllers (Sun et al., 2016a,b) to be designed. Therefore, these parameters also need to be estimated over a wide speed range and included in estimation algorithms for obtaining robust estimations of v m and the fluxes. Nevertheless, there are some difficulties associated with the estimation of these states and parameters:  As v m approaches zero, measurement errors become larger and, more importantly, the electromagnetic induction from the stator to the rotor becomes wea- kened; thus, the measured stator currents, which are utilized in the design of a speed-sensorless observer or estimator, stop carrying the v m -information, espe- cially at very low and zero speeds (Holtz, 2002), which is known as the worst case for the state and parameter estimation of induction motors.  Another problem is the difficulty of performing simul- taneous estimation of both R s and R r in the speed-sensorless case (Faiz and Sharifian, 2001). If the number of measured states is increased, in other words, if rotor speed is measured together with the stator currents, then it is possible to estimate or com- pensate for changes in both resistances by the existing methods, as in Demir et al. (2017), using EKF, and Arunachalam et al. (2016), utilizing a model reference adaptive system scheme with inclusion of offline or online trained artificial neural networks. However, when only the measured stator currents are available, the simultaneous estimation of both R r and R s becomes challenging. Ozsoy et al. (2010) have tried to estimate i sa , i sb , u ra , u rb , v m , t L , R r and R s using an EKF algorithm for an eighth-order induction motor model in the speed-sensorless case, and declare that eighth-order EKF results are not as successful as seventh-order EKF results without R r or R s estimation Department of Electrical and Electronics Engineering, Nigde Omer Halisdemir University, Turkey Corresponding author: Murat Barut, Department of Electrical and Electronics Engineering, Nigde Omer Halisdemir University, 51200-Nigde, Turkey. Email: muratbarut27@yahoo.com, mbarut@ohu.edu.tr