Immersion based observer for the switched reluctance motor ? Alejandra de la Guerra * Alejandro Guti´errez–Giles ** Gildas Besan¸con *** * Metropolitan Autonomous University (UAM), Av. San Pablo Xalpa 180, Reynosa Tamaulipas, Azcapotzalco, CDMX, Mexico (e-mail: ale delaguerra@comunidad.unam.mx). ** Center of Research and Advanced Studies of the National Polytechnic Institute (Cinvestav), Av. Instituto Polit´ecnico Nacional 2508, CDMX, Mexico (e-mail: alejandro.giles@cinvestav.mx) *** GIPSA-Lab, CNRS Univ. Grenoble Alpes, CNRS, Grenoble INP GIPSA-lab, 38000 Grenoble, France (e-mail: gildas.besancon@gipsa-lab.grenoble-inp.fr) Abstract: This article presents a high–gain Kalman–like observer for the switched reluctance motor that can reconstruct the angular position and velocity from the measured variables, i. e., stator currents and voltages. This is possible after the results obtained from an observability analysis, which in turn permits the application of an immersion to transform the motor model into a suitable form for the observer implementation. Numerical simulations are incorporated to explain and validate the observer design. The procedure includes an active phase detection stage and a current–based speed controller that defines the commutation required for this kind of motors. The proposed observer is capable of reconstructing the mechanical variables (rotor position and speed) by employing only electrical ones (currents and voltages). Keywords: switched reluctance motors, observers, observability, sensorless control 1. INTRODUCTION This paper is motivated by the great qualities of the Switched Reluctance Motor (SR motor), such as its simple and endurable construction, no need of permanent mag- nets, and its torque-speed characteristics. In particular, the SR motor has the unique property that it can still ope- rate during certain fault conditions (Saha and Choudhury, 2016), which makes this motor a reliable source of motion. Moreover, these properties make it a serious candidate for traction in Electric Vehicles (EV) and Hybrid Electric Vehicles (HEV), as documented in Rahman et al. (2000), Zeraoulia et al. (2006), and Ehsani et al. (2018). In the case of wind energy generation, there are cost problems associated to the employment of rare-earth elements to produce permanent magnets, such as neodymium (Nd). One of the solutions proposed by Jacobson and Delucchi (2011) is to avoid the use of this element by replacing the permanent magnet generators (PMG) with generators of similar performance and size such as the Switched Reluc- tance Generator (SRG). However, there are some disad- vantages for using this motor. For example, the control of this machine is more complicated than for other motors. In particular, it cannot be open loop controlled. Another drawback is that most industrial controllers present speed ripple and noise. Additionally, it is not as commercially available as other motors. ? Alejandra de la Guerra thanks CONACYT under scholarship CVU: 332351. Alejandro Guti´errez–Giles thanks CONACYT under scolarship CVU: 334785 Another control challenge is the sensorless control, i.e., the implementation of a speed control by using only electrical input (voltages) and output (currents) measurements, without mechanical sensors. The main motivation behind the sensorless control is of economic nature, since it can diminish the cost of the entire drive while, from a technical point of view, the elimination of mechanical sensors decreases the complexity and maintenance of the system. The sensorless problem for the switched reluctance motor has been extensively studied and partially solved using different methods, which can be classified as: 1) active- phase detection, 2) current gradient, 3) flux linkage re- construction, and 4) state observers. In general, the first two methods are signal-based and the last two ones are model-based methods. However, a review of the state of the art shows several problems associated with the de- sign of these estimators: most of the reported estimation techniques were made for a determined speed interval, e.g., low, medium or high speed. Also, few state observers have been reported, which includes the Luenberger-like observers proposed by Lumsdaine and Lang (1990) and Elmas and Zelaya-De La Parra (1996), and the sliding mode observer introduced by McCann et al. (2001) and lately modified by Brandstetter and Krna (2013). In more recent articles, such as Ouddah et al. (2015), Brandstetter et al. (2016), Xiao Wang et al. (2016), and Li et al. (2017), the authors work with very simplified models of the motor and the observers employed are based on the works already Preprints of the 21st IFAC World Congress (Virtual) Berlin, Germany, July 12-17, 2020 Copyright lies with the authors 6012