62 PRZEGLĄD ELEKTROTECHNICZNY, ISSN 0033-2097, R. 88 NR 12a/2012 Mateusz DYBKOWSKI, Grzegorz TARCHAŁA, Teresa ORŁOWSKA-KOWALSKA 1) 1) Wroclaw University of Technology, Institute of Electrical Machines, Drives and Measurement Experimental analysis of the sensorless traction drive system with DTC-SVM algorithm and MRAS CC estimator Abstract. In the paper the sensorless DTC-SVM induction motor drive system for tram application is described. For the rotor speed, electromagnetic torque and stator flux vector reconstruction the MRAS CC estimator was applied. This estimator is less sensitive to the motor parameter variation than other observers and is stable in the wide range of the speed reference changes as well as it is simple in practical implementation. Experimental tests, typical for traction applications, were performed for 50kW induction motor drive in the whole speed and torque ranges. Streszczenie. W pracy przedstawiono analizę bezczujnikowego układu wektorowego sterowania silnikiem indukcyjnym z wykorzystaniem metody DTC-SVM dla trakcji miejskiej. Do estymacji strumienia stojana, momentu elektromagnetycznego i prędkości kątowej wykorzystano estymator MRAS CC , charakteryzujący się zwiększoną odpornością na zmiany parametrów schematu zastępczego silnika indukcyjnego oraz stabilnością w szerokim zakresie zmian prędkości kątowej wirnika i prostotą realizacji sprzętowej. Wykonano obszerne badania eksperymentalne na stanowisku laboratoryjnym z silnikiem indukcyjnym o mocy 50kW w całym zakresie zmian prędkości i momentu obciążenia, typowe dla napędów trakcyjnych. (Analiza bezczujnikowego układu wektorowego sterowania silnikiem indukcyjnym z wykorzystaniem metody DTC-SVM) Keywords: induction motor, sensorless drive, speed estimation, MRAS estimator, traction drive Słowa kluczowe: silnik indukcyjny, napęd bezczujnikowy, estymacja prędkości, estymator MRAS, napęd trakcyjny Introduction Recently the AC motors are more and more popular in the industrial drives, because of their robust construction and relatively low manufacturing cost (especially in a case of induction motor drives), practically maintenance-free, and at least well-matured vector control methods, which assure as good dynamical performance as in the case of DC motor drives [1]-[4]. So, these control ideas have been introduced in traction drives also, like drives of trams, trolleybuses, or metro train sets, connected with higher efficiency and reliability with simultaneous reduction of the drives’ weight and size [6]-[8]. The induction motor (IM) or permanent magnet synchronous motor (PMSM) drives are used nowadays in such applications [7], [8]. Control methods, used in the traction drives systems, mainly Field Oriented Control (FOC) or Direct Torque Control (DTC), require information about state variables of the motor, especially stator or rotor flux and electromagnetic torque [1], [4], [5], [10], [11], thus fast digital signal processors (DSP) have to be applied for practical implementation [9]. To obtain the proper reconstruction of the internal electromagnetic variables of the IM, the speed information is required. Moreover the information about the rotor speed must be used in the field weakening algorithm of the traction drive control structure [7], [8] as well as in the diagnostic process [11]. Information about the rotor speed is necessary at zero or low speed operation, to realize properly the electrical braking [3]. During the start up to the nominal speed with nominal (or bigger) load torque, as well as during the breaking operation, electromagnetic torque must be controlled perfectly [12]. In traction drive systems without speed control loop, information about the rotor speed must be used in the field weakening algorithm [7], [8] and in the diagnostic process, as the drive reliability and safety are one of the key features of the traction vehicles. The temporary sensorless drive operation under emergency conditions (broken feedback from rotor position sensor) is another demand for the smart traction drive control. The main goal of this paper is to present the experimental sensorless traction drive system with speed and flux estimator based on the current-based Model Reference Adaptive System (MRAS CC ) [5], [12]. Dynamical properties of the sensorless 50kW traction drive are investigated and examined under experimental test in the whole speed range, including field weakening and low speed regions. Sensorless control structure One of the most popular control drive algorithm applied in the modern traction drive systems is based on the Direct Torque Control technique with Space Vector Modulation – DTC-SVM [1], [2]. The general scheme of this control structure is presented in the Fig. 1. Fig. 1. Sensorless DTC-SVM with MRAS type estimator In the DTC-SVM structure, the electromagnetic torque and stator flux vector magnitude are controlled by PI controllers in the synchronous reference frame. The essential stator flux angle information, necessary for the coordinate transformation, is obtained from the speed and flux observer. The estimated speed signal is used in the field-weakening algorithm and in the diagnostic process of the traction drive (Fig. 1). Information about the rotor speed, stator flux vector, electromagnetic torque, DC bus voltage, signals from the inverter and stator currents are used in the diagnostic process of the traction drive system. When these signals become bigger than the safety values, the drive should be stopped. Especially in an extremely low speed region these values must be observed. Any oscillations and abnormal transients are not allowed, too. The speed and flux estimator MRAS CC [5] applied in the analyzed control structure is based directly on the IM mathematical model [4]. ref e m m  A S C S B S s u e e m e s  ref s u  y x   s i ref s  ref sx u ref sy u ss  e m  est m  e m  e s  e e m d u C B A S , , C B sA i , , d u d u ABC    ref s u  sA i sB i e m  m  3