Controllo sensorless di motori sincroni a riluttanza alimentati tramite VSI e matrix converter

Synchronous reluctance (SyR) motor drives provide several advantages compared to induction and permanent magnet motor drives. Compared to induction motors, synchronous reluctance drives pose higher efficiency due to the absence of rotor cage losses. In comparison with permanent magnet motors, SyR motors are cost effective solutions since there is no magnet on their rotor's structures. If field orientation control approach is adopted for speed or torque control of SyR motors, the knowledge of rotor position is inevitable for appropriate performance of such control. Although direct torque control (DTC) technique is known an inherent sensorless control technique, however, flux observation deteriorates at low speed where back-electromotive force (EMF) information is too poor to be used to estimate the position of the stator flux linkage. Thus, rotor position information is also mandatory for DTC techniques. Conventionally, position information is obtained by means of a shaft mounted position sensor. These sensors, not only increase the system cost, but also decrease the reliability. To tackle these problems, many researches have been done in the past decades to develop sensorless control techniques for AC motor drives. Normally, at low speed and standstill, rotor position is estimated by means of high frequency injection based techniques. At medium and high speed levels, model based approaches are applied. Therefore, to cover a wide speed range sensorless control, high frequency signal injection based technique is combined with model based approach. Dealing with sensorless control of pure synchronous reluctance motors, many researchers have done interesting works for both high frequency injection based and model based techniques to estimate the rotor position. However, in all sensorless control techniques presented for pure SyR motors, maximum torque per ampere (MTPA) strategy was not adopted due to different problem and difficulties devoted to SyR motors which will be discussed in this dissertation. This work investigates the sensorless control of synchronous reluctance machine drives, focusing on the direct flux vector control method. The control operates in stator flux oriented coordinates, using constant switching frequency. Torque control is pursued through closed-loop control of the amplitude of stator flux and the current component in quadrature to the flux linkage vector. At standstill and low speed levels, rotor position is obtained through high frequency signal injection method. However, compared to commonly used high frequency current response demodulation, flux response is demodulated in this dissertation. This will lead to inherent cross-saturation effect compensation while in other techniques, extra model based approaches are required to compensate the effect of cross-saturation. At medium and high speeds, two methods are adopted to achieve the rotor position: one is based on flux estimation and the other is based on active flux concept. Furthermore, it will be extensively discussed that around zero load, sensorless control of pure SyR motors suffers from low saliency in terms of high frequency injection based technique and and also suffer from low back-EMF in term of model based technique. Using this analysis, a minimum flux plus MTPA strategy is proposed suitable for sensorless control of SyR motor drives. A hybrid position and speed observer is proposed based on the combination of the model based approaches and high frequency signal injection and demodulation. The two methods are fused together to form a unique position and speed estimate signals, with seamless transition between the two models. The proposed sensorless method covers a wide speed range from standstill to flux weakening. Also, the tuning aspects of the observer is addressed in this dissertation. In addition, sensorless control of matrix converter-fed synchronous reluctance motors is studied in this dissertation. Although the basic concept of the sensorless control technique is that of proposed for VSI-fed SyR motors, it will be shown that matrix converter nonlinear voltage errors coming from dead-time and voltage drop of IGBT switches is different from conventional inverters. Therefore, a self-commissioning technique is proposed to identify and compensate the effect of nonlinear voltage errors, no requiring of knowledge of IGBT switches a priori. This technique is suitable for sensorless control techniques at low speeds where the amount of voltage is comparable with nonlinear voltage error. Therefore, these errors should be compensated to obtain an accurate flux estimation. Results from simulations and experiments are provided to verify the effectiveness of the proposed methods.