Wide speed sensorless control of pmsm drive with smooth transition between HFSİ and extended luenberger observer

Avcı, Mustafa Mus Ab
Süreli Yayın başlığı
Süreli Yayın ISSN
Cilt Başlığı
Graduate School
It is well-known that the most efficient way to generate mechanical power is to use electric motors. In the beginning, conventional DC motors with commutator excitation are used. However, with the spread of AC power distribution, AC motors have become popular because they have superior efficiency and low maintenance and installation cost. Thus, induction motors have become a dominant factor in the industry. Moreover, although induction motors have satisfied the needs of the industry for a while, the energy shortage and the efficiency criteria caused a reorganization with respect to efficiency classes among electric motors. Thus, using new-generation motors with permanent magnets to create magnetic excitation has become indispensable worldwide. However, the transition to permanent magnet motors requires a motor controller for power electronics. Since, unlike induction motors, electromechanical commutation must be carried out electronically. With the improvements in semiconductor technologies and digital signal controllers, controller chips are becoming more available with low cost and high horsepower. In addition, sophisticated digital signal controllers allow engineers to develop superior control algorithms. At first, AC motors could only be controlled via scalar control. Since only controllable quantities in scalar control are voltage and frequency, naturally controlling bandwidth and dynamical responses are poor compared to vector control schemes. One of the well-known vector control schemes is field-oriented control, which enables control of motor armature and excitation voltages independently and on a vector basis. Thus, new-generation motors are driven with new-generation controllers. Although field-oriented control is one of the suitable control methodologies, it requires geometric knowledge of the position of the rotor flux vector during the operation. Besides, there are direct and indirect sensing operations to achieve rotor position information. Although direct sensing of rotor position information via a position sensor directly mounted on the shaft is the simplest way, there are some drawbacks. Especially, position sensors increase total system cost and decrease the reliability of the drive system according to environmental conditions like temperature, moisture, and altitude. On the other hand, indirect sensing methods rely on mathematical manipulations and computation via machine parameters like voltage, current, resistance, and inductance. Regarding self-sensing methods, they could be categorized in a manner of machine model-based and saliency tracking-based models. Machine model-based approaches use observers to get rotor flux position information by iteratively computing machine model state equations. Unlike model-based solutions, saliency tracking solutions require a signal injection concept to detect rotor position information from the demodulation of modulated injection frequency through the motor itself. In summary, although model-based approaches have promising performance in medium to high-speed regions, they have poor or moderate performance and controllability in the low-speed region; because of that, the magnitude of back electromotive force is proportional to the rotating speed. Contrarily, saliency-tracking approaches have better performance in terms of the quality of estimated position information in zero, low and nearly zero-speed regions. However, as the rotational frequency increases, signal processing becomes a burden and uncontrollable with the same sampling infrastructure and dynamics. Consequently, hybrid observer structures have become more popular since the two approaches have beneficial features with respect to operational speed region. A soft transition method must be developed to combine two self-sensing methods within the stable operation in terms of transients and steady-state operation. In this thesis scope, a hybrid sensorless field-oriented control methodology is proposed. First, the literature is reviewed in the aspects of machine model-based sensorless algorithms and saliency-tracking approaches. Two of the self-sensing methods are further investigated. One is the Luenberger observer that computes back electromotive force to estimate rotor flux position, and the other is high-frequency signal injection. Also, a soft transition from one method to the other is proposed. Furthermore, verification is made by modeling, analysis, and simulation using the MATLAB®/Simulink® environment. In the simulation, the motor is started with the estimated position referenced by the outcome of the high-frequency signal injection method. Next, beyond a defined transition point, the estimated position reference is changed to the observer algorithm. It is observed that a soft transition is necessary to keep the system under stable conditions. After the hybrid method is realized and implemented, a motor driver is designed via Altium. All component selection, design of gate drive circuits, design of current and voltage measurement circuits, as well as digital and analog interfaces are described. Beyond the hardware design phase, embedded software is developed to run hybrid control algorithms. Besides, embedded software flow and control loop descriptions are detailed. Also, a test bench including two identical motors, a dummy load, and a rectifier circuit for the generator side is prepared for the experiments. Experimental results are recorded, discussed and presented.
Thesis (M.Sc.) -- İstanbul Technical University, Graduate School, 2023
Anahtar kelimeler
electric motors, elektrik motorları