Speed estimation based vector control of PMSM for high speed applications

Permanent magnet synchronous motors (PMSM) are widely preferred today in both industrial applications and in areas where advanced technologies are used, such as the defense industry, due to their high efficiency, compact structure and superior control capabilities. It is of great importance to obtain rotor position information accurately for the precise control of these motors. This information is usually provided by position sensors such as encoders or Hall sensors. However, disadvantages such as increasing system cost, complicating the structure and increasing maintenance requirements have brought sensorless control methods to the forefront. Sensorless control both reduces system cost and increases system reliability by reducing the risk of failure due to environmental conditions. For this reason, interest in sensorless control methods has increased in recent years, especially in applications that are critical in terms of cost and reliability. In this thesis, two sensorless speed estimation methods commonly used for PMSM, namely Model Reference Adaptive System (MRAS), Sliding Mode Observer (SMO) and hybrid observer structures combining the two methods, have been examined in detail. The developed structures have been modeled in MATLAB/Simulink environment and tested in scenarios where both reference changes and engine parameters change. According to the results obtained, it has been observed that the SMO method is more robust against changing engine parameters, while the MRAS structure can operate with less processing load. In line with the findings obtained from simulation analyses, the SMO method was preferred for experimental studies. For this purpose, an embedded control system based on TMS320F280025 microcontroller was designed and the real-time control capabilities of the system were evaluated. In the developed software, field-oriented control (FOC) structure was applied; rotor position and speed information were estimated via the SMO observer and integrated into the control algorithm. As a result, this thesis study demonstrates the applicability of sensorless control structures at both theoretical and practical levels and supports the field usability of SMO-based approaches, especially in PMSM driving systems.