Initial Position Identification of Permanent Magnet Synchronous Motor Rotor Based on High-frequency Square Wave Injection

Objective Since the development of the permanent magnet synchronous motor, there have been position sensorless control algorithms for various operating conditions. Among these, identifying the initial position of the permanent magnet synchronous motor plays a crucial role in the motor’s starting performance. If the initial position cannot be accurately positioned, the motor can experience shaking, reversal, or even failure to start. At medium and high speeds, the back electromotive force information in the motor model is prominent, allowing the rotor position information to be indirectly obtained by extracting the back electromotive force data. However, at low speeds or even zero speed, the back electromotive force information is weak, and the rotor position can no longer be effectively identified. Therefore, studying the sensorless control algorithm of the motor at zero low speed is particularly important. Among all kinds of rotor position identification methods, the high-frequency injection method based on rotor salient pole characteristics is widely used and effective. It does not depend on the motor back electromotive force and has good control effects at zero low speed. The high-frequency pulse vibration injection method, in particular, is a widely used algorithm in the high-frequency injection approach. Traditionally, this method involves injecting a high-frequency sine wave voltage signal on the estimated rotor d-axis and using a filter to isolate the high-frequency response current of the motor under high-frequency excitation, obtaining the current information corresponding to the rotor position. However, the presence of the filter can introduce phase lag, which affects the accuracy of rotor position identification.Methods This study proposed an initial position identification method for permanent magnet synchronous motor rotors based on high-frequency square wave injection. Initially, the mathematical model of the system under high-frequency excitation was analyzed. When the motor operates at low speed or is stationary, and the injected high-frequency square wave signal’s frequency is significantly higher than that of the fundamental wave signal, the influence of back electromotive force and resistance voltage drop in the model can be disregarded, thus simplifying the motor model and the expression for response current. High-frequency square wave signals were then injected into the estimated rotor d-axis, and the response current separation method was utilized to differentiate between the high-frequency response current components and the fundamental wave current components of the response current. The fundamental wave current was used for current closed-loop control, while the high-frequency response current carried the rotor position information. The forms of the high-frequency square wave signal and the high-frequency response current are illustrated in Fig.1. Then, for the isolated high-frequency response current carrying rotor position information, a rotor position observer was designed without filters. The tangent function processed the high-frequency response current and decoupled the rotor position error information, which was then phase-locked to identify the rotor position. This method avoids the phase lag problem caused by the filter and uses the advantages of high frequency and large bandwidth utilizing square wave injection. After the preliminary positioning of the rotor’s initial position was completed, it was also necessary to identify the rotor polarity to determine whether the initial positioning of the rotor position and the actual rotor position were the same or opposite. This study, based on the magnetic saturation effect of the motor, applied current bias in different directions and equal amplitude to the reference d-axis in the identification of rotor polarity. The rotor polarity was determined by comparing the average value of the high-frequency response current amplitude of the d-axis during the operation time of the current bias. Therefore, the results of rotor polarity judgment are reliable and accurate.Results and DiscussionsThe entire permanent magnet synchronous motor (PMSM) position sensorless field-oriented control (FOC) system was designed, and a PMSM simulation and experimental platform was constructed to verify the feasibility of the proposed algorithm. The initial position identification of the permanent magnet synchronous motor in a static state and the motor’s performance in a running state were tested and verified. The algorithm accurately identified the rotor position in a static state and correctly determined the rotor’s polarity when the actual rotor position was 285°. The initially identified rotor position was compensated from 100.8° to 280.8°, which confirmed the accuracy of the position identification algorithm and the correctness of the polarity identification algorithm. In addition, the error in the initial position identification results of the proposed algorithm at various rotor angles was compared to that of the traditional high-frequency vibration injection. The proposed algorithm achieved more accurate identification results than the traditional method. After the initial positioning of the rotor was completed, the dynamic response performance of the motor was tested as it started and accelerated to the given speed, including when there was a sudden increase in speed. The steady-state performance of the motor at a constant speed and the response performance of the motor to abrupt forward and reverse rotation commands during the operation were also tested. Conclusions All experimental results demonstrated that this algorithm is suitable for permanent magnet synchronous motors at low speeds, and the dynamic control effect is good.