Rotor Position Error Compensation in Sensorless IPMSM Motor Drives

This paper presents a new method to compensate rotor position errors caused by parameter mismatching in model-based sensorless interior permanent magnet synchronous motor drives. According to the proposed approach, the error in estimating the rotor position is detected by injecting low-frequency current signals. Subsequently, the estimated rotor angular position is adjusted using a gradient descent algorithm until minimizing the ripple generated on the estimated rotational speed. A rotor position correction can be fully online accomplished coping with saturation and cross-coupling effects, as well as with unmodeled phenomena and unpredictable variations of the stator resistance with the temperature. The correction procedure can be accelerated by starting from an approximated correction table which can be offline built using the same technique. The proposed approach does not need additional current or voltage sensors, no motor design modifications, thus providing an inexpensive mean to improve the control perf