Sensorless Control of IPMSM Based on Regression Model

Abstract

A position-sensorless algorithm is developed for an interior permanent-magnet synchronous motor, while reflecting the saliency in an extended electromotive force term. Active flux is converted into a new linear regression form, and a high-pass filter was applied. Then, the gradient algorithm is applied to derive an estimate of the active flux in the stationary frame. The proposed observer is an inherently sensorless type because it does not require the rotor speed or position information. It is practically attractive since it does not require the use of pure integrator that is marginally stable. In addition, the observer can be constructed without the phase modulation flux linkage constant. It is basically a model-based method. However, it is easily extended to a signal injection method that is robust in the low-speed region. Ultimate boundedness is established using the persistency of excitation condition. Validity of the algorithm is demonstrated experimentally via torque and speed control with a full load in the low-speed range.