A Rotor Flux Observer for SPMSM Sensorless Drive Based on Linear Regression Model

A rotor flux observer is developed based on the linear regression model in surface-mounted permanent magnet synchronous machine. To estimate the rotor flux using voltage model, the additional feedback loop should be used to attenuate an unknown initial value, dc-offset, and dc-drift problems instead of the pure integrator. Using the Swapping lemma, a linear regression model was derived in a simple way using the relationship between back-EMF and rotor flux. Using the gradient decent method, the simply additional feedback loop was straightforwardly derived. Then, the observer stability was analyzed using Lyapunov stability. To compare the performance with the existing back-EMF-based observer, the effects of the 5th and 7th harmonics in the rotor flux and inverter harmonics on the sensorless positioning performance were analyzed. The validity was demonstrated by simulation and experimental results.