Online identification of load angle compensation for anisotropy based sensorless control

This paper proposes an online learning method that compensates for the load dependent saliency displacement. During normal sensorless drive operation the deviation between saliency based and fundamental model based position estimation is observed and used to adapt the parameters of a compensation function over load. The theoretical derivation comprises of linear and nonlinear approximation which both are investigated by experiments. In conclusion the proposed method may discard the usual need for a position sensor during the commissioning process of a sensorless drive.