High performance sensorless solutions for induction motor control

This paper presents a comparison between four sensorless solutions for high performance induction motor control. The presented solutions are developed around two classical schemes based on direct flux oriented control (DFOC). The first one is based on the Luenberger rotor flux observer and the second one on the model reference adaptive scheme (MRAS) with voltage-current (UI) and current-speed (I/spl omega/) as rotor flux observers. For both cases have been tested the classical methods to obtain the estimated rotor speed (LNBPI), (MRASPI). In addition for Luenberger based scheme have been tested an original method to obtain the estimated rotor speed (LNBG), and for the MRAS based scheme a neural network have been developed in order to perform the estimation tasks (MRASNN). All four methods have been implemented and tested using a fixpoint Motorola DSP56F805.