KKL observer design for sensorless induction motors
Résumé
We propose a novel observer for speed and torque estimation in induction motors, using only electrical measurements and assuming the parameters known. The design is based on a novel representation of the motor model taking the form of a cascade of a flux subsystem and a velocity subsystem with known injection terms. After giving sufficient conditions for the uniform strong observability of those subsystems, we exploit the KKL approach to design a globally asymptotically stable observer without relying on any time-scale separation. We provide a method to optimize the observer parameters and illustrate its performance on simulation in a realistic scenario.