Extended state estimation of nonlinear modeled induction machines

Abstract

The authors report on the extended state estimation of induction machines. Based on a smoothly nonlinear machine model, local observability is analyzed for operation both with and without a speed sensor. An extended Kalman filter is designed so as to estimate not only electromagnetic quantities and the rotation speed, but also the load torque. Simulations demonstrate its excellent performance for sensorless field-oriented control.<>