An Extended Kalman Filter and an Appropriate Model for the Real-time Estimation of the Induction Motor Variables and Parameters

摘要

This paper presents an efficient discrete-time second-order model of an induction motor for the rotor flux and parameters real-time estimation using an Extended Kalman Filter. Compared with the usual model, this model offers many advantages for real-time identification and fault diagnostic of the induction motor. Indeed not only, it remains stable and accurate for large sampling periods, but also it offers a better linearity with respect to the estimated parameters and it reduces the computational cost of the Kalman filter. Experimental results show the great accuracy and the fast convergence of the estimated parameters, even for sampling periods larger than 10 ms. In practice, the optimization of the measurement procedure and the realistic tuning of the Kalman filter has allowed the real evaluation of the parameters uncertainties and reveals the errors due to the model simplification of the motor such as the iron losses.