Methodological Insights for Online Estimation of Induction Motor Parameters

This paper presents contributions for online estimation of states and parameters of an induction motor with Kalman filter. In order to ensure a good level of confidence of the estimation, a suitable methodology is proposed and two of its main points are investigated. First, an original method is used for tuning the covariance matrices, relying on the evaluation of the state noise due to modeling errors. Second, an observability analysis is developed, allowing to validate the model and the proposed excitation trajectory. Experimental results show that, with the chosen input signal, the parameters can be estimated with good accuracy in less than two seconds.