Online Identification of Induction Machine Parameter Deviations for Aging Detection - A Comparative Study Using Recursive Least Squares Algorithm and Extended Kalman Filter

Abstract

The use of electrical machines in automotive traction systems is rapidly increasing. To ensure operational safety, the machine behavior is monitored to detect failures or aging effects. Besides other approaches, online parameter identification is suited for real-time observation of the machine condition during operation. Two of the most established online parameter identification algorithms are the recursive least squares and the extended Kalman filter algorithm. In existing approaches the algorithms identify the absolute parameter values. In this paper the used identification models are modified to directly identify the parameter deviation related to the reference values. This results in an additional advantage in identifying operational parameter changes because nonlinear behavior is provided by the respective parameter reference. The performance of the proposed algorithms to monitor different electrical parameter changes is compared using an extended analytical induction machine model.