Using an autoregressive model in the detection of abnormal characteristicsof squirrel cage induction motors

The problem of failures in large electrical machines is of great concern du e to its decisive influence over industrial productivity. This paper descri bes an autoregressive moving average (ARMA) estimation algorithm that uses as input/output data the starting current of an induction motor under test, The large current flowing in the motor may provide evidence of faults at a n earlier stage in its development. In fact, during the starting period it is observed that there is an imbalance of each phase impedance, a phenomena which is incremented when the saturation level is increased. The proposed autoregressive model allows the calculation of a quality factor to compare the harmonic level during the transient and steady state regimen. This para meter and other harmonic amplitudes, call be easily determined using the mo del and can be considered as true representatives of the machine condition. (C) 2000 Published by Elsevier Science S.A. All rights reserved.