Robust multivariate control charts for early detection of broken rotor bars in an induction motors fed by a voltage source inverter

In this paper, we present a work in progress about the development of an expert system to detect incipient faults related to asymmetries in the rotor cage in induction motors fed by Voltage Source Inverters. The expert system is based on a condition monitoring technique based on robust statistical tools, and it makes use of the Fast Fourier Transform to obtain the spectrum of the motor line current and a multiresolution technique using wavelet functions to detect significant peaks in the spectrum and to measure the height of these peaks with respect to the "baseline" signal. Finally, a Quality Control approach based on robust multivariate control charts is applied to detect a progressive deterioration of the rotor cage. This technique has been successfully applied to the early detection of broken bars in induction motors fed by a sinusoidal voltage supply and it is to be adapted to the case of voltage source inverter supply. A case-study is shown in which a cage fault condition was provoked by drilling a hole in one of the bars of an induction motor obtaining different fault conditions by progressively making the hole deeper and a great deal of laboratory tests were performed for each fault condition being the motor supplied by a power converter with a fundamental frequency of 35 Hz.