Orbit Analysis From a Stray Flux Full Spectrum for Induction Machine Fault Detection

Abstract

Magnetic flux sensors in particular, combined with appropriate signal processing techniques, create excellent opportunities for extracting significant information for fault diagnosis of induction motors. Recent research has opened the possibility of simultaneously obtaining magnetic flux signals displaced approximately 90° in time. This makes it possible to correlate the stray magnetic flux from different transducers. In this scenario, the Full Spectrum technique, widely used for vibration signals, can become a useful tool for displaying the relationship between different magnetic flux components in the frequency domain. Taking this into account, this work proposes the use of the filtered orbit, modified and obtained from the Full Spectrum, for interpreting the magnetic flux signals. This was done for the motor working under inter turn short circuits (three, six and nine turns short), unbalanced voltage fault or without fault conditions at different loads. Two characteristic frequencies of faults were considered, 180 Hz and 420 Hz. The experimental results showed the viability of the application of the technique and made it possible to identify the severity and the distinction of the faults studied as the orbital area and the direction of rotation were changed due to the presence of the fault. For example, for the incipient failure (three turns short), considering all loads, the orbit area at 180 Hz increased no less than 92% and the unbalanced voltage by no less than 885%. Additionally, the pattern of the orbit rotation at 420 Hz, was also affected by the operating condition of the motor under unbalanced voltage.