Detection of Broken Rotor Bars in Induction Motors Under Oscillating Load Through Bidimensional Entropy and Time-Frequency Maps

Induction Motors are key elements of most realworld processes, their common use is no coincidence, but rather the result of a combination of reliability, simplicity, efficiency, and low maintenance costs. Ideally, Induction Motors operate under constant load conditions but in real applications there exist several factors that produce oscillating or fluctuating loads that trend to mask fault-related patterns generating false alarms that lead to unnecessary interventions. Thus, to overcome this issue, this work proposes a diagnosis methodology for detecting the occurrence of broken rotor bars in Induction Motors under oscillating load, the contribution of this work lies in analyzing time-frequency maps by means of bidimensional entropy features. The time-frequency maps are obtained by Short-Time Fourier Transform from stator currents and the computation of bidimensional entropy features allows its characterization under transitory and steady-state regimes. The obtained results demonstrate the potential of bidimensional entropy features to assess the occurrence of broken rotor bars in Induction Motors under oscillating load, also, it is demonstrated that bidimensional entropy features are immune to oscillating load conditions.