Cyclostationary-based bearing diagnostics under electromagnetic interference

Rolling element bearings are common but crucial components of rotating machines, being the interface between stationary and rotating parts and are responsible for many machine failures and breakdowns. Condition monitoring of machinery using vibration analysis often provides insight and early detection of damage on the bearings. The most widespread method for detection of bearing faults is Envelope Analysis, where the raw signal is first filtered around an excited frequency band presenting a high Signal-to-Noise Ratio (SNR) and is further demodulated, extracting the fault information. The selection of this band is commonly done using the Fast Kurtogram which detects the filter band by maximizing the spectral kurtosis of the filtered signal. In some cases the vibration signals are polluted by Electromagnetic Interference (EMI), commonly emitted by the Variable-Frequency Drives of motors. EMI presents an impulsive behaviour similar to the behaviour of faulty bearing signals. As a result the Kurtogram might lead Envelope Analysis to erroneous results. Recently, Cyclic Spectral Coherence (CSC) has been proposed as a powerful diagnostic tool, revealing in the frequency domain all the information of second-order cyclosta-tionary signals. In this paper an automated methodology for bearing fault detection under strong EMI is proposed based on CSC. The vibration signals are processed and the CSC map (cyclic frequency, frequency) is extracted. The integration of the CSC map along the frequency axis results in an enhanced version of the traditional squared envelope spectrum (i.e. spectrum of the squared envelop). Furthermore an optimization criterion is proposed in order to select the optimum integration limits leading practically to the selection of the optimal demodulation band, avoiding the influence of the EMI effect. The methodology is further validated and evaluated on vibration signals captured on an experimental planetary gear-box operating under strong EMI.

ISBN:9781510868458

Jaar van publicatie:2018

Toegankelijkheid:Closed