< Back to previous page

Publication

Demodulation of vibration signals generated by defects in rolling element bearings under variable speed using morphological analysis and complex shifted morlet wavelets

Book Contribution - Book Chapter Conference Contribution

Rolling element bearings are the most common, widespread and important machine elements in rotating machinery, since they act as the ”interface” where the rotor forces are transferred to the stationary part of the machinery. The appropriate analysis of the vibration signals emitted by defected rolling element bearings can lead to the clear identification of the nature of the fault. The envelope detection or demodulation methods have been established as the dominant analysis methods for this application, since they can separate the useful part of the signal from its redundant contents. The classical methods are applied on signals captured under steady speed and reach their limits if the speed starts to vary. In this paper a novel demodulation method, generally applicable on non-stationary modulated signals and based on the morphological analysis and the Complex Shifted Morlet Wavelets, is introduced. The vibration signals, emitted by a rolling element bearing presenting a defect in the inner or on the outer race under a variable shaft rotation speed, are firstly processed using the morphological analysis. The raw signals are transformed by the morphological filtering through their interaction with another signal of simple geometry called structural element. The Beucher Gradient morphological operator is applied on the signals. A flat signal is used as a structural element SE and its length is selected using a kurtosis based maximization criterion. One or more Complex Shifted Morlet Wavelets are further used in order to obtain the wavelet transform of the signals. The center frequency and the bandwidth of each CSMW are firstly selected such as the frequency band of the first harmonic of the defects is covered. Exploiting the rotational invariance of the transformed signal, the time curve of the characteristic bearing fault frequency is extracted, following the change of the shaft rotation speed. The effectiveness and the robustness of the novel alternative method are evaluated in a simulated and an experimental case involving defective rolling element bearings.
Book: International Conference Surveillance
Publication year:2017