Transformée en ondelettes empirique (EWT) : Application en traitement de signal de défaut de machines tournantes

Abstract

This dissertation focuses on the use of Empirical Wavelet Transform (EWT), a time-frequency analysis method, for the detection of bearing and gear faults in rotating machines. After a general presentation on signals in mechanics, the advantages and disadvantages of non-stationary stationary signal analysis methods, we presented the basic principles and advantages of EWT compared to other analysis techniques time, frequency and time frequency. The application of the periodogram and spectrogram on stationary and non-stationary signals shows the different analysis possibilities, making it possible to compare the EWT method to other methods. Finally, the results obtained are synthesized and the importance of the EWT in monitoring of frequencies and amplitudes is emphasized.