Skip to Main content Skip to Navigation

Génération de symptômes flous par FFT et SWT : applications à la détection de vibrations

Samir Taleb 1
1 GIPSA-SA-IGA - GIPSA - Signal et Automatique pour le Diagnostic et la Surveillance
GIPSA-DIS - Département Images et Signal, GIPSA-DA - Département Automatique
Abstract : The purpose of this work is fault detection through changes detection or anomalies in the measured signals. Our work presents a method making possible to build a behavioral model of the process, using historical process data (reference signals), in order to detect abnormal situations resulting from dysfunctions. This method consists in generating on line progressive symptoms, making possible to follow the temporal evolution of the system operation to be supervised. These symptoms take into account uncertainties on the decision by aggregating the result of the detection test with that of its validity. By their properties of frequency and time-frequency localization, Short Time Fourier Transform and Stationary Wavelets Transform are two methods adaptable to fault diagnosis. The choices of the weighting window length and type for the STFT, of the mother wavelet and the number of decomposition levels for the SWT are the freedom degrees which allow this method to extract relevant information from the analyzed signals. A gradual symptom is obtained according to the type and from of the extent fault. The fault detection method conceived was initially tested on signals containing simulated faults was then validate on industrial signals to detect the torsional vibrations in a reversing rolling mill (hot rolling). The freedom degrees of the suggested method confer flexibility to the developed methods with regard to different applications.
Document type :
Complete list of metadatas

Cited literature [41 references]  Display  Hide  Download
Contributor : Patricia Reynier <>
Submitted on : Tuesday, February 5, 2008 - 2:12:53 PM
Last modification on : Thursday, November 19, 2020 - 1:00:53 PM
Long-term archiving on: : Friday, May 7, 2010 - 6:25:25 PM


  • HAL Id : tel-00239310, version 1


Samir Taleb. Génération de symptômes flous par FFT et SWT : applications à la détection de vibrations. Automatique / Robotique. Institut National Polytechnique de Grenoble - INPG, 2007. Français. ⟨tel-00239310⟩



Record views


Files downloads