Démarche d'Analyse Spectrale en vue d'une Interprétation Automatique, Application à un Signal d'Engrenages

Matthieu Durnerin (1) , Nadine Martin (1) , Corinne Mailhes (2) , Gérard Lejeune (1) , Joël Liénard (1) , Jean-Louis Lacoume (1) , Francis Castanié (2) , Pascale Prieur (3) , Georges Goullet (4) , André Silvent (1)

1 CEPHAG - Centre d'Études des Phénomènes Aléatoires et Géophysiques
2 IRIT-SC - Signal et Communications
3 DER/SDM
4 CTSN/DLSM

This paper describes a complete spectral analysis process, using the knowledge of both experts in signal processing as well as experts in fields of application (e.g. acoustic, vibratory mechanics) with the objective to estimate a signal spectral content. The contribution of this paper is the setup of a decision process based on the comparison of several methods in spectral analysis (e.g. Fourier, Maximum Likelihood, parametric, deflation) along with a spectral interpretation. A preanalysis consists in evaluating some signal properties of the signal (stationarity, gaussian hypothesis) and in studying its autocorrelation. Then, the process is builtfrom a succession of analyses, leading to the detection and identification of the different spectral patterns. The interpretation of each step is simplified by the definition of objective criteria. Finally, some additional analysis are considered (cyclostationarity, harmonic research).

Domains: Signal and Image processing; Signal and Image Processing

Matthieu Durnerin, Nadine Martin, Corinne Mailhes, Gérard Lejeune, Joël Liénard, et al.. Démarche d'Analyse Spectrale en vue d'une Interprétation Automatique, Application à un Signal d'Engrenages. GRETSI 1997, Sep 1997, Grenoble, France. pp.539-542. ⟨hal-00263829⟩