Marco Buzzoni - Ph.D Student, Engineering Department in Ferrara - Séminaire du LVA

Blind deconvolution via maximum cyclostationarity

Blind deconvolution is an effective tool for the extraction of impulsive components from noisy observations having significant implications on the diagnostic of rotating machines. In this scenario, a general purpose method is proposed based on the eigenvalue decomposition. Furthermore, a cyclostationary criterion is introduced and validated using simulated and real signals. This criterion has turned out to be effective on the identification of early stage gear tooth faults as well as the monitoring of bearing fault development.