Sparsity Analysis of a Gearbox Vibrations Signals

JAVIER ERNESTO KOLODZIEJ; FERNANDO BOTTERÓN

Curitiba - PR

Seminario; XIX SEMANA DA ENGENHARIA DE PRODUÇÃO E MECÂNICA SULAMERICANA: XIX SEPROSUL; 2019

Universidad Federal de Paraná - Curitiba

In this paper, an analysis of gearbox vibration signals sparsity is proposed, considering a time-frequency domain, defined from de discrete Fourier Transform (DFT), and the definition of a hermitian angle between a complex elements vector, grouping together the samples of the DFT of signals acquired from multiple sensors, and a reference vector. This analysis contributes to understand the vibration signal separation algorithms performance. A particular case considered is vibration signals coexisting in the same frequency bin and coming from different mechanicals pieces. It could occur due to the design or construction characteristics and due to nonlinearity. The final goal of this work is to contribute with the implementation of embedded processing, or in sensor processing, needed to the manufacturing equipment conditions monitoring using the structure proposed to the internet of things, considered the really first step toward the future smart industry materialization. A measurement for vibration signals sparsity is also proposed to asses, the best operation condition for blind source separation applying time-frequency masking.