OZR opvangmandaat post doc Cédric Peeters

Project Details


Traditionally, most signal processing methods for vibration or instantaneous angular speed analysis of rotating machinery rely on the Discrete Fourier Transform or DFT. When it comes to the analysis of high-frequency data, the DFT has several advantages for the purpose of condition monitoring. Notably, the DFT is non-parametric, meaning no signal model assumptions need to be made to obtain a reliable spectral estimate. This property is especially useful in an industrial setting where an operator typically has no knowledge of the correct signal model. Nonetheless, making no assumptions can also be disadvantageous as the DFT does suffer from high variance, high sidelobe levels, and low resolution. Recently however, it was shown that there is significant potential for a novel type of high-resolution, semi-parametric spectral estimators to improve the performance of conventional signal analysis tools for machine monitoring. To the best of this author’s knowledge, no research has been done yet investigating these techniques for the purpose of condition monitoring. This project explores these spectral estimators on multi-sensor data of rotating machinery. This entails both the adaptation of these tools as well as the development of a novel concept for a high-resolution magnitude-squared coherence estimator and a sparse discrete-random separation method. The developed methods will be validated on experimental data coming from test bench data and offshore wind turbine data.
Effective start/end date1/10/2330/09/24


Explore the research topics touched on by this project. These labels are generated based on the underlying awards/grants. Together they form a unique fingerprint.