Abstract
Acoustic source imaging has a wide range of applications in machine fault diagnosis, sources identification and vehicle NVH (Noise, Vibration and Harshness). One specific source of interest is cyclostationary source generated by rotating machine, which is rarely addressed by the conventional methods (Beamforming or Acoustical Holography) especially for a scenario of low SNR (Signal-to-Noise-Ratio). In this paper, the cyclostationary source is drowned by an unknown number of interferences and additive noise, and proposed cyclostationary source imaging technique aims at obtaining only the image of cyclostationary sources without any other interferences and noise. The proposed method consists in integrating the propagation function (Green’s function specifically) into the reduced-rank cyclic wiener filter based on the Minimum Mean Square Error (MMSE) criteria, which extended the classic reduced-rank cyclic wiener filter from the filtration (on the measurements) to the reconstruction (on the sources). After the theory is explained, an industrial application with engine bay on-site measurements is elaborated to validate the proposed method: a cyclostationary source radiated only from the High Pressure Pump (HPP) is reconstructed (only the image of HPP is obtained), and the noise interference from the engine is filtered (not shown in the final image) by the proposed method.
Original language | English |
---|---|
State | Published - 2017 |
Externally published | Yes |
Event | 46th International Congress and Exposition on Noise Control Engineering: Taming Noise and Moving Quiet, INTER-NOISE 2017 - Hong Kong, China Duration: 27 Aug 2017 → 30 Aug 2017 |
Conference
Conference | 46th International Congress and Exposition on Noise Control Engineering: Taming Noise and Moving Quiet, INTER-NOISE 2017 |
---|---|
Country/Territory | China |
City | Hong Kong |
Period | 27/08/17 → 30/08/17 |
Keywords
- Bayesian imaging
- Cyclostationary signal extraction
- Inverse acoustical problem