Technical Program



Title

An Average Spectral Coherence Approach for Feature Extract of Fan Vibrations


Topic

2.2 Experimental Methods for Localizing / Characterizing Sources


Authors

CHU Ning
Zhejiang University

Hangzhou - China
LI Shyang
Zhejiang University

Hangzhou - China
WU Dazhuan
Zheijang University

Hangzhou - China

Abstract

Acoustic noises produced by fan aerodynamic performance are generally transient, non-stationary and broadband. To extract key features from these complicated noises, this paper proposes an improved cyclostationary approach combining the compressed sensing and Kurtosis spectrum. Mechanical interference and background noise can be attenuated and filtered by sparsity-based compressed sensing technique. Meanwhile acoustic source can be accumulated and enforced by adequate linear array of microphone sensors. Thus informative characterization of fan acoustics can be drawn up comprehensively by Kurtosis spectrum from cyclostationary statistics of fan acoustic noises.