Publication Date:
2019-07-13
Description:
Special techniques must be applied when analyzing acoustic noise data from nonstationary sources such as aircraft flyover measurements. Since the Fourier transform is time dependent, the noise signal is divided into short time segments by introducing a window function so that the spectral characteristics remain reasonably stationary. Reducing the window width reduces the frequency resolution while increasing the window duration can lead to spectral smearing. A trade-off must be made between time resolution and frequency resolution. The effects of varying the window duration on narrow-band acoustic spectra and thus the frequency bin width are addressed in this study. The influence of window functions (rectangular, Hamming, etc.) are also investigated. Both a tonal noise source, XV-15 aircraft in the airplane mode, and a broadband noise source, a F-18 aircraft, are considered. When dealing with flight test data, not only is the signal nonstationary, often it is contaminated by both ambient background noise and internal noise generated by the data acquisition system and power generators. Since generator noise is highly tonal, this can be particularly troublesome when computing tone-corrected perceived noise level (PNLT). A scheme is presented in this paper to eliminate unwanted background and internal noise.
Keywords:
ACOUSTICS
Type:
(ISSN 0736-2935); : The use of EOS for
Format:
text
Permalink