New Time-Frequency Analysis of Acoustic Data
Yih Nen Jeng, Jiaming Huang, Jia‐Lin Tsai, You-Chi Cheng · 2006
A new time-frequency transform is employed to examine acoustic data of an old motorcycle, an Unmanned Aerial Vehicle (UAV), and an F-16 fighter. For a given data string, zero points around the two ends are identified by searching and interpolation. After dropping segments beyond two zero ends, the corresponding Fourier sine spectrum is obtained. The time-frequency transform imposes finite bandwidth window upon the Fourier sine spectrum centered at a given frequency. The inverse Fourier transform of the band-pass limited spectrum gives the real part of the resulting spectrogram. This procedure avoids the error induced by the finite data length and excessive average other than the Fourier transform. By carefully inspecting the real part plots of the spectrogram, one can pick many information about the structure vibration and aero-acoustic data. For the motorcycle data, a series of structure vibration modes and random vibration from low to high frequency oscillations are grasped. The minor frequency shifts of high frequency modes with respect to time and random vibration show that the motorcycle is not at a good condition. The UAV is a helicopter and the examined data covers the fuel run out of point. Around the low frequency regime, the wind produces ultrasonic acoustic signal and is captured by the spectrogram. After the fuel run out of instant, propellers do not gain power to rotate again and gradually lost the rotation speed so that all the corresponding modes show a linearly decayed frequency. The spectrogram of F-16 noise recorded in the cockpit reflects the harmonic and sub-harmonic modes of