基于Wigner-Ville分布和Wavelet时间尺度的飞机非平稳抖杆背景声分析

NONSTATIONARY QUIVER SPINDLE BACKGROUND SOUND ANALYSIS OF AIRPLANE VIA WIGNER-VILLE AND WAVELET TIME-SCALE DISTRIBUTION

飞机黑匣子中舱音记录器记录的舱音主要是由话语音、噪声和具有不同物理意义的背景声组成的复杂混合体,增加了舱音分析的难度,用传统的辨听和分析方法难以解决舱音特征等技术难题。以舱音记录器记录的抖杆背景声为例,针对传统Fourier变换的一次型时间/频率和短时Fourier变换的二次型能量在舱音辨听、分析的不足,将Wigner-Ville时频和Wavelet时间尺度引入到飞机舱音分析中,扩展了舱音分析方法。对比分析结果表明:平滑伪Wigner-Ville分布(SPWVD)及其时频分布的重排具有高的时频分辨率、时频凝聚性,而且没有交叉项干扰,适合分析非平稳舱音背景声;同时,离散小波变换的时间尺度分析技术确定抖杆背景声基频频率及位置,为获得非平稳舱音背景声特征提供一种值得借鉴的途径。

Cockpit voice recorder （CVR） in aircraft black box records many cockpit voices, such as speaker voices, noises, special meaning background sounds. Cockpit voices＇ complexity exacerbates analysis difficulty through traditional differentiating ＆ hearing methods. These methods＇ key techniques puzzles such as cockpit voices＇ characteristics identification are not resolved. According to the shortage in cockpit voice heating and analyses by traditional Fourier transformation（FT） single time/frequency and short time Fourier transformation （STFT） double energy, the new methods ,which can applied to analysis of no stationary cockpit background sound, is set up by means of introducing Wigner-Ville time-frequency and wavelet time-scale technology respectively into airplane cockpit voice analyses, as an example of quiver spindle background sound recorded by CVR. Compared with these methods, it is demonstrated that the spectrogram of both smooth pseudo Wigner-Ville distribution（SPWVD） and its time-frequency redistribution features high time-frequency resolution, immunity of cross term interference; and that the resonate frequency and response time are obtained by time-scale analysis of discrete wavelet transformation（DWT）.