民机驾驶舱舱音小波包分析及特征提取

Wavelet analysis and feature extraction for sound in civil aircraft cockpit

针对民航事故调查中的舱音背景声分析,利用小波包分析方法,对不同种类开关手柄声进行了三层分解,提取了不同开关手柄声的各频带归一化能量特征用于对开关手柄进行识别。最后,以B737客机的常用开关手柄的声音为研究对象,采集了大量舱音信号,利用该方法,对舱音信号进行了分析识别,并对结果进行统计验证。

Aiming to aircraft cockpit background sound analysis in civil aviation accident investigations,the wavelet packet analysis method was used to carry out time-frequency analysis of different types of switches'sounds,and then ex-tract the frequency band-normalization energy characteristics to identify various types of switches'sounds.Finally,the typ-ical switches'sounds of a B737 aircraft were chosen as the study objects,and a number of samples of various types of switches'sounds were collected from experiments.The sound samples were analyzed and the energy features were extracted and the results were statistically verified,they fully showed the validity of this method.