基于独立分量分析的内燃机噪声信号分离  

Separation of Engine Acoustic Signal Based on Independent Component Analysis

作　　者：刘月辉[1] 张俊红[1] 郝志勇[2] 

机构地区：[1]天津大学机械工程学院,天津300072 [2]浙江大学机械与能源学院,浙江杭州310027

出　　处：《内燃机学报》2005年第1期63-66,共4页Transactions of Csice

基　　金：国家自然科学基金资助项目(50175078)。

摘　　要：采用独立分量分析的方法对内燃机的噪声信号进行了研究。首先对独立分量分析的基本原理进行了简要的叙述,在FastICA算法的基础上建立了内燃机噪声的独立分量分析模型,并编制了相应的计算程序。以某四缸柴油机为研究对象测量了不同工况下的噪声信号,计算了这些噪声信号的统计峰度,确认其为非高斯信号,满足独立分量分析的基本要求。然后对测得的柴油机噪声信号进行了独立分量分析,将其分解为一系列不同的独立分量。为了了解各独立分量的时频分布,用小波变换的方法对它们进行分析,分析结果表明,这些不同的独立分量对应着不同的内燃机噪声源信号。A study on identifying the engine noise source from engine acoustic signal based on the independent component analysis was carried out.The principle and algorithm of independent component analysis were briefly introduced,and the independent component analysis model for engine noise was given based on the FastICA algorithm. The kurtosis of the recorded acoustic signals of a diesel engine was calculated and a characteristic of non-Gaussian signals was presented,which meets the requirement for the independent component analysis.The independent component analysis was conducted in engine where the acoustic signals were decomposed into several independent components. Furthermore,the wavelet transforming was applied on these components for getting their time-frequency information.The analysis showed that these independent components would correspond to different engine noise source signals.

关 键 词：内燃机 噪声 独立分量分析 小波分析 

分 类 号：TK402[动力工程及工程热物理—动力机械及工程]