基于经验模态分解和独立成分分析的柴油机噪声源识别技术  

Identification of Diesel Engine Noise Source Based on Empirical Mode Decomposition and Independent Component Analysis Using EMD-ICA

作　　者：张俊红[1] 李林洁[1] 刘海[1] 王健[1] 王凯楠[1] 

机构地区：[1]天津大学内燃机燃烧学国家重点实验室,天津300072

出　　处：《内燃机学报》2012年第6期544-549,共6页Transactions of Csice

基　　金：国家自然科学基金资助项目(50975192);教育部博士点基金资助项目(20090032110001);国家高技术研究发展计划资助项目(2012AA1117064)

摘　　要：为有效地控制整机噪声能量和提高整机噪声品质,采用经验模态分解(EMD)和独立成分分析(ICA)技术,通过将EMD分解后的本征模函数作为ICA方法中的多个虚拟通道,解决了对单一采样信号进行盲源识别的欠定问题.将该思路应用于柴油机辐射噪声的主要噪声声源的识别研究,同时利用相干分析与时频分析技术实现柴油机噪声声源的准确识别.结果表明,EMD-ICA联合的噪声声源分离识别技术,可用来识别柴油机燃烧噪声、机械噪声声源,有效地克服了EMD技术在噪声声源识别中的模态混叠问题,降低了ICA技术对单一采样信号进行准确识别的难度.Noise source identification of diesel engine is important to noise control and noise quality im- provement. The techniques of empirical mode decomposition （EMD）and independent component analy- sis （ICA） are adopted in the study. The intrinsic mode functions resulting from EMD are used as the virtual channels of ICA, which solves the undetermined problem of blind source identification from single sam- piing signal. The method is applied to separate the main noise sources of diesel engine radiation noise, the techniques of coherence analysis and time frequency analysis are used to identify the noise sources. Result shows that the combined technology of EMD-ICA can be used to identify combustion and mechanical noise sources of diesel engine, which eliminates the mode mixing in EMD, and simplifies noise sources identification when ICA is used for accurate noise sources identification for single sampling signal.

关 键 词：柴油机 经验模态分解 独立成分分析 噪声源识别 

分 类 号：TK421.6[动力工程及工程热物理—动力机械及工程]