基于VMD-ICA-CWT的内燃机噪声源识别方法

Noise sources identification method of internal combustion engine based on VMD-ICA-CWT

提出了基于变分模态分解(VMD)、独立分量分析(ICA)和连续小波变换(CWT)相结合的内燃机噪声源识别算法.首先,对内燃机进行铅覆盖,只裸露待测的第6缸部分,测量裸露部分缸盖位置处的单一通道噪声信号;然后采用变分模态分解算法将其分解为各变分模态分量,并用FastICA算法提取各独立成分,解决了对单一通道噪声信号进行盲分离的欠定问题,同时克服了传统的经验模态分解处理噪声信号时出现的模态混叠缺陷;最后利用连续小波时频分析和相干分析,对分离结果进行进一步识别.研究结果表明:该算法能有效地分离识别出内燃机的燃烧噪声和气阀机构敲击噪声.

The variational mode decomposition（VMD）,independent component analysis（ICA）and continuous wavelet transform（CWT）algorithms were proposed to identify the noise sources.Firstly,the internal combustion engine was covered by lead,and only the sixth cylinder part was bared.The single channel′s cylinder head noise signal of the exposed part was measured,and then the variational mode decomposition algorithm was used to decompose the noise signal into several mode components,and FastICA algorithm was used to extract the independent components,which could solve the underdetermined problem of the single channel noise signal blind source separation,and overcame the modal aliasing defects of the traditional empirical mode decomposition meanwhile when processing the noise signals.Finally,continuous wavelet time-frequency analysis and correlation analysis were used to further identify the separating results.The results show that the proposed algorithm can effectively identify the combustion noise and the valve train knock noise of the internal combustion engine.