柴油机的缸盖噪声信号含有很多能诊断柴油机故障的有用信息，如燃爆、气门开启和关闭信息。但是提取的缸盖噪声信号中也含有柴油机工作时产生的排气噪声和机体其它部位辐射噪声等干扰信号，且干扰信号与有用信号的频率成分在20KHz以下的频带内重叠在一起。因此如何能有效的提取有用信号十分关键。本文通过实验选取气门室罩距气门较近一侧的侧表面1CM外作为测点。分析了小波分解、FIR滤波、小波降噪和FIR分频带阈值处理降噪四种方法的优缺点后，发现小波降噪既能保留有用信息，又能有效将不需要的噪声信号去除，在提取缸盖噪声中的有用信息时具有明显的优势。将缸盖噪声信号降噪后发现，各种气门间隙和失火故障均可通过缸盖噪声判断出。

The acoustic pressure signal near the engine valve cover contains many useful information which can diagnose the fault of diesel engine, these information includes combustion, the valve open and close information. But this signal also contains many interfere noise, like exhaust noise and other noise eradiated from the other part of the diesel engine, and the frequency composition of useful signal and interfere noise are largely overlapped in the frequency band under 20KHz. So how to abstract the useful signal is very important. In this paper,1CM outside the valve side of the engine valve cover was selected as the measure point. Four kinds of signal processing methods: wavelet decomposition, FIR filter, wavelet denoise and FIR divided frequency bands threshold valve denoise was investigated . We find wavelet denoise can reserve useful signal and eliminate interfere noise in the same time, have the obvious advantage in cylinder head acoustic signal processing. After the cylinder head acoustic signal was denoised, the misfire and valve clearance related fault could be diagnosed through the cylinder head acoustic signal.