中文摘要：

以4120SG柴油机为研究对象，通过故障模拟试验测试了柴油机在气阀正常状态、气阀漏气状态下的缸盖声发射信号，对燃烧段声发射信号进行经验模态分解（ExperienceModeDecomposition，EMD），计算了各个本征模分量（Intrinsic Mode Function，IMF）的能量分布及EMD能量熵。研究表明声发射信号能量主要集中在前3个IMF分量内，随着气阀漏气故障严重程度的增加，代表最高频IMFl分量能量相对增加，而能量熵H减少。通过自定义特征参数，提高了声发射应用于柴油机气阀漏气诊断方法的灵敏度。

英文摘要：

A 4120SG diesel engine is taken as the research object. The acoustic emission signals of the cylinderhead are measured when the diesel engine under normal and valve leakage conditions with different leakagedegrees. The signals are decomposed based on the Experience Mode Decomposition method （EMD）. The energydistribution of each Intrinsic Mode Function （IMF） component and the EMD energy entropy are calculated. Thecomparative analysis shows that the signal energy is mainly concentrated on the top three IMF components. Theenergy of IMF1, which represents the first high-frequency component, increases while the energy entropy Hdecreases with the severity of leakage increasing. On the basis, a self-defined characteristic value is put forwardto increase the sensibility of application of acoustic emission on the diagnostic method for the valve leakage.