研究了将经验模式分解(Empirical Mode Decom position,EMD)、遗传算法及BP神经网络相结合对柴油机振动信号进行故障诊断的方法。首先运用经验模式分解方法对柴油机缸盖表面振动信号进行分解并提取特征参数;然后利用遗传算法对得到的特征参数进行选择,找到对于故障诊断最为敏感的参数;最后建立了BP神经网络模型对柴油机典型故障进行诊断。通过对某型柴油机的验证,表明该方法能够准确识别柴油机供油系统的典型故障。 The method of fault diagnosis of diesel engine vibration signals by combining empirical mode decomposition (EMD), genetic algorithm and BP neural network is studied. First of all, decompose the vibration signal of diesel cylinder head surface by using the empirical mode decomposition method and extract the characteristic parameters; then use the genetic algorithm to select the characteristic parameters to find the most sensitive parameters for fault diagnosis; finally establish the BP neural network model of diesel engine Typical fault diagnosis. The verification of a diesel engine shows that this method can accurately identify the typical fault of diesel engine oil supply system.