针对坦克变速箱故障时振动信号的复杂性和不确定性,提出了总体局部均值分解(Ensemble Local Mean Decomposition,ELMD)能量矩的故障特征提取方法,对坦克实车试验中测取的变速箱正常、主轴滚动轴承滚动体点蚀和3挡被动齿轮断齿3种状态下的振动信号进行计算。结果表明:该方法提取的特征能很好地反映变速箱的状态,从而实现对变速箱的故障诊断。 Aiming at the complexity and uncertainty of vibration signal in tank gearbox fault, a method of fault feature extraction for energy moment of Ensemble Local Mean Decomposition (ELMD) was proposed. The gearbox measured in actual tank test was normal , Spindle rolling bearing pitting and 3-speed passive gear tooth three kinds of state vibration signal is calculated. The results show that the features extracted by this method can well reflect the state of the gearbox, so as to realize the fault diagnosis of the gearbox.