发动机轴承故障诊断方法的研究

针对EMD（Empirical Mode Decomposition）方法中存在的端点效应和IMF（Intrinsic Mode Function）虚假分量过多的问题，提出了基于互相关的EMD方法。首先，对非平稳性信号进行互相关延拓消除端点效应；其次，对消除端点效应的信号进行EMD分解，并将分解后的IMF与原信号作互相关，保留与原信号最相关的IMF；最后，作出信号的Hilbert边际谱，识别信号的频域特征。仿真结果表明该方法能够有效地克服端点效应，分离出真实IMF。将其应用于船舶发动机滚动轴承故障诊断中，能有效地识别出故障特征。

Study of bearing fault diagnosis method in marine engine

An EMD method based on cross-correlation is proposed in view of the end effects and excessive false component of IMF. First, extension using cross-correlation analysis of non-stationary signal is applied to eliminate the end effect. Second, signal after eliminating end effect is decomposed with EMD, and keep the best IMF through cross-correlation between IMF and original signal. Finally, the Hilbert marginal spec-trum of signal can be obtained and frequency feature is identified. The simulations reveal that the EMD method based on cross-correlation can solve the end effects perfectly and true IMF can be separated out effectively. The method is applied in rolling bearing faults diagnosis in marine engine. The result shows that the fault can be identified exactly.