Abstract: Hilbert-Huang Transform(HHT) is regarded as a breakthrough method in the signal analysis domain. It is suitable to analyze nonlinear and non-stationary data. The vibration signal from the reciprocating engine is non-stationary, time-variation, so, application of HHT is a good choice. However, the technique cannot be expressed by an analytical formulation so that it is difficult to conduct theoretical analysis and performance evaluation. A multicompomnet signal was built. And then, compared with the Fast Fourier Transform(FFT) and Wavelet Transform(WT) and Wigner-Ville Distribution(WVD), HHT has the capability of the high efficient decomposition signal and accurately describing it, by which a reducing cross terms interference Time-Frequency Distribution(TFD) method was developed. Applying of HHT to analyze knock cylinder signal, result shows that it can effectively extract fault signal characteristic information and give a way of the condition monitoring and fault diagnosis for reciprocating machinery.