Feature Extraction from Vibration Signals of Engine Based on Hilbert-Huang Transformation
Hao Da-hai · Ordnance Industry Automation · 2007
HHT (Hilbert-Huang Transformation) was combined with wavelet analysis as a method of feature extraction. Cylinder head vibration signal was decomposed and restructured in low, middle, and high frequency bands by wavelet. Signal in each frequency band was divided into some IMFs (Intrinsic Mode Function) by EMD (Empirical Mode Decomposition). Then Hilbert spectrum of signal in each frequency band was obtained by HHT. According to characteristics of impulse response in vibration signal of engine, energy integration along frequency axes was projected to the time-amplitude plan. An example of practical application shows that faults can be judged and located through different energy distribution of each frequency band and impulse response.