Adaptive Analysis of Nonlinear Signal Base on EMD

Based on the principle of EMD, a novel method of constructing the time-space filter banks is proposed in which the adaptive filter cutoff gradient is determined when the mean significantly departs from zero. It is employed to analyze the turbine shaft vibration signal of Boeing 737 plane successfully. The component representing the rotation speed and the low-frequency component are extracted from the background noise. Compared with wavelet transform analysis, more satisfactory result is attained in the axis displacement measurement. This method can be applied in the analysis of other nonlinear and nonstationary signals as well.