Research In Gear Box Fault Diagnosis Based On Improved Hilbert-Huang Transform

In the view of fault vibration signal non-linear, non-stationary nature, we introduced the Hilbert-Huang transform method, which includes the empirical mode decomposition (EMD) and Hilbert transform of two parts. The signal was decomposed into a limited number of Intrinsic Mode Functions (IMFS), then for each IMF into Hilbert transform.In this paper, in order to improve the envelop overshoot and Hilbert-Huang spectrum endpoint effect, we presented a least-squares method to the trend items and cubic B-spline based on a sliding average method of improved Hilbert-Huang algorithm. By summing up the strengths and weaknesses in various methods of interpolation and extension, we use a least-squares method to the trend before seeking IMF, and further use the cubic B-spline based on the moving average method for directly fitting the mean sequence to replace the original method seek to strike a mean way, avoiding the use of spline interpolation and then seek envelope caused by many problems. According to the new algorithm is to make flow charts, we introduced a simulation signal in the comparative study, that the new method can reduced the number of false weight and energy leakage, and showed an initial advantage.Study on the Gearbox vibration for the mechanism, we applied the improved Hilbert-Huang Transform method into the fault diagnosis. We used the improved Hilbert-Huang Transform for precise diagnosis, Hilbert-Huang spectrum and marginal spectrum, and then carried out detailed analysis. In the meshing frequency and gear shaft rotation frequency nearby, it presented the breakdown characteristic, the high-order harmonic amplitude of the relative increase more, occured the many score overtones and frequency fluctuations, and found the obvious frequency modulation effect. According to the signal characteristics of a successful diagnosis of the fault that occurred is the failure of tooth surface wear.