An Improved Method Of Hilbert Vibration Decomposition And Its Application In Mechanical Fault Diagnosis

The vibration signal of rotating machinery is unsteady,and its fault diagnosis is a difficult problem.On the basis of studying the Hilbert vibration decomposition algorithm,this paper improves it and applies it to the fault identification of gears and bearings,and verifies the effectiveness of the proposed method.The work and achievements in this paper are as follows:1.The decomposition ability of Hilbert vibration decomposition is studied,and the shortcomings of Hilbert vibration decomposition are found out.Then a Hilbert vibration decomposition method based on mirror continuation is proposed to solve the end-effect shortcomings of Hilbert vibration decomposition method.The decomposition effects of HVD method and EMD method are compared by simulation signal decomposition,and it is found that HVD method overcomes the end-effect shortcomings and mode mixing.Overlapping is more advantageous.2.A mechanical fault diagnosis method based on fast kurtosis diagram and Hilbert vibration decomposition is proposed.The method is verified by the diagnosis examples of gear fault and generator bearing.Using HVD to decompose the signal of gear or rolling bearing,separates the high frequency modulation signal,highlights the fault signal characteristics,then decomposes the fault signal of gear or rolling bearing,and further uses the fast kurtosis map to analyze the high frequency component,which can enhance the fault feature extraction ability of the fast kurtosis map.3.A mechanical fault diagnosis method based on Hilbert vibration decomposition and envelope spectrum is proposed.The mechanical signal can be decomposed by HVD and the mechanical fault features can be extracted by combining envelope spectrum.The envelope spectrum method based on HVD is better than single envelope spectrum analysis,and this method is simple and applicable.4.A method of mechanical fault diagnosis based on HVD and threshold de-noising is proposed.The principle of threshold de-noising and the selection of threshold are analyzed.The signal is decomposed by HVD method,and then the decomposed components are de-noised by threshold.The mechanical fault can be diagnosed by combining Fourier transform.