Machinery Fault Diagnosis Method Based On Empirical Wavelet Transform

As the main components of machinery,rotor, gears and bearings is play a decisive role.Therefore, the fault diagnosis method research of these components is of great significant.At present, the time-frequency analysis method for processing non-stationary machinery fault vibration signal is the main method. This paper uses Experience Wavelet Transformthis which is a new kind of signal processing method,combined with some other means of signal processing method for machinery fault vibration signal processing.The main contents are as follows:1 Time-frequency Analysis Based on Empirical Wavelet Transform in Fault Diagnosis of MachineryEmpirical Wavelet Transform(Empirical Wavelet Transform, EWT) is an adaptive signal processing method for Dividing and filtering the signal spectrum.At first, the Fourier spectrum maximum of the signal is divided adaptively,then establishing a set of wavelet filter bank to filter the divided spectrumin and get a group of single component ingredients. Hilbert transform every single component ingredients you can get the instantaneous frequency and instantaneous amplitude. At last applied this method to the fault diagnosis of rotating machinery,in order to verify the effectiveness of the method for simulation of signal and several groups of typical experimental rotor fault signal, this paper compares the performance of the EWT method and the Empirical Mode Decomposition method.The result shows that:the EWT method can accurately analyze mechanical fault signal and the failure eigenvalues are evident.It can be effectively applied to rotating machinery failure.2 The gear and rolling bearing fault diagnosis based on improved Empirical Wavelet Transform and fast spectral kurtosis filteringImproved Empirical Wavelet Transform(IEWT) is a new kind of self-adaptation method of signal analysis. Combined this method with Fast Spectral Kurtosis filtering,we can achieved the purpose of gear and rolling bearing fault diagnosis. At first, the signal is decomposed using the IEWT method, and the two components which have the most obvious fault characteristics are extracted. Then, the reconstructed signal is filtered using the Fast Spectral Kurtosis filtering method. In the end, the filtered signal is analyzed by spectral envelope method and the fault features of signal are extracted. By analyzing the gear teeth broken and rolling bearing fault signals, it is indicated that the method based on IEWT is more distinctive than the method based on EMD. It can effectively identify the fault types of gear and rolling bearing.3 Roller Bearing Fault Diagnosis Based on IEWT, SVD and FCMA method for mechanical fault diagnosis based on IEWT, SVD and FCM is proposed.Firstly, fault signal is decomposed by IEWT, taking several component which have higher correlation coefficient to make up the initial feature vector matrix.Then decompose the initial feature vector matrix to obtain singular values to make up the singular value feature vector matrix.Finally, input the singular value feature vector matrix to the FCM to diagnosis the fault mode.The method is applied to recognise the rolling bearing fault pattern. Compared with the method based on EMD and FCM, it is more accurate and distinguishable. So it can be effectively applied to rotating machinery failure.