Research On Bearing Fault Diagnosis Method Based On Morlet Wavelet And Scale Space

As one of the key components of rotating machinery,rolling bearings are widely used in various fields.Its running state will directly affect the operating performance of the whole machinery.Due to the harsh working environment and long-term bearing of alternating load,rolling bearings are extremely vulnerable to damage.At the same time,the vibration signals collected from bearings often contain a variety of interference components,which affect the accuracy of fault diagnosis.In order to improve the accuracy of bearing fault diagnosis,this dissertation has conducted in-depth research on Morlet wavelet and scale space theory,and proposed a fast and effective method for bearing fault diagnosis,this method can accurately find out the types of bearing faults and realize bearing fault diagnosis.The main contents of this dissertation are as follows:Firstly,this dissertation introduces the research status of fault diagnosis technology of rolling bearing at home and abroad,summarizes the current processing methods of rolling bearing vibration signal,and analyses the problems of each method in processing the signal.Also,the basic structure and common failure forms of rolling bearing are summarized,the vibration mechanism of rolling bearing is studied,then the calculation formula of fault characteristic frequency in different components of rolling bearing is presented.Secondly,the basic theory and properties of Wavelet Transform(WT)are introduced in detail,the influence of the change of the scaling factor a and the translation factor b on the time frequency window of the wavelet function is studied;the definition and function expression of Morlet wavelet are given,and the influence of shape factor?and center frequency f_c on the time and frequency domain waveforms of Morlet wavelet is discussed.Meanwhile,the filter range of the Morlet wavelet is derived by using shape factor and center frequency;based on the time domain convolution theorem and Parseval theorem,a fast calculation method of Morlet wavelet transform coefficients is derived.Morlet wavelet can be regarded as a band-pass filter,and its decomposition effect heavily depends on the center frequency and bandwidth,that is,the preset frequency band boundary.Scale Space can adaptively divide the frequency band boundary of the signal,according to this,the scale space is applied to divide the frequency band in bearing vibration signals.Scale Space(SS)theory is also introduced and discussed from three parts:continuous scale space,discrete scale space,and frequency band boundary.Based on these charateristics,a Morlet wavelet and Scale Space(SSM)bearing fault diagnosis method is proposed,and the validation of this method is verified by simulation signals.The results show that this method can detect the characteristic frequency of bearing failure,but when identifying the resonance frequency band in the bearing failure signal,there is an over-division problem in the scale space,thus losing the failure information.Finally,in order to solve the problem of scale space over division,a method of scale space optimization spectrum is proposed to divide the signal band boundary more accurately.Correlated Kurtosis(CK)is introduced,and Envelope Correlation Kurtosis(ECK)is used as the evaluation index to identify the optimal resonant frequency band of the fault signal.Based on these two indices,the Scale Space Optimal Spectral Envelope Correlated Kurtosis(SSOSECK)method is proposed.The effectiveness of this method is verified by simulation signals and measured data,also performance comparison has been made with Spectrum Kurtosis(SK).The results show that the method proposed in this dissertation can accurately identify the resonance frequency band of single and compound bearing fault,and realize bearing fault diagnosis.The performance of the diagnosis is significantly better than the conventional spectrum kurtosis method.