Novel Synchroextracting Transform Method And Its Application In Mechanical Fault Diagnosis

This dissertation was supported by the National Natural Science Foundation of China(Grant No.52075236,51675258),and the Natural Science Foundation of Jiangxi Province,China(Grant No.20212ACB202005),Mine Intelligent Equipment and Technology Anhui Provincial Key Laboratory Open Fund Key Project(Grant No.201901002),A key project of the Shaanxi Provincial Key Laboratory of Intelligent Monitoring of Mine Electromechanical Equipment Intelligent Monitoring Open Fund(Grant No.SKL-MEE-IM201901),Equipping Pre-Research Projects(Grant No.6142003190210).Based on the deficiencies of Synchroextracting transform method(SET)and its application in mechanical fault diagnosis,the project put forward some new improved Synchroextracting transform methods and applied to mechanical fault diagnosis.At meanwhile,a comparative study with the traditional SET method is conducted.The simulation and experiment results showed that these methods were effective,and the proposed methods are somewhat innovative.The main contents of this paper include the following aspects:1.To address the weaknesses of the traditional time-frequency method of processing signals to acquire time-frequency features that are ambiguous,time-frequency energy is not aggregated and there is insufficient feature extraction when dealing with weak signal faults.Based on the high time-frequency resolution capability of the Synchroextracting transform method,the SET method was applied into the mechanical fault diagnosis of rolling bearing,A comparison wth the classical short-time Fourier transform and Synchrosqueezed short-time Fourier transform time-frequency analysis methods is conducted in terms of time-frequency accuracy and anti-interference,among other things,and the results of the simulation processing show that,when setting the same window width and length,the SET method has the superior performance in time-frequency resolution and the advantage in interference immunity.Finally,the SET method was applied to the feature extraction of weak faults pf rolling bearings.The SET method can effectively extract the weak features of rolling bearings,and compared with the classical short-time Fourier transform method and the Synchrosqueezed short-time Fourier transform method,the SET method is more capable of revealing the characteristic frequency of the weak bearing fault.2.Empirical wavelet transform(EWT)has the advantage of extracting AF-FM components with tightly supported Fourier spectrum,therefore it can extract frequency characteristic components of each component of multi-component signal.Combining this method with the advantage of high-precision time-frequency of the traditional SET method,a mechanical fault diagnosis method based on empirical wavelet transformSynchroextracting transform(referred to as EWT-SET method)is proposed,then this proposed method is compared and analyzed with the classical SET method in terms of anti-interference and ability to handle multicomponent signals.The anti-aliasing performance of the proposed EWT-SET method in processing multi-component FM and AM signals is verified by simulation analysis.Finally,the proposed method is applied to the fault diagnosis of rotors with different degrees of rubbing faults.The results further verify the superiority of the proposed method.3.When dealing with multi-component non-stationary complex signals for the traditional SET method,the instantaneous frequency difference of adjacent components should be greater than twice the frequency support limit of the set Gaussian window function,otherwise frequency aliasing is inevitable,thus making the SET method limited in processing multi-component FM signals.Based on this deficiency,the variational nonlinear chip mode decomposition(VNCMD)is combined with SET in this paper,then a fault diagnosis method of VNCMD-SET is proposed.In the proposed method,the nonlinear FM modal decomposition adopts the demodulation means,combined with the joint optimization scheme of variable modal decomposition to deal with the multi-component signals whose component frequency components are close to each other or even intertwined,and the decomposed signal components are analyzed and processed by the SET method.And the proposed method is compared with the classical SET method in terms of time-frequency resolution,anti-interference and anti-frequency aliasing.The simulation results show that the VNCMD-SET method has high time-frequency resolution,and overcomes the problem of frequency overlap when the traditional SET method process the multi-component signals.Finally,the VNCMD-SET method was applied into the bearing outer ring fault diagnosis of two fault modes of weak damage and serious damage,and made a comparation with the traditional SET method,the experimental results further verify the effectiveness of the VNCMD-SET method.4.Based on the the shortcomings of traditional Synchroextracting transform(SET)method or FM signals and the occurrence of discontinuities in time-frequency characteristics under strong noisy environments,this paper constructs a novel rotation synchroextracting chirplet transformation(RSECT)under the traditional SET framework.The proposed method retains the advantage of fitting the time-frequency characteristics of the original signal of the generalised linear Chirplet transform,and retains the high precision time-frequency analysis ability of SET.The simulation results show that the proposed method is obviously superior to the generalized linear Chirplet transform(GLCT)and the traditional SET method.The RSECT method can adaptively process frequency modulation(FM)and amplitude modulation(AM)signals and multicomponent signal with crossed frequency components,and has high time-frequency analysis ability and anti-interference ability.Finally,the proposed method is applied to mechanical fault diagnosis,And the experimental results further verify the effectiveness of the proposed method,and the RSECT method can effectively extract the characteristic frequency of fault signal.In order to effectively evaluate the time-frequency aggregation performance of the proposed rotation synchroextracting chirplet transformation method,the Rényi entropy index was used to effectively evaluate the proposed method.And compares RSECT method with the Rényi entropy index value of the generalized linear Chirplet transform and the traditional SET method,the results show that the proposed rotation synchroextracting chirplet transformation has the lowest Rényi entropy value under different noise levels,and the results show that the proposed rotation synchroextracting chirplet transformation method has better time-frequency aggregation performance.