The Method Of Autogram And Its Application Research In Rotating Machinery Key Parts Fault Diagnosis

As key transmission components,rolling bearing and gear play a very important role in mechanical systems and they have been widely used in industrial fields.Transmission components are often subjected to dynamic loads during operation due to the complexity of the environment,and the failures and damages will often occur during this process.Once the bearing or gear fails in operation,the normal operation of the equipment will be seriously affected,and even lead to accidents and casualties.Therefore,it is of great significance to study the early failure detection and diagnosis of rotating machinery.Resonance demodulation is a commonly used method of signal processing and fault diagnosis.The weak signal submerged in the background noise can be effectively extracted,but it is difficult to find a suitable demodulation frequency band in this method.Autogram is a recently proposed method for demodulation band selection.In this method,the maximal overlap discrete wavelet packet transform is utilized to divide the frequency domain of signal,and the unbiased autocorrelation kurtosis of the square envelope signal is used as the characteristic index to obtain the optimal frequency band.Autogram has been studied in detail and its related theory has also been improved in the paper funded by the National Natural Science Foundation of China(No.51975004).The main research work and innovations of the thesis are as follows:(1)A fault diagnosis method based on minimum entropy deconvolution in association with Autogram is proposed to overcome the issue that the characteristic frequency of frequency band filtered signal is not obvious due to the interference of strong background noise in Autogram method.In this method,first,the minimum entropy deconvolution is used to suppress noise,then the optimal demodulation frequency band of the denoised signal is extracted through Autogram.The proposed method is adopted to analyze the simulated and measured bearing signals,the results reveal that the fault characteristic frequency of the filtered signal can be effectively highlighted by the new method.(2)A fault diagnosis method based on order statistic filtering Autogram is proposed to overcome the shortcoming that Autogram cannot adaptively segment the frequency band according to the signal characteristics.In this method,order statistic filtering(OSF)and smoothing are used to improve the empirical wavelet transform,and based on that the frequency domain signal is adaptively segmented.The proposed method is verified by the simulated and measured data.The results show that the optimal demodulation frequency band can be accurately obtained by the proposed method according to the characteristics of signal itself,and the obtained fault characteristics of filtered signal are more obvious.(3)In the process of adaptive band segmentation,the influence of noise and irrelevant components on the diagnosis results cannot be suppressed by the improved Autogram method based on OSF.For this,first,an adaptive frequency band segmentation method-Maximum envelope-based Autogram(MEAutogram)method is proposed.The segmentation position can be adaptively determined by this method according to the characteristics of the frequency domain signal.Then,symplectic geometry mode decomposition(SGMD)is introduced to suppress noise and irrelevant components.And based on this,a fault diagnosis method based on MEAutogram and SGMD are proposed to improve the accuracy of detecting the optimal demodulation band.Finally,the superiority of the proposed method is verified through simulated and gear measured signal.(4)To solve the problem that the frequency band of the low amplitude area in the frequency domain cannot be effectively extracted by MEAutogram,the traverse symplectic correlationgram(TSCgram)is proposed.In this method,traversal segmentation model is applied to process signal of frequency domain at first,and a series of demodulation bands with different center frequency and bandwidth are obtained.Then,the symplectic correlation kurtosis was established to select the optimal frequency band.In this process,noise and irrelevant components are removed to improve the stability of the traditional kurtosis index.At last,the effectiveness and advantage of the proposed TSCgram method are verified through the analysis results of simulated and measured data.(5)Aiming at the issue that inaccuracy of selecting the optimal frequency band by single index,and the traditional envelope spectrum is easily disturbed by irrelevant components.A fault diagnosis method called Traverse index enhanced-gram(TIEgram)is proposed.The kurtosis,correlation coefficient and negative spectral entropy are weighted and fused by this method to obtain a new index,which is used as the basis for selecting the optimal demodulation frequency band.Then,to improve the effectiveness of envelope spectrum,an enhanced adaptive multi-scale weighted morphological filtering-based envelope spectrum is proposed,which has improved the diagnosis accuracy by removing irrelevant components.Finally,the proposed method also was applied to bearing fault diagnosis under non-stationary speed conditions and was compared with the existing fast kurtogram and Autogram methods.The results verified the feasibility and superiority of TIEgram in the fault diagnosis of non-stationary speed bearing.