Research On Fault Diagnosis Of Rotating Machinery Based On Singular Spectrum Decomposition Theory

Rolling element bearings(REB)which are the research objects of the dissertation are widely used in rotating machinery.Unexpected fault occurring in bearings may lead to fatal breakdown of machines and result in enormous economic losses in industry.There-fore,it is significant to accurately detect the early defect of bearings.Vibration signals are widely used for bearing fault diagnosis and the key issue is extracting the features of bearing fault.However,these features are not easy to extract due to the heavy background noise and other source of vibration in the rotating system such as gears,couplers,etc.To enhance fault feature extraction,a series of studies are carried out in this dissertation,which can improve the weak fault feature extraction ability in strong background noise environment.The main contents are as follows:1.Some background knowledge of the selected topic is discussed from the viewpoint of theoretical analysis and engineering application.The process of condition monitoring and fault diagnosis of rolling element bearings is introduced and the development of bearing fault feature extraction methods based on vibration signal analysis is reviewed.The research contents of the dissertation are presented based on the above analysis and summary.2.Aiming at the problem that the fault characteristics of rolling bearings are difficult to detect under the influence of strong background noise and interference signal,a new adaptive signal processing method based on singular spectrum decomposition(SSD)is proposed.This method can divide non-stationary signal into several single-component signals from high frequency to low frequency by constructing a trajectory matrix,and adaptively select embedding dimension length.The singular spectral components with obvious impact characteristics and maximum kurtosis are selected according to the kurtosis criterion.The singular spectral components are separated into envelope components and pure frequency components with the empirical AM-FM decomposition method,and the instantaneous frequencies are calculated by using energy operators.To solve the problem of endpoint effect in the instantaneous frequency,a data extension method using support vector regression is proposed.The experimental results show that the instantaneous frequency extraction can be better applied to the vibration signal of rotating machinery in the background of strong noise,and the instantaneous characteristics can be extracted.3.This paper researches the synchrosqueezing wavelet transform algorithm and introduces it into the rolling bearing fault diagnosis.Firstly,I build a rolling bearing fault experimental platform to collect the rolling bearing fault datas,analyze it using wavelet transform,and then use synchrosqueezing transform to compress the coefficients after wavelet transform.Compared with short-time Fourier transform and wavelet transform,the experimental results show that the synchrosqueezing wavelet transform can effectively extract the characteristic frequency of rolling bearing.4.Through the fault diagnosis experimental test system for gearbox and rolling bearing,including rotor comprehensive fault test platform(HZXT-004),rotor power control system(HZXT-003),data acquisition instrument(HADM-I),acceleration sensor(HD-YD-216)collects the vibration data at the time of failure and performs processing verification.The experimental results show that the instantaneous frequency extraction of singular spectrum decomposition in the background of strong noise can be better applied to the vibration signals of key components of rotating machinery and extract the instantaneous characteristics.