Research On Fault Diagnosis Of Low Temperature And High Speed Bearing Based On Noise-assisted Empirical Mode Decomposition

Low temperature and high speed rolling bearings are often disturbed by various background noises during operation.Fault features are often covered up by strong noise when there are early weak faults in the bearing,which makes fault features difficult to obtain.Therefore,it is of great significance to improve the safe and reliable operation of liquid rocket by using effective diagnostic techniques and methods,analyzing bearing vibration data,extracting fault features and identifying fault location.In this thesis,the"vibration signal of low temperature and high speed rolling bearing" is taken as the research object;the signal analysis method is Complete Ensemble Empirical Mode Decomposition with Adaptive Nosie(CEEMDAN);and Support Vector Machine(SVM)is used as classification and recognition method.The main research contents include signal denoising,feature extraction and bearing operation status recognition.Specifically as follows:(1)For the problem that the early weak faults of bearings are easily submerged by noise and the nonlinear non-stationary characteristic of bearing vibration signals,Empirical Mode Decomposition(EMD)is introduced and the Empirical Mode Decomposition interval threshold denoising method is studied.Aiming at the modal aliasing problem of EMD algorithm,an improved CEEMDAN interval threshold denoising method is proposed.CEEMDAN algorithm can effectively reduce the phenomenon of modal aliasing and improve the performance of the denoising method.Aiming at the deviation and continuity problems of soft and hard threshold functions,an improved threshold function which is suitable for interval threshold noise reduction is proposed.In addition,the proposed method only denoises the noise-dominated Intrinsic Mode Function in order to improve the computational efficiency,so the autocorrelation function is introduced as the criterion to distinguish whether Intrinsic Mode Function is noise-dominated or signal-dominated.The improved denoising method is applied to the determination signal,typical signal and bearing engineering data of Case Western Reserve University.The experimental results show that the improved denoising method can effectively improve the signal-to-noise ratio and highlight the fault characteristics of the signal under the conditions of low signal-to-noise ratio and high signal-to-noise ratio.(2)Based on CEEMDAN algorithm,it is found that the energy distribution of Intrinsic Mode Function can reflect the state type of signal,so an energy feature extraction method is introduced.In view of the limited number of fault samples of bearing test bench,SVM is adopted to classify and identify fault states,which can solve the problem of classification and recognition of small samples.An intelligent diagnosis method based on CEEMDAN and SVM is proposed.The proposed method is applied to the bearing engineering data of Case Western Reserve University;the experimental results show that the proposed method can accurately identify the running status of bearings.(3)The improved CEEMDAN interval threshold denoising method and the intelligent diagnosis method based on CEEMDAN and SVM which is proposed in this thesis are applied to the measured vibration data of low temperature and high speed rolling bearings.The experimental results show that the noise reduction method proposed in this thesis can effectively reduce the impact of noise and highlight the fault features.Based on the existing bearing datas,the intelligent diagnosis method proposed in the thesis can accurately identify the running state of the bearing.