Fault Diagnosis Of Bearing-rotor System Based On Sparse Representation

Fault diagnosis of bearing-rotor system is of great significance to ensure the safe operation of aeroengine.Using the collected vibration signal for fault diagnosis is the most effective way at present,which has the advantages of accuracy,convenience and quickness.In this paper,the vibration signal of bearing-rotor system is taken as the main research object.Aiming at the problem of fault feature extraction and fault type recognition of bearing rotor system,the fault feature extraction technology based on wavelet transform,wavelet packet transform,Hilbert Huang transform and sparse representation is studied.Mechanical fault recognition method based on support vector machine is studied.The fault diagnosis method of bearing-rotor system based on wavelet transform is studied.Taking the vibration signal of bearing-rotor system as the research object,the feasibility of using relative wavelet energy and sample entropy as feature quantity is analyzed.The experiment of fault classification based on support vector machine verifies the effectiveness of wavelet transform and wavelet packet transform for fault diagnosis of bearing-rotor system,and further analyzes the influence of penalty factor and kernel function width on the recognition results,and optimizes the classifier.In the framework of sparse representation theory,a fault diagnosis method of bearing rotor system based on the Laplace wavelet and OMP algorithm is proposed.The Laplace wavelet is similar to the impulse waveform of bearing fault signal,while OMP algorithm has simple structure and strong feasibility.Laplace wavelet correlation filtering method is used to construct dictionary,OMP algorithm is used to reconstruct signal,Hilbert transform is used to obtain envelope spectrum,fault feature frequency is extracted,and support vector machine is used to identify fault type.The simulation results show that the method has the ability to identify the time when the transient impact component is generated and the ability to resist strong noise.The experimental results show that the feature extraction effect of this method is due to the wavelet packet transform,and the recognition rate is high.Aiming at the low computational efficiency of OMP algorithm,a fault feature extraction method of bearing-rotor system based on SWOMP algorithm is proposed.In the process of atom selection,SWOMP algorithm reduces the number of iterations by setting a threshold,and the threshold is determined by the weak selection strategy,which enhances its rationality.The simulation results show that compared with the traditional time-frequency analysis method,the time-frequency map constructed by this method has higher time-frequency resolution;Compared with other greedy algorithms,SWOMP algorithm has faster calculation speed and higher reconstruction success rate,especially when the signal complexity is high.The effectiveness of the proposed method for fault feature extraction of bearing rotor system is verified by the analysis of bearing rotor system experiment.