The Research Of Aero-Engine Intershaft Bearing Fault Diagnosis Based On Local Mean Decomposition

Intershaft Bearing is the key component of the aero-engine, it works at the bad condition of high speed, high temperature, high pressure, overloaded. The early weak fault of the intershaft bearing will develop into uncontrollable bearing damage phenomenon, then the rotor system will be stuck and the aero-engine will stop work, and it may cause aviation accidents, so the research of aero-engine intershaft bearing fault diagnosis has great realistic significance. The aero-engine intershaft bearing dual-rotor test stand is the important carrier for the research of intershaft bearing. The test stand and the aero-engine have the common feature:the transmission path of the vibration signal is complex, the SNR is very low because of the existence of the environmental noise and other interference components, and the intershaft bearing fault signals often exhibit non-stationary, nonlinear features, so it is difficult to extract fault characteristic from the vibration signal. For this issue, the article carries out the following research content:1. Discuss the developing situation of aero-engine intershaft bearing fault diagnosis, describe the application of the local mean decomposition in fault diagnosis field. Develope the fault test program about the test stand and do a lot of pitting, scratches fault test against the intershaft bearing NU1013 and develop the acquisition system for the test stand.2. One fault diagnosis method combining local mean decomposition and manifold learning is proposed in this paper against the difficulty of fault feature extraction caused by the feature of non-stationary, non-linear, low SNR. First, the source signal is decomposed into a series of components utilizing local mean decomposition, the high-dimensional spatial data sets are constructed; second, low-dimensional manifolds compositions are reconstructed using the manifold learning methods for high-dimensional data sets, noise sub-components are abandon, the attractor components which represent the essential characteristics of the signal are withhold; finally, the noise reduction reconstructed signal is obtained by adding the attractor components. The effectiveness of the method is verified by analysing the test signal.3. One intershaft bearing fast independent component analysis method based on the local mean decomposition is researched in this paper against the underdetermined blind source separation problem that the number of effective measurement channels is less than the number of the vibration source. Firstly, the source signal is decomposed by local mean decomposition. Second, the number of vibration source'n'is determined by calculating eigenvalues cumulative contribution rate, the component signal whose correlation coefficients are the first n maximum are selected as the optimal observation signals. Finally, the optimal observation signals are utilized as the input of fast independent component analysis, the best estimated source signals are received. The feasibility of the method is verified by the simulation and the test signal analysis.4. Develop the aero-engine intershaft bearing dual-rotor test stand fault diagnosis system utilizing labview and matlab combining rotor and intershaft bearing failure mechanism. The common signal processing algorithms and the fault diagnosis method described above are integrated into the system. The applicability of the system is verified through a large number of examples of signal analysis.