Fault Diagnosis Method Research On Intershaft Bearing Of Aero-Engine

Aero-engine is an important part of the aircraft, it directly determines the overall performance of the aircraft. Due to its long run in a variety of complex environment, the requirement of security and reliability of the aero engine is quite high. As the core part of the aero-engine, the intershaft bearing supports the high pressure rotor and the low pressure rotor. Once the failure of intershaft bearing, it will cause the increase of the rotor vibration, and even cause parking lock in the engine. Therefore, on the one hand, the research on the fault diagnosis method of the aero-engine intershaft bearing can ensure the stability of the aircraft in the course of operation; on the other hand, it can also reduce the economic loss.At present, the diagnosis of the intershaft bearing is mainly based on the analysis of the vibration signal. Different from general rolling bearing signal, the intershaft bearing signal contains a large amount of background noise. It is easily affected by the unbalance response of the surrounding high and low pressure rotor, and the effective signal is submerged by the long path transmission, so it is difficult to identify the frequency of failure.Embarks from the actual situation, the research on the intershaft bearing of aero-engine is carried out in this paper from the aspects of signal preprocessing, construction and optimization of feature set and intelligent pattern recognition. In the aspect of preprocessing, this paper introduces the C-C algorithm from the view of phase space reconstruction, and improves the algorithm. Aiming at the characteristics of the complexity of the intershaft bearing signal itself, the de-noising method based on phase space reconstruction is put forward. Firstly, the low dimensional acquisition signal is reconstructed into the high dimension phase space by the improved C-C algorithm, and then the main stream is extracted from the phase space, and finally the data is reconstructed by the main stream to achieve the effect of noise reduction.In the construction and optimization of feature parameter set, this paper firstly extracts the original feature set from the phase space by using singular value decomposition, and then carries on the reduction of the feature set. This paper uses the nonlinear manifold learning method LLE for reduction. According to the characteristics of the signal of intershaft bearing, the LLE algorithm is improved to NSLLE algorithm based on nonlinear monitoring distance. This improved method is helpful to mining the low dimensional structure of the data, and the dimension reduction effect is better than that of KPCA and other manifold learning algorithms. In addition, the definition of the geodesic distance based on the discriminant formula to find the nearest neighbor parameters in the NSLLE algorithm, to avoid the randomness of the choice and the uncertainty caused by the constant test; and using the Treelets transform to optimize the R-NSLLE parameters, we find the optimal intrinsic dimension in the dimension reduction process.Finally, in the aspect of intelligent pattern recognition, this paper chooses the ELM algorithm, which has the characteristics of short training time and strong generalization ability. In this paper, ELM is improved to Semi-Supervised ELM, which can be used to train and test the signal with labeled feature and unlabeled feature to improve the practicability of fault recognition. And the on-line ELM is studied, which can update the ELM network parameters at any time according to the new data, and make full use of the characteristics of historical data and new data to monitor the running state of the system in real time.In the whole research process, this paper deeply studies the theory of the algorithm, the simulation data and real experimental data are combined for analysis, and from the qualitative and quantitative point of view for further discuss. The experimental results fully prove the validity of the research method, and it has a certain significance to the fault diagnosis of the intershaft bearing.