Abstract:

Generally, there isn't a large number of fault samples in early aero-engine fault diagnosis and early weak fault signals is hard to identify. In order to solve this problem, a new method for early fault identification on an aero-engine rotor system (AERS) is proposed which is able to implement fault intelligent selfrecovery monitoring (FISRM). This method is based on the analysis of early fault characteristics of AERS, and it combines the support vector machine (SVM) with the stochastic resonance theory (SRT) and wavelet packet decomposition (WPD). First, the SRT is employed to zoom the early fault feature signals. Then, the multi-resolution analysis of the wavelet packet is used to extract early fault features. Finally, the early fault feature vector is inputted to the classifier, and the early fault of AERS can be identified and monitored by intelligent self-recovery. In this article, the structure of the intelligent diagnosis system, the method of extracting early fault characteristics, the construction of multi-fault classifier and the FISRM are analyzed. The results show that the proposed method can identify an early fault of AERS in small samples, and its algorithm is simple, the identification effect is satisfactory, and an early fault of AERS can be monitored by self-recovery.

Key words: aero-engine, rotors, early fault feature extracting, fault classification identification, fault intelligent self-recovery monitoring (FISRM)