Early fault classification identification and fault self-recovery on aero-engine

In order to increase the safety of aero-engine and solve the fault sample shortage in aero-engine fault diagnosis, we put forward a new method. This method can efficiently identify the early fault of aero-engine and it has function of fault self-recovery. Fault classification identification and fault self-recovery are adopted. It is combined with the Stochastic Resonance (SR), Wavelet Packet Analysis (WPA) and Support Vector Machine (SVM) and the fault self-recovery method of multi-modules cooperation is used. In this paper, the basic composition of system, the way of weak fault feature zoom, the extraction of fault feature vector, the principle of fault classification, the structure of multi-faults classifier and realization of fault self-recovery are studied. It provides a new technical way to increase ability on identification and protection for aero-engine early fault. The results show that the method can effectively identify the aero-engine early fault in shortage of fault samples numbers and it can realize the fault self-recovery of aero-engine.