Fault Early Recognition and Health Monitoring on Aeroengine Rotor System

Abstract

A new method is presented to improve the safety and reliability of an aeroengine rotor system (AERS) in early fault recognition and health monitoring, addressing the problems that fault samples are not sufficient and that early weak faults are not easy to recognize. First, the stochastic resonance system is used to refine the early weak feature signal so as to amplify the fault information. Second, the early fault features are extracted by multiresolution performance of wavelet packet analysis, and the fault characteristic vector can be constructed. Finally, the extracted eigenvectors import the support vector machine (SVM) classifier to carry on the fault recognition and then make use of intelligent monitor module to monitor early faults in AERS. Experimental results have shown that this method can not only early recognize faults in AERS but also monitor the fault online. It provides a new way to increase the safety of AERS and predict the sudden fault.