Signal Diagnosis of Aircraft Engine Rotor Vibration Fault

Abstract

With the development of modern aviation technology, the structure of aircraft engine as the heart of the plane is more and more complicated. The problem of people’s concern is how to monitor engine condition and realize the fast fault diagnosis of engine. In this paper, the method of support vector machine (SVM) combing evidence theory decides whether a signal is a fault signal on the base of researching the engine vibration mechanism and the characteristics of rotor fault vibration signals. This method avoids the defects giving the single information though the traditional method which is unable to predict the tendency of the engine safety. Evidence theory meets weaker conditions than Bayesian probability theory. It can express "uncertain" and "unknown" directly. Therefore the paper makes information fusion in combination with evidence theory in different measuring points and different working conditions of engine. This method not only can identify small sample and nonlinear system, but also fuse information which gets more evidence samples effectively. At the same time, the posterior probability diagnosis results can predict the development trend of the fault accurately. Output in the form of probability can deepen the cognition about the present situation of the engine and better observe the safety of the trend of engine simultaneously. It can facilitate the management and protection in time.