A Jet Engine Prognostic and Diagnostic System Based on Bayesian Classifier

Abstract

In this work, a predictive maintenance system is discussed as a modern solution for reducing downtimes in complex systems such as airplanes’ jet engines. The developed predictive maintenance system is based on prognostic and predictive algorithms which will be constructed by using machine learning techniques. Bayesian theorem is specially studied and employed for this purpose in this paper. The design and implementation of a Naïve Bayesian classifier will be presented to demonstrate and challenge the practicality of the method. A turbofan jet engine health check system was chosen as a complex and live industrial testbed example. We also demonstrate that the system in question has a high entropy and despite this, the Bayesian approach is sufficient enough to eliminate the critical errors as well as maintain a satisfactory overall accuracy.