Bayesian belief networks for fault identification in aircraft gas turbine engines

Abstract

Describes the methodology for usage of Bayesian belief networks (BBNs) in fault detection for aircraft gas turbine engines. First, the basic theory of BBNs is discussed, followed by a discussion on the application of this theory to a specific engine. In particular, the selection of faults and the means by which operating regions for the BBN system are chosen are analyzed. This methodology is then illustrated using the GE CFM56-7 turbofan engine as an example.