Statistical estimation of multiple faults in aircraft gas turbine engines

Abstract

This article presents estimation of multiple faults in aircraft gas-turbine engines, based on a statistical pattern recognition tool called symbolic dynamic filtering. The underlying concept is formulated by statistical analysis of evidences to estimate anomalies (i.e. deviations from the nominal values) in multiple critical parameters of the engine system; it also presents a framework for sensor information fusion. The fault estimation algorithm is validated on a numerical simulation test-bed that is built upon the NASA C-MAPSS model of a generic commercial aircraft engine.