APPLICATION OF ROBUST FAULT DETECTION METHODS TO F404 GAS TURBINE ENGINES

Abstract In this paper we examine the problem of the detection of 5% final nozzle miscliedulings in F404 gas turbine engines from transient data. We conclude that a model based approach using the transfer function between fuel flow and compressor outlet temperature does in fact provide reliable detection. We validate this claim on real data collected in test cell conditions by demonstrating that the method proposed both detects faults, and does not give false alarms. We also report the results obtained from modelling several other input-output pairs in the engine. This illustrates that the fuel flow to compressor outlet temperature pair is the best to choose for the purposes of failure detection.