Gas-turbine diagnostics using artificial neural-networks for a high bypass ratiomilitary turbofan engine

摘要

The Tristar aircraft, operated by the Royal Air Force, fly many thousands ofhours per year in the transport and air-to-air refuelling roles. A large amountof engine data is recorded for each of the Rolls-Royce RB211-524B4 engines: itis used to aid the maintenance process. Data are also generated during test-bedengine ground-runs after repair and overhaul. In order to use recorded enginedata more effectively, this paper assesses the feasibility of a pro-activeengine diagnostic-tool using artificial neural networks (ANNs). Engine-healthmonitoring is described and the theory behind an ANN is described. An enginediagnostic structure is proposed using several ANNs. The top level distinguishesbetween single-component faults (SCFs) and double-component faults (DCFs). Themiddle-level class includes components, or component pairs, which are faulty.The bottom level estimates the values of the engine-independent parameters, foreach engine component, based on a set of engine data using dependent parameters.The DCF results presented in this paper illustrate the potential for ANNs asdiagnostic tools. However, there are also a number of features of ANNapplications that are user-defined: ANN designs; the number of training epochsused; the training function employed; the method of performance assessment; andthe degree of deterioration for each engine-component's performance parameter.