Fuel System In Certain Aero-Engine Failure Diagnosis Based On Bayesian Network (BN)

With the rapid development of modern aviation transport, the accuracy and comprehensiveness of fault diagnosis in civil aviation are getting people's attention.Aero-engines are the direct power supply unit of an aircraft, playing an important role in deciding wether it is safe during aircraft operating. Meanwhile, the fuel system is the unit of energy supply in aero-engine, ensuring the operation of aero-engine is safe, so fault diagnosis and repairt of aero-engine fuel system is vital. Which lies on the key issues according to fault diagnosis in the aero-engine fuel system is how to accurately, quickly and comprehensively determine the correct position of component at which is failure occurred. In this research, the Bayesian Network fault diagnosis method is used for in-depth analysising in CFM56-5B engine. And the troubleshooting cost function is introduced to analyze the diagnostic costs,then the type of aero-engine fault diagnosis decisions are taken.The article firstly presents CFM56-5B aero-engine fuel system with simple structures and principles, and then parts of common faults in aero-engine fuel system through historical information and experience are introduced; Secondly,probability theory and Bayesian theory are introduced. And the Bayesian fault diagnosis model is made through establishing of a Bayesian network topology, and then the diagnostic Bayesian Network topology of aero-engine fuel system is established through the mechanism and symptoms of failure with analyzing its qualitative and quantitative information; After that, the troubleshooting cost function is introduced into troubleshooting fault diagnosis decision. The consideration of the safety indicators and economic indicators of fault diagnosis should be made before ensuring that Bayesian probability of failure have been obtained and the data are available, after that,the system troubleshooting steps should be reasonably ranked, so that the steps which possess the relatively high probability and the relatively low-cost diagnostic diagnosis relatively should be made sure to be priority.Finally, the analysis application is achieved through a specific failure of the CFM56-5B aero-engine fuel system with fault diagnosis and fault diagnosis decision; The special amended entropy method is used during the establishing of specific troubleshooting cost function, using the actual time and cost values to determine the final cost of the diagnosis, andultimately the diagnostic decisions based on Bayesian Network fault diagnosis are determined,and the Bayesian Network diagnostic procedure is compared with TSM(Trouble Shooting Manual) to confirm the superiority of fault diagnosis method decision in aero-engine fuel system based on Bayesian network with the standard of troubleshooting cost function through TOPSIS comprehensive evaluation.