Research On Fault Diagnosis Technology Of Aircraft System Based On Fault Propagation Mechanism And Petri Network

System fault diagnosis is an important means to ensure the safety and reliability of the aircraft,and plays an important role throughout the life of the aircraft.With the development of technology,the structural complexity of the aircraft system is increasing.How to prevent system fault and quickly locate the fault source have become an urgent problem to be solved.Considering the hierarchical complexity,propagation,relevance,uncertainty of the system fault,this paper studies fault propagation mechanism and fault diagnosis technology of the aircraft system,and selects a certain type of aircraft pneumatic system as a typical application object.The main contents of this paper are as follows:Firstly,aiming at the inherent attribute of the aircraft system failure mode and the characteristics of the hierarchical transmission,the fault propagation mechanism of the aircraft system is analyzed by the method of functional fault analysis and FMMEA analysis.A hierarchical functional structure tree is established for the corresponding subsystems of the civil aircraft pneumatic system,which lays the foundation for the subsequent modeling and diagnosis of Petri net.Secondly,an improved fault diagnosis model-CFFPN is proposed.The fuzzy Petri net is combined with the fault Petri net effectively based on the traditional Petri net theory and introduces token,place,transition colored rule and fuzzy production rule.The CFFPN model is applied to the fault diagnosis of aircraft system,which can not only reflect the characteristics of fault propagation and fault location,but also can determine the maximum propagation path.Thirdly,the fault diagnosis algorithm based on CFFPN model is studied.When there is no fault in the system,the model uses the transition ignition discrimination matrix and the MYCIN confidence method to deduce the fault phenomena.When the system fails,the minimum cut set theory is used in the process of reverse reasoning to diagnose the fault,and it can be used to trace the fault source leading to the fault phenomenon and assist the maintenance decision.Furthermore,a parameter optimization method of CFFPN model based on BP error back propagation algorithm is proposed.This method improves the oscillation and convergence of the learning process by introducing the momentum term and adaptively adjusting the learning rate so that the CFFPN model can have the same self-learning function as the BP neural network.The feasibility of the algorithm is demonstrated by examples.Finally,a fault diagnosis system based on CFFPN model is developed.Through the visual GUI graphical modeling tools,the element database of place,transition,arcs and Token is set.The fault diagnosis model is stored in XML format.In addition,the system designs a fault diagnosis analysis module to realize the state assessment of fault propagation and the minimum cut set analysis.