Engine Reliability Prediction Based On Fault Tree And Bayesian Network

Reliability research has been more and more widely used in engines,and reliability is one of the most important indicators of engines.These studies are based on to failure data which are obtained during working time,and by using computer aided technology the distribution can be summarized,thus to improve the reliability of the product.The database based on SQL Server has been established,at the same time the function of database management system has been analyzed to build the fault tree analysis module,to complete the user-oriented operation interface,and to deepen the design of the BN module which could further optimize the background program and realize the function of the reliability prediction.During the reliability analysis of automotive diesel engine system,FTA is applied in the given example of fault tree model.Based on the functional module that has been developed,the fault locations,failure modes,fault causes and failure criticality have been pointed out.The example regards "diesel engine doesn't work" as the top event,according to the model,crank connecting rod mechanism,fuel supply system,gas distribution system,cooling lubrication system and inlet exhaust pressurization system have been included.Based on the FT,the BN method is proposed,and this process is demonstrated with the example of "engine failure".Through calculation of the failure probability in the fault tree,the relationships between the different levels in the events are expressed.By developing the interface of Bayesian network model,failure correlations of different subsystem parts have been found out,which could roughly guide the range of reliability prediction.With the help of computer aided technology,the reliability fuzzy prediction method of the engine system is used in supercharging system as an example to verify the feasibility.The two sub-systems Crank connecting rod mechanism and the cooling lubrication system have been compared,suitable triangular fuzzy numbers are selected for fuzzy range,by comparing the results of reliability prediction model with BN model,the reliability database system module is proved to be appropriate.