Study On Electronic Fault Diagnosis Method And Application Of Automotive Gasoline Engine Based On FTA-FuBN

Because a single diagnosis method or algorithm is difficult to deal with the diagnosis of complex situations,the development of Bayesian Network(BN)model based on Fault Tree Analysis(FTA)logic structure has gradually become a hot spot in the field of fault research,fault tree and Bayesian network complement each other in modeling and quantitative calculation,and have strong applicability and engineering application value.Due to the closeness of the manufacturer’s data,it is rare to introduce the massive data of the Repair Case Knowledge(Re CK)system.Therefore,the current research on the development of Bayesian network based on fault tree(FTA-BN)in the field of automobile fault is still blank.Therefore,fitting the advantages of fault tree qualitative evaluation and Bayesian quantitative reasoning,solving the modeling problem of qualitative and quantitative analysis of automobile fault reasoning,and diagnosing vehicle faults quickly and efficiently can greatly improve the competitiveness and sustainable development ability of automobile manufacturers.In this paper,the whole cycle from no power to good operation of the gasoline engine is divided into three stages: starting power supply,failure to start and poor operation.Along the idea of "Research on automobile fault reasoning and diagnosis algorithm → research on three-stage fault FTA-FuBN(Fuzzy Bayesian)diagnosis method → construction of Hy BN(Hybrid Bayesian)intelligent diagnosis model",the intelligent diagnosis method and application of FTA-FuBN for automobile gasoline engine electrical fault are studied in detail.According to the characteristics of hierarchy,relevance and uncertainty of automobile fault,the complex equipment reasoning diagnosis algorithm of influence factor fault tree fuzzy Bayesian network is studied,including the rules of fault tree structure and parameter transformation into Bayesian network,the construction method of fuzzy Bayesian network structure mapping and fuzzy parameter assignment algorithm.The method and steps of Bayesian network fault reasoning diagnosis based on fault tree and machine learning are given.Aiming at the characteristics of limited reference information and few auxiliary means to collect information when the gasoline engine is not powered on,a multi fault symptom fuzzy Bayesian network starting power supply fault reasoning diagnosis method based on fault tree is studied.The mechanism of 47 fault nodes of automobile gasoline engine start-up power supply fault in one stage is explored,the FTA path of 5 significant fault influence factors is fitted,the optimized multi fault symptom start-up power supply fault tree model is constructed,and the fuzzy Bayesian network model of start-up power supply fault with 23 fault symptoms and 9 types of fault nodes is mapped.According to 3633 fault data sets of the Re CK system in recent five years,the fault model of poor engine operation is verified and solved.Aiming at the characteristics of limited fault symptoms,wide involved surface and complex traction surface when the gasoline engine can’t start,the reasoning diagnosis method of less fault symptoms fuzzy Bayesian network can’t start fault based on fault tree is studied.The causal logic mechanism of 28 fault nodes of the second stage failure of automobile gasoline engine is revealed;The FTA paths of six significant fault influence factors are extracted,the improved failure tree model with few fault symptoms and failure to start is constructed,and the fuzzy Bayesian network model of failure to start fault with three fault symptoms and 10 types of fault nodes is transformed and established.Aiming at the characteristics of large amount of fault data,complex interference relationship and working mechanism when the gasoline engine is not running well,the reasoning diagnosis method of multi fault node fuzzy Bayesian network based on fault tree is studied.The mechanism of 54 fault nodes of poor three-stage operation of automobile gasoline engine is found out;Integrating the FTA paths of seven significant fault impact factors,the improved multi fault node misoperation fault tree model is constructed,and the misoperation fault fuzzy Bayesian network DAG of four fault symptoms and 17 types of fault nodes is mapped.On the basis of FTA-FuBN reasoning diagnosis of three-stage faults in the whole operation cycle of automobile gasoline engine,and making full use of the advantages of hybrid modeling,Hy BN intelligent diagnosis model based on machine learning and expert experience is constructed.Based on the solution parameters of the sample set of the fault data source of the Rick system,the state of the fault node is predicted to verify the effectiveness and applicability of the Hy BN model of the three-stage fault of starting power supply,failure to start and poor operation,and the accuracy presents a certain difference.