Reliability Assessment Of Car Engine Systems Based On Dynamic Bayesian Network

The engine is the core of the car and it is the source of power that makes the car move.With more requirements for low-carbon environment and cars performance,car engine structure is becoming more and more complex.Moreover,the performance parameters of the car engine are increasing and the environmental conditions are more stringent,which puts forward higher requirements on car engine.Reliability is one of the important quality indexes of car engine.It is necessary and of great significance for the reasonable design,efficient test,safe operation and scientific maintenance plan of the engine to analyze and evaluate the engine reliability accurately.The traditional reliability analysis methods are usually under the binary assumption for each element and system.Each element and system are considered to be invalid or in good condition.However,there may be multi-state for some elements and systems in practical applications.Fault symptoms and fault cases do not share a one-to-one mapping,but have a strong randomness and uncertainty in the process of fault diagnosis.In addition,the existing reliability analysis methods can only be used in the static system reliability assessment.It is difficult to consider the influence of time related factors on system reliability.To cure the above problems,a new method based on Dynamic Bayesian Network is proposed to assess the car engines reliability.Firstly,the current research status of the system reliability and DBN were introduced.Secondly,the internal structure of the car engine was analyzed particularly and failure modes and effects analysis(FMEA)for each subsystem of the car engine was defined.Thirdly,the basic principle of Dynamic Bayesian Network was introduced to prove the feasibility and superiority of Dynamic Bayesian Network in dealing with uncertain problems in theory.And on the basis of FMEA table,the Dynamic Bayesian Network models of the common failure modes of the car engine were established.By using the advantages of Dynamic Bayesian Network in dealing with polymorphism and uncertainty,the reliability evaluation and fault diagnosis of the engine system were carried out using Matlab FullBNT toolbox.Lastly,the design and development of the reliability evaluation software of car engine system was completed by using Matlab language.Research results show that,compared with the existing reliability analysis methods,the method based on Dynamic Bayesian Network is able not only to compute the reliability indices of car engines at some point in the future but also to handle multiple states system,carry out the fault detection easily,and recognize the weakness of the system.