Research On Fault Diagnosis Of Marine Diesel Engine Lubrication System

In recent years,the level of ship intelligence and automation has been continuously improved,and stricter requirements have been imposed on ship reliability and safety.As the core equipment of ships,diesel engines play a vital role in ship safety.However,the structure of the diesel engine is complicated,there are many parts,and the body is in a high temperature and high pressure environment,so the possibility of failure is relatively high.The traditional fault diagnosis is mostly empirical method,thermal parameter method,oil analysis method,etc.These methods cannot accurately locate the fault diagnosis of the lubricating oil system,and take a long time,and some require special detection tools.This paper uses Bayesian noisy-OR/AND model to develop a fault diagnosis system,which can diagnose faults accurately and quickly,and can also provide maintenance measures for staff reference,can restore the operation of the equipment in the shortest time that the fault occurs The safe operation of ships is of great significance.In this paper,the Asian network is used as an experimental model to compare the influence of different elimination sequences on inference time.The main reason for the reasoning speed of variable elimination method is the construction of elimination order.At present,there are mainly four search methods:minimum degree,maximum potential,minimum missing edge and minimum increase complexity to construct the elimination order.Experiments have found that the minimum increase complexity search method is superior to other search methods,which can shorten the inference time and improve the inference efficiency.Establish a diagnosis model for the lubricating oil system of WARTSILA 6L34DF diesel engine.According to the type of failure of the lubrication system,the entire lubrication system can be divided into four sub-faults:abnormal oil pressure in the intake engine,abnormal temperature of the intake oil,excessive oil consumption rate and early failure of the oil.Through the analysis of the system,six fault trees were established,namely,too high oil pressure,too low oil pressure,too high temperature,too low temperature,too high oil consumption rate and early lubricating oil Failure.Using the method of transforming fault trees into Bayesian networks,the Bayesian network diagnosis model of the above six fault trees is constructed.Development of fault diagnosis software for diesel engine lubricating oil system.Using Visual Studio 2017 and SQL Server 2017 as the development environment,the diagnosis system is developed based on the C#language.Establish a knowledge base of the lubricating oil system in the database,and store the a priori probability in the corresponding data table;in Visual Studio 2017,the program realizes the entire fault diagnosis function.The system includes 4 menu bars.Finally,two fault examples are used to prove that the system can accurately and quickly diagnose the cause of the fault.The research shows that when combined with the observation information of the staff on the equipment,the fault diagnosis system can accurately and quickly locate the cause of the fault and give the corresponding maintenance strategy.In Bayesian network inference,the use of the minimum missing complexity search method can increase the inference speed of the variable elimination method,can shorten the running time of the system background calculation,and reduce the system's stall.The Bayesian network diagnosis model is superior to the existing diagnosis methods,and can truly diagnose the cause of the failure,helping the staff to quickly and accurately locate the failure.