Research On Fault Diagnosis System Of Marine Diesel Engine Based On Bayesian Network

Marine diesel engine is the key equipment of the ship power plant. Once the engine went wrong, it would cause serious effect on the operation of ship and huge economic losses, even endanger the personal safety. Uncertainty of fault causes as the complexity of marine diesel engine can brings about the failure in finding the fault location and elimination in time. So it is significant to introduce Bayesian Network into fault diagnosis of marine diesel engine, searching for the accurate fault location and elimination in time.Bayesian network is considered as one of the most efficient theoretically models on uncertain knowledge representation and inference field. Due to its solid theoretical foundation, efficient reasoning algorithm and learning ability, Bayesian network is used in the fault diagnosis of marine diesel engine to solve those problem of uncertainty.At the beginning of this paper, the common faults of marine diesel engine are diagnosed. According to the characteristics of Bayesian networks, the information involved in fault diagnosis of marine diesel engine is described as fault symptom nodes, operation nodes and fault cause nodes. The diagnosis Bayesian networks based on the structure of "fault symptom node-operation node-fault cause node" are set up. Then, under the requirement of fault diagnosis with gathering data as evidence information, the data acquisition and monitoring system of marine diesel engine based on PLC is established. The main function of the system is to collect the state information of diesel engine. On one hand, the information can be used as evidence for fault diagnosis. On the other hand, the information can be used to determine the current state of the diesel engine. At the end of the paper, the diagnosis system based on Bayesian network with background database established by Access is carried out with Visual C++6.0. The system includes knowledge base management module, network building module, inference and explanation module, learning module evidence collection module and interface module. In addition, the feasibility and effectiveness of the method in fault diagnosis of marine diesel engine were proved by practical experiments.