Research On The Method For Fault Diagnosis Of Steam Discharge System

Nuclear power is one of important tendency of World Future Energy. As an importantmeasure to ensure the security and stability of nuclear power system, fault diagnosis hasbeen highly valued by the theory and professional field. But the existing intelligentdiagnosis methods some needs a large number of fault samples, some are lack of reasonableexplanation, some can’t diagnosis unknown fault, so they can’t meet the requirements ofsecurity and reliability of nuclear power plant.With steam discharge system as an example, a fault diagnosis model based onmultilevel flow model was proposed in this thesis, and the mission reliability in thecondition of common-mode failure has been researched.First of all, the principle and parameters of steam discharge system are analysed. Byusing the theory of multilevel flow model, the fault diagnosis model of steam dischargesystem is set up. By determining the spread rules of consequences, constructing the searchalgorithm of flow structure and using symbol directed graph method in the main alarmreasoning, the fault diagnosis model of steam discharge system is set up, and framework andprocess of diagnosis is designed. Through the example, the effectiveness of the diagnosismethod is verified.Secondly, by using multilevel flow model reliability algorithm, steam discharge systemreliability model is established. For the unit in a single failure and common-mode failure, byusing the Markov theory the fault state is researched. After solving and drawing the chart ofprobability distribution of common mode fault state, the mission reliability of each target iscalculated, and finally get the mission reliability of whole system.Finally, according to the analysis to the problems of human-computer interface indigital master control room and its influence to the fault diagnosis system and alarm system,the design steps of intelligent diagnosis system interface is determined. Based on task levelof nuclear power plant intelligent fault diagnosis system, this paper completes theman-machine function distribution, and uses ecological interface theory to improve displaymode of the parameter, and designs the interface of intelligent fault diagnosis system tomeet the human factor standard. Finally using eye movement equipment information collection efficiency of traditional interface and new interface diagnosis is compared, andthe results show that the new interface effect is good.