Research On Fault Diagnosis In Nuclear Power Plants Based On Signed Directed Graph

Extremely high safety standards are required during nuclear power plants (NPPs)working. So it is critically important for process monitoring and fault detection and diagnosis(FDD). Through research on online FDD in NPPs. it helps operator timely to know the state of system to take the proper operation to prevent the accident further deterioration.Due to the complex system of NPPs, a mathematical model is hard to be established in FDD. Signed directed graph (SDG) can show the complex relationship between the parameters and has advantages of establishing model conveniently, inference flexible and so on. So SDG is adopt for FDD in this paper. Firstly, systematically SDG modeling method have been studied in this paper. Loss of coolant accident, steam generator tube rupture, the main steam line break, loss of feed water, station black - out are taken as example to set establish SDG model. Secondly, experts knowledge are achieved by the SDG model, because the same parameters have different influence on faults, the parameters weight are applied in FDD. But as a fact. process monitoring is the first step of SDG-based FDD. Additionally. the threshold in NPPs increased difficultly of FDD for incipient fault diagnosis. In order to solve these problems. threshold combined qualitative trend analysis are proposed for process monitoring in this paper. Finally, according to the results of diagnosis, "status matrix" and the breadth-first search are applied to achieve SDG online inference, it can explain the reason of faults.According to the SDG-based FDD design scheme. C# .net 4.0 and SQL Server 2008 are utilized to develop the FDD system for NPPs. The cases studies show that the scheme are feasibility. The system has the advantages of interface friendly. operation easily, and diagnosis rapidly.