Abstract:Rapid and accurate diagnosis of nuclear accident is the primary link of nuclear emergency. Signed directed graph SDG has good completeness as a common method for fault diagnosis, but the traditional SDG model has low diagnosis efficiency. As a representation method of causal relationship, information flow can be calculated and analyzed from a quantitative point of view, and the application of information flow to causal fault diagnosis can improve the diagnosis efficiency. In this paper, the information flow is introduced to improve the SDG model and the info-flow(information flow) SDG model is constructed, and a fault location method based on information flow is researched by using this model. In this method, the Coefficient of Variation is introduced to compare the jump amplitude of the information flow of the key causal pairs on each compatible path, so as to obtain the fault source with the highest probability of occurrence. In this paper, a simulation experiment is carried out in the PCTran AP1000 system by taking the coolant loss accident (LOCA)in the primary loop system of a nuclear reactor as an experimental case.

Research on Nuclear Power Fault Location Method Based on Info-Flow SDG Model