Research On Fault Diagnosis In Primary Loops System Of Nuclear Power Plant Based On Petri-SDG Method

Nuclear energy has become the current energy development trend because of its cleanliness,economy,and efficiency.At the same time,nuclear power safety is particularly important for its development,research on nuclear power operation support technology based on condition monitoring and fault diagnosis has become indispensable to ensure the safe operation of nuclear power.The condition monitoring and fault diagnosis technology is studied for Fuqing NPP Unit2(M310 reactor type),and based on research,a fault diagnosis system is developed in this article.The main research contents are as follows:(1)Research on the basic theory and analysis methods of traditional Petri nets,combined with its development and application in the field of fault diagnosis,a fault diagnosis method based on fuzzy fault Petri,and an analysis method of the correlation matrix and state equations are proposed for the characteristics of the research object.(2)With the loss of coolant in the primary coolant system and the chemical and volume control system in the primary circuit system of the nuclear power plant,the leakage of the charging line,the leakage of the drain line,the accidental insertion of the control rod,the accidental lifting of the control rod,and the regulator The heater was turned on by mistake and the regulator was sprayed with seven kinds of faults.A fuzzy fault Petri net model was established,and the data of Fuqing full-range simulation simulator was used to verify the fault model.(3)The principle of k-Nearest Neighbor(kNN)algorithm classification is studied,the key parameters and calculation principles of the kNN classification algorithm applicable to the selection of seven kinds of fault parameter data are calculated and analyzed.Then train the kNN classification algorithm and classification effect is tested using test data.(4)Establish seven Signed Directed Graph(SDG)models of faults,use forward reasoning to evaluate the fault results,and the basic process of fault verification are give.(5)Based on the above research,C # is used to develop a fault diagnosis system that includes functions such as state monitoring based on kNN classification algorithm,fault diagnosis of fuzzy fault Petri nets,and use the simulator data to test the function of the fault diagnosis system.Through test verification,the above research content can realize the effective diagnosis of seven kinds of faults in the nuclear power plant’s primary loops system,including the main coolant system and the chemical and volume control systems.