Research On Sbloca Prediction Method Based On Information Flow

With the development of nuclear power industry,nuclear power energy has become an efficient,clean and economical energy.More and more countries and regions use it as a substitute for traditional energy sources,and the proportion of nuclear power generation has reached 16%of the world’s total power generation.However,the nuclear power system uses the energy generated by the nuclear reaction to generate electricity,and a large amount of radioactive materials will inevitably be produced during the operation of the nuclear power system.The issue of nuclear power safety is undoubtedly an important research field in the development of nuclear power.Once the nuclear power system has an accident,it will mean that it may pose a great threat to the surrounding environment and biological safety.Therefore,when a nuclear power system accident occurs,the sooner it is predicted to find the fault and deal with it in time,the possible risk of the accident can be reduced to a lower level.The Loss of Coolant Accident(LOCA)is one of the typical accidents in the nuclear power system.The occurrence of this accident will have a huge impact on the coolant system of the primary loop reactor of the nuclear power system.And with the development of the big data era,from the research of big data correlation gradually to the research of big data causality,industrial control systems are also developing towards intelligent control,Many data-driven failure prediction methods or methods combined with traditional predictions are used in the failure prediction of nuclear power systems.The LOCA prediction method based on information flow proposed in this paper is based on the causality of the physical variables in the nuclear power system,and uses the Liang-Kleeman information flow method to predict the loss of coolant accidents.The main research results have the following five points:(1)A cyber-physical causal flow model of nuclear power control system is proposed.Based on the reactor coolant system in the nuclear power system,Firstly,it is divided according to the structure,and then the structural theory and operating principle are analyzed separately,and the relevant physical variables are extracted.Finally,a cyber-physical causal flow model of the industrial control system corresponding to the reactor coolant system components is constructed to model,and the information and physical relationship within the corresponding system can be obtained,and the causal relationship between the internal variables can be fully expressed(2)A method of causality discovery is proposed.Through the analysis of the coolant loss accident,the causal relationship existing between the physical characteristics is obtained according to the changes of physical variables,and the key causal relationship that can produce sensitive changes in response to the fault is discovered,that is,the coolant volumethe water level of the regulator.(3)Research and system implementation of small break water loss accident prediction method based on information flow.Using the simulation data in PCTran AP1000,calculate the causal pair information flow between the coolant volume sensitive to the coolant loss accident and the water level of the regulator,through observation,it is found that the information flow when there is no fault is a curve with small fluctuations.When an accident breach occurs,the originally relatively stable information flow will jump immediately,and it is considered that the point in time when the jump occurs is the point in time when the coolant loss accident occurs.In addition,the small additional noise in the data does not affect the results.Finally,the applicability of the information flow to the prediction of coolant loss accidents is verified.According to the results of the experiment,it can be found that the coolant loss accident prediction method based on information flow can be applied to the smallest breach size of 0.05 cm2,and found the law of the jump degree of the information flow and the size of the breach,that is,the jump degree of the information flow is smaller,and the size of the breach has also occurred.Finally,a small break water loss accident prediction system based on information flow is realized.