Method for constructing fault model of nuclear power plant reactor coolant system, fault diagnosis method and related device

Abstract

The application discloses a nuclear power plant reactor coolant system fault model construction method, a fault diagnosis method and related devices, and relates to the technical field of nuclear power plant safety monitoring. The method comprises the following steps: obtaining historical operation data of a nuclear power plant reactor coolant system; determining evidence layer nodes and fault layer nodes of a dynamic Bayesian network model and a directed acyclic graph structure based on the historical operation data; calculating fuzzy observation data corresponding to the evidence layer nodes based on interval type-2 fuzzy sets and historical operation data corresponding to the evidence layer nodes; iteratively calculating a conditional probability table between the evidence layer nodes and the fault layer nodes based on the fuzzy observation data, obtaining an optimal conditional probability table, and constructing a fuzzy dynamic Bayesian network model based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table. The application has good robustness and adaptability in nuclear power plant reactor coolant system fault diagnosis, and has high fault diagnosis precision.

Description

Technical Field

[0001] This application relates to the field of nuclear power plant safety monitoring technology, and in particular to a method for constructing a fault model, a fault diagnosis method, and related devices for a nuclear power plant reactor coolant system. Background Technology

[0002] Nuclear power plants are highly uncertain and complex systems comprised of interconnected and spatially dispersed structures, components, and human factors. The Reactor Coolant System (RCS), as its core subsystem, directly impacts the overall safety of the nuclear power plant. Therefore, early detection of RCS faults and identification of their type, location, and severity are crucial for their safe operation.

[0003] Existing fault diagnosis methods can be broadly categorized into model-based methods and data-driven methods. Model-based methods rely on precise physical equations to construct fault thresholds, such as a coolant pressure drop exceeding 15% triggering a coolant loss alarm. However, in multi-fault coupled scenarios within the RCS (Real-Time Cross-Section) system, they suffer from large modeling errors and poor adaptability. Data-driven methods employ traditional machine learning algorithms (such as random forests and support vector machines). While capable of handling massive amounts of sensor data, these algorithms, being "black box" models, lack interpretability and cannot provide causal chains for fault diagnosis.

[0004] Bayesian networks (BNs) provide a probabilistic "white-box" model, but traditional BN models have limitations when applied to complex systems that change over time. Dynamic Bayesian networks (DBNs) are a method that extends BNs to time-series frameworks. By discretizing time and incorporating previous states into the reasoning process under current conditions, they can effectively model dynamic systems. However, DBNs rely on precise probabilistic inputs and cannot handle uncertainties such as sensor noise and missing data in the RCS (Restricted Series Cross-Section); topology construction depends on expert experience, leading to insufficient model generalization. Summary of the Invention

[0005] The purpose of this application is to provide a method for constructing a fault model, a fault diagnosis method, and related devices for a nuclear power plant reactor coolant system, which can improve the accuracy of fault diagnosis for a nuclear power plant reactor coolant system.

[0006] To achieve the above objectives, this application provides the following solution:

[0007] Firstly, this application provides a method for constructing a fault model for a nuclear power plant reactor coolant system, including:

[0008] Acquire historical operating data of the reactor coolant system of a nuclear power plant; the historical operating data includes historical normal operation data and historical fault operation data; the historical normal operation data is multi-parameter time-series data under normal operating conditions; the historical fault operation data includes various fault conditions and multi-parameter time-series data corresponding to each fault condition;

[0009] Based on the historical operating data, multiple evidence layer nodes and multiple fault layer nodes of the dynamic Bayesian network model are determined.

[0010] Calculate the correlation coefficient between the evidence layer nodes, the first causal relationship between the evidence layer nodes, and the second causal relationship between the evidence layer nodes and the fault layer nodes, and determine the directed acyclic graph structure of the dynamic Bayesian network model based on the evidence layer nodes, the fault layer nodes, the correlation coefficient, the first causal relationship, and the second causal relationship;

[0011] Based on the interval type II fuzzy set and the historical running data corresponding to all the evidence layer nodes, calculate the fuzzy observation data values ​​corresponding to all the evidence layer nodes in the dynamic Bayesian network model.

[0012] Based on the fuzzy observation data values, the conditional probability table between the evidence layer nodes and the fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed. The fuzzy dynamic Bayesian network model is used to determine the fault diagnosis results of the reactor coolant system of a nuclear power plant.

[0013] Optionally, the correlation coefficients between the evidence layer nodes, the first causal relationship between the evidence layer nodes, and the second causal relationship between the evidence layer nodes and the fault layer nodes are calculated. The directed acyclic graph structure of the dynamic Bayesian network model is then determined based on the evidence layer nodes, the fault layer nodes, the correlation coefficients, the first causal relationship, and the second causal relationship. Specifically, this includes:

[0014] Calculate the Kendall correlation coefficient between the nodes in the evidence layer, and generate an undirected acyclic graph structure of the dynamic Bayesian network model based on the Kendall correlation coefficient, the nodes in the evidence layer, the nodes in the fault layer, and the R.T. Copula model.

[0015] Calculate the first causal relationship between the nodes of the evidence layer;

[0016] Calculate the second causal relationship between the evidence layer node and the fault layer node;

[0017] The directed acyclic graph structure of the dynamic Bayesian network model is determined based on the undirected acyclic graph structure, the first causal relationship, and the second causal relationship.

[0018] Optionally, based on the interval type-2 fuzzy set and the historical running data corresponding to all the evidence layer nodes, the fuzzy observation data values ​​corresponding to all the evidence layer nodes in the dynamic Bayesian network model are calculated, specifically including:

[0019] For each of the evidence layer nodes, the historical operation data corresponding to the evidence layer node is evenly divided into multiple sub-intervals;

[0020] For each of the sub-intervals, Latin hypercube sampling is performed on the sub-interval to obtain the sample value corresponding to the sub-interval;

[0021] For each sample value corresponding to the sub-interval, based on the interval type II fuzzy set, the fuzzy observation data of the sample value corresponding to the sub-interval is calculated to obtain the fuzzy observation data of the sample value corresponding to each sub-interval;

[0022] The fuzzy observation data of the sample values ​​corresponding to all the sub-intervals are determined as the fuzzy observation data values ​​corresponding to the evidence layer nodes, thus obtaining the fuzzy observation data values ​​corresponding to all the evidence layer nodes.

[0023] Optionally, based on the fuzzy observation data values, the conditional probability table between the evidence layer nodes and the fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Then, based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed, specifically including:

[0024] Based on the fuzzy observation data values, the expectation-maximization algorithm is used to iteratively calculate the conditional probability table between the evidence layer nodes and the fault layer nodes to obtain the initial conditional probability table.

[0025] Based on the initial conditional probability table, a set of candidate conditional probability tables is randomly generated;

[0026] Based on the whale optimization algorithm and the set of candidate conditional probability tables, the optimal conditional probability table is obtained;

[0027] Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, the fuzzy dynamic Bayesian network model is constructed.

[0028] Optionally, based on the historical operational data, multiple evidence layer nodes and multiple fault layer nodes of the dynamic Bayesian network model are determined, specifically including:

[0029] Using the random forest algorithm, the operating conditions and several variables that affect the fault diagnosis of the reactor coolant system in nuclear power plants under each operating condition are selected from the historical operating data.

[0030] Set the variables as evidence layer nodes and the operating conditions as fault layer nodes;

[0031] In this system, one variable is set as one evidence layer node, and different variables are set as different evidence layer nodes. One working condition is set as one fault layer node, and different working conditions are set as different fault layer nodes.

[0032] Secondly, this application provides a method for diagnosing faults in a nuclear power plant reactor coolant system, including:

[0033] Real-time acquisition of multi-parameter time-series data of the reactor coolant system in nuclear power plants;

[0034] The real-time collected multi-parameter time-series data is input into the fuzzy dynamic Bayesian network model to obtain the fault diagnosis results of the nuclear power plant reactor coolant system.

[0035] Optionally, the method for diagnosing faults in the nuclear power plant reactor coolant system further includes:

[0036] Based on the fault diagnosis results of the nuclear power plant reactor coolant system and the fuzzy dynamic Bayesian network model, the fault layer nodes corresponding to the multi-parameter time-series data of the nuclear power plant reactor coolant system collected in real time are determined.

[0037] Based on the fault layer nodes corresponding to the nuclear power plant reactor coolant system and the fuzzy dynamic Bayesian network model, the evidence layer nodes corresponding to the multi-parameter time-series data of the nuclear power plant reactor coolant system collected in real time are determined.

[0038] Thirdly, this application provides a computer device, including: a memory, a processor, and a computer program stored in the memory and capable of running on the processor, wherein the processor executes the computer program to implement the steps of the method for constructing a fault diagnosis model of a nuclear power plant reactor coolant system as described in any one of the above claims or the method for diagnosing faults in a nuclear power plant reactor coolant system as described in any one of the above claims.

[0039] Fourthly, this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system as described in any one of the preceding claims, or the method for diagnosing a fault in a nuclear power plant reactor coolant system as described in any one of the preceding claims.

[0040] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system as described in any one of the preceding claims, or the method for diagnosing a fault in a nuclear power plant reactor coolant system as described in any one of the preceding claims.

[0041] According to the specific embodiments provided in this application, this application has the following technical effects:

[0042] This application provides a method for constructing a fault model, a fault diagnosis method, and related apparatus for a nuclear power plant reactor coolant system. First, based on historical operating data of the nuclear power plant reactor coolant system, the correlations between evidence layer nodes and between evidence layer nodes and fault layer nodes are calculated to determine the directed acyclic graph structure of the dynamic Bayesian network model. This replaces manual experience in defining the dynamic Bayesian network model, improving the construction efficiency and ensuring that the constructed model more closely reflects the actual system's operating logic, resulting in high model reliability. Second, based on interval type II fuzzy sets and historical operating data corresponding to all evidence layer nodes, the fuzzy observation data values ​​corresponding to all evidence layer nodes in the dynamic Bayesian network model are calculated. For RCS systems with high uncertainty... The system enables more accurate modeling, effectively eliminates sensor noise interference, and enhances the model's noise resistance and robustness. Finally, based on fuzzy observation data values, the conditional probability table between evidence layer nodes and fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed. The constructed fuzzy dynamic Bayesian network model deeply integrates the ability of interval type II fuzzy sets to express uncertainty with the temporal reasoning ability of dynamic Bayesian networks. In RCS system fault diagnosis and state assessment, it has good robustness and adaptability, and high fault diagnosis accuracy. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is an application environment diagram of a method for constructing a fault model of a nuclear power plant reactor coolant system according to an embodiment of this application;

[0045] Figure 2 A flowchart illustrating a method for constructing a fault model of a nuclear power plant reactor coolant system, provided in an embodiment of this application;

[0046] Figure 3 This is a structural diagram of a fuzzy dynamic Bayesian network model constructed in one embodiment of this application;

[0047] Figure 4 This is a comparison chart of the fault diagnosis accuracy between the fuzzy dynamic Bayesian network model constructed in one embodiment of this application and the traditional fault diagnosis model;

[0048] Figure 5 A flowchart illustrating a method for diagnosing faults in a nuclear power plant reactor coolant system, provided as an embodiment of this application;

[0049] Figure 6 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0050] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0051] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0052] This application provides a method for constructing a fault model of a nuclear power plant reactor coolant system, which can be applied to, for example... Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on other servers. Terminal 102 can send historical operating data of the nuclear power plant reactor coolant system to be processed to server 104. Server 104 receives the historical operating data of the nuclear power plant reactor coolant system to be processed. Based on the historical operating data, server 104 determines multiple evidence layer nodes and multiple fault layer nodes of the dynamic Bayesian network model; calculates the correlation coefficient between the evidence layer nodes, the first causal relationship between the evidence layer nodes, and the second causal relationship between the evidence layer nodes and the fault layer nodes, and then, based on the evidence layer nodes and the... The fault layer nodes, the correlation coefficients, the first causal relationship, and the second causal relationship determine the directed acyclic graph structure of the dynamic Bayesian network model. Based on the interval type-2 fuzzy set and the historical running data corresponding to all the evidence layer nodes, the fuzzy observation data values ​​corresponding to all the evidence layer nodes in the dynamic Bayesian network model are calculated. Based on the fuzzy observation data values, the conditional probability table between the evidence layer nodes and the fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed. The server 104 can feed back the obtained fuzzy dynamic Bayesian network model to the terminal 102. Furthermore, in some embodiments, the method for constructing a fault model of a nuclear power plant reactor coolant system can also be implemented separately by the server 104 or the terminal 102. For example, the terminal 102 can directly process the historical operating data of the nuclear power plant reactor coolant system to construct a fuzzy dynamic Bayesian network model, or the server 104 can obtain the historical operating data of the nuclear power plant reactor coolant system to be processed from the data storage system and process the historical operating data of the nuclear power plant reactor coolant system to be processed to construct a fuzzy dynamic Bayesian network model.

[0053] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.

[0054] In one exemplary embodiment, such as Figure 2As shown, a method for constructing a fault model of a nuclear power plant reactor coolant system is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 205. Wherein:

[0055] Step 201: Obtain historical operating data of the nuclear power plant reactor coolant system; the historical operating data includes historical normal operation data and historical fault operation data; the historical normal operation data is multi-parameter time-series data under normal operating conditions; the historical fault operation data includes various fault conditions and multi-parameter time-series data corresponding to each fault condition.

[0056] Step 202: Based on the historical operating data, determine multiple evidence layer nodes and multiple fault layer nodes of the dynamic Bayesian network model.

[0057] Step 203: Calculate the correlation coefficient between the evidence layer nodes, the first causal relationship between the evidence layer nodes, and the second causal relationship between the evidence layer nodes and the fault layer nodes, and determine the directed acyclic graph structure of the dynamic Bayesian network model based on the evidence layer nodes, the fault layer nodes, the correlation coefficient, the first causal relationship, and the second causal relationship.

[0058] Step 204: Based on the interval type II fuzzy set and the historical running data corresponding to all the evidence layer nodes, calculate the fuzzy observation data values ​​corresponding to all the evidence layer nodes in the dynamic Bayesian network model.

[0059] Step 205: Based on the fuzzy observation data values, iteratively calculate the conditional probability table between the evidence layer nodes and the fault layer nodes to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, construct a fuzzy dynamic Bayesian network model. The fuzzy dynamic Bayesian network model is used to determine the fault diagnosis results of the nuclear power plant reactor coolant system.

[0060] Implementing steps 201 to 205 involves two steps: First, acquiring historical operating data of the nuclear power plant's reactor coolant system. Based on this data, calculating the correlations between evidence layer nodes and between evidence layer nodes and fault layer nodes, and determining the directed acyclic graph structure of the dynamic Bayesian network model. This replaces manual experience in defining the dynamic Bayesian network model, improving its construction efficiency and ensuring the model better reflects the actual system's operating logic, resulting in high reliability. Second, based on the interval type-II fuzzy set and the historical operating data corresponding to all evidence layer nodes, calculating the fuzzy views corresponding to all evidence layer nodes in the dynamic Bayesian network model. For RCS systems with high uncertainty, the measured data values ​​enable more accurate modeling, effectively eliminating sensor noise interference and enhancing the model's noise resistance and robustness. Finally, based on the fuzzy observation data values, the conditional probability table between the evidence layer nodes and the fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed. The constructed fuzzy dynamic Bayesian network model deeply integrates the ability of interval type II fuzzy sets to express uncertainty with the temporal reasoning ability of dynamic Bayesian networks. In RCS system fault diagnosis and state assessment, it has good robustness and adaptability, and high fault diagnosis accuracy.

[0061] In another exemplary embodiment of this application, in order to quickly and accurately screen out various operating conditions of the nuclear power plant reactor coolant system during its historical operation and several variables that affect the fault diagnosis of the nuclear power plant reactor coolant system under each operating condition, so as to determine the fault layer nodes and evidence layer nodes of the dynamic Bayesian network, the above step 202 is replaced by the following steps 2021 to 2022:

[0062] Step 2021: Based on the historical operating data of the nuclear power plant reactor coolant system, the random forest algorithm is used to filter the operating conditions and several variables that affect the fault diagnosis of the nuclear power plant reactor coolant system under each operating condition from the historical operating data. The variables are multi-parameter time series data that affect the fault diagnosis of the nuclear power plant reactor coolant system under each operating condition.

[0063] As an example, we screened out normal operating conditions and seven fault operating conditions, and then screened out 18 variables that affect the fault diagnosis of the reactor coolant system in nuclear power plants under normal operating conditions and seven fault operating conditions.

[0064] Step 2022: Set variables as evidence layer nodes and operating conditions as fault layer nodes; wherein, one variable is set as one evidence layer node, and different variables are different evidence layer nodes, one operating condition is set as one fault layer node, and different operating conditions are different fault layer nodes.

[0065] As an example, the selected normal operating conditions and 7 fault operating conditions are set as fault layer nodes (Z1-Z8) of the dynamic Bayesian network model. The 8 operating conditions corresponding to the fault layer nodes (Z1-Z8) of the dynamic Bayesian network model are shown in Table 1:

[0066] Table 1. Fault layer nodes (Z) of the DBN model 1- Z8 corresponds to 8 working conditions

[0067]

[0068]

[0069] As an example, 18 variables that influence fault diagnosis of the reactor coolant system in nuclear power plants under normal operating conditions and 7 fault conditions are set as evidence layer nodes (x1-x) in the dynamic Bayesian network model. 18 ), Evidence layer nodes (x1-x) of the dynamic Bayesian network model 18 The 18 corresponding variables are shown in Table 2:

[0070] Table 2 Evidence layer nodes (x1-x) of the DBN model 18 ) corresponding to 18 variables

[0071] P <![CDATA[Pressure (x1) of the Reactor Coolant System]]> PSGA <![CDATA[Pressure of steam generator A (x2)]]> PSGB <![CDATA[Pressure of steam generator B (x3)]]> QMWT <![CDATA[Total thermal power (x4)]]> SCMA <![CDATA[Loop A undersaturation margin temperature (x5)]]> SCMB <![CDATA[Loop B Undersaturation Margin Temperature (x6)]]> TAVG <![CDATA[Average temperature (x7) of the reactor coolant system]]> TCA <![CDATA[Temperature of cold pipe A (x8)]]> TCB <![CDATA[Temperature of cold pipe B (x9)]]> THA <![CDATA[The temperature of heat pipe A (x 10 ) <!-- 6 -->]]> THB <![CDATA[The temperature of heat pipe B (x 11 )]]> VOL <![CDATA[Volume of reactor coolant system liquid (x 12 )]]> WFWA <![CDATA[Flow rate (x 13 ) of feed water to steam generator A WFWB <![CDATA[Flow rate (x 14 ) of feed water to steam generator B WRCA <![CDATA[Flow rate of reactor coolant loop A (x 15 )]]> WRCB <![CDATA[Flow rate of Reactor Coolant Loop B (x 16 )]]> WSTA <![CDATA[The flow rate of steam in steam generator A (x 17 )]]> WSTB <![CDATA[Flow rate of steam in steam generator B (x 18 )]]>

[0072] In another exemplary embodiment of this application, in order to make the constructed dynamic Bayesian network model more closely resemble the operating logic of the real system and have high model credibility, the correlation coefficient between the evidence layer nodes, the first causal relationship between the evidence layer nodes, and the second causal relationship between the evidence layer nodes and the fault layer nodes can be calculated. Based on the evidence layer nodes, the fault layer nodes, the correlation coefficient, the first causal relationship, and the second causal relationship, the directed acyclic graph structure of the dynamic Bayesian network model is determined. In this case, step 203 is replaced by steps 2031 to 2034.

[0073] Step 2031: Calculate the Kendall correlation coefficients between the evidence layer nodes, and generate an undirected acyclic graph structure of the dynamic Bayesian network model based on the Kendall correlation coefficients, the evidence layer nodes, the fault layer nodes, and the R-Teng Copula model. Specifically:

[0074] The joint cumulative distribution function of the m-dimensional variable X is defined as follows:

[0075] F(X) = F(x1,…,x) i ,…,x j ,…,x m) = C(F1(x1),…,F i (x i ),…,F j (x j ),…,F m (x m ))(1);

[0076] In the formula: F(X) = F(x1,…,x i ,…,x j ,…,x m ) is the joint cumulative distribution function of the variable X, where X = {x1,…,x i ,…,x j ,…,x m} is all the evidence layer nodes, and each variable x i represents an evidence layer node; C(·) is the Copula function; F i (x i ), F j (x j ) are the marginal cumulative distribution functions of the variables x i and the variable x j respectively.

[0077] Let u i = F i (x i ), u j = F j (x j ), and calculate the Kendall correlation coefficient τ i,j between x i and x j . The formula is as follows:

[0078]

[0079] In the formula: c(·) is the probability density function of the Copula.

[0080] Define the objective function τ, and use the objective function τ to generate the first layer tree structure in the R vine Copula model, that is, the undirected acyclic graph of the dynamic Bayesian network model. The formula is as follows:

[0081]

[0082] In the formula: ε = {(i,j)|1 ≤ i < j ≤ m} is the edge set composed of all two-dimensional variables; ε′ is a subset of the edge set, satisfying connectivity and acyclicity, and |ε′| = m - 1.

[0083] Step 2032, calculate the first causal relationship between the evidence layer nodes. Specifically:

[0084] Define x i x j Time lag embedding vector at time t x i ( t) and x j ( t) Construct shadow manifold and The formula is as follows:

[0085]

[0086] In the formula: t = 1 + (B - 1) γ ,…,T, where T represents a total of T time points; B is the embedding dimension, and γ is the lag parameter.

[0087] As an example, B = 2, γ = 1.

[0088] From shadow manifold Found on x i The B+1 nearest neighbors of (t) are denoted as x i (t g ), from the shadow manifold Found on x j The B+1 nearest neighbors of (t) are denoted as x j (t g ), calculate x i Cross-mapping estimate of (t) The formula is as follows:

[0089]

[0090] Where: μ g It is an intermediate variable; ||·|| is the Euclidean distance; ω g Indicates the weight.

[0091] calculate With x i The Pearson correlation coefficient ρ between (t) and T The formula is as follows:

[0092]

[0093] In the formula: For x i The mean of (t), for The mean.

[0094] The Pearson correlation coefficient ρ is obtained through z-transform. Tz-scores mapped to a normal distribution The formula is as follows:

[0095]

[0096] Calculate the correlation coefficient after z-transformation and The difference G between them is given by the following formula:

[0097]

[0098] In the formula: T max and T min These are the maximum and minimum sample values ​​for calculating the Pearson correlation coefficient, respectively.

[0099] If G>G α / 2 Then x i It is x j The reason.

[0100] As an example, choosing a significance level α = 0.01, the corresponding critical value G α / 2 =2.58.

[0101] Step 2033: Calculate the second causal relationship between the evidence layer node and the fault layer node.

[0102] As an example, each evidence layer node is the cause of each failure layer node.

[0103] Step 2034: Determine the directed acyclic graph structure of the dynamic Bayesian network model based on the undirected acyclic graph structure, the first causal relationship, and the second causal relationship.

[0104] In another exemplary embodiment of this application, in order to enable the constructed dynamic Bayesian network model to better handle the high uncertainty of the nuclear power plant reactor coolant system, the fuzzy observation data values ​​corresponding to all evidence layer nodes can be calculated based on the interval type II fuzzy set and the historical operating data corresponding to all evidence layer nodes, so as to effectively eliminate sensor noise interference and enhance the noise resistance and robustness of the model. In this case, step 204 above is replaced by the following steps 2041 to 2044:

[0105] Step 2041: For each of the evidence layer nodes, the historical running data corresponding to the evidence layer node is evenly divided into multiple sub-intervals.

[0106] Step 2042: For each sub-interval, perform Latin hypercube sampling on the sub-interval to obtain the sample value corresponding to the sub-interval.

[0107] Step 2043: For each sample value corresponding to the sub-interval, calculate the fuzzy observation data of the sample value corresponding to the sub-interval based on the type II fuzzy set of the interval, and obtain the fuzzy observation data of the sample value corresponding to each sub-interval.

[0108] Step 2044: Determine the fuzzy observation data of the sample values ​​corresponding to all the sub-intervals as the fuzzy observation data values ​​corresponding to the evidence layer nodes, and obtain the fuzzy observation data values ​​corresponding to all the evidence layer nodes.

[0109] As an example, the implementation process of steps 2041 to 2044 is as follows: The variable x corresponding to the evidence layer node... i The domain [I,J] (i.e., the historical running data corresponding to the evidence layer nodes) is uniformly divided into r first sub-intervals {λ1,…,λ k ,…,λ r}, perform Latin hypercube sampling on r first subintervals to obtain variable x i The sample values ​​in each first sub-interval are used to obtain the variable x. i sample set s i ={s i1 ,…,s ik ,…,s ir The sample values ​​of each first sub-interval are solved using the interval type-II fuzzy set and cubic spline interpolation methods to obtain the fuzzy observation data value s of the evidence layer node in the dynamic Bayesian network model. i For each evidence layer node, the above process is repeated to obtain the fuzzy observation data values ​​S={s1,…,s} of all evidence layer nodes in the dynamic Bayesian network model. i ,…,s m}

[0110] As an example, for the k-th first subinterval Sample values ​​s ik The solution process is as follows:

[0111] The k-th first subinterval is blurred using a Gaussian kernel function, and the fuzzy probability density function of the k-th first subinterval is defined. The formula is as follows:

[0112]

[0113] In the formula: K(·) is the Gaussian kernel function.

[0114] Fuzzy probability density function for the k-th first subinterval Perform iterative calculations to obtain the fuzzy cumulative distribution function corresponding to the kth first sub-interval. The formula is as follows:

[0115]

[0116] In the formula: and They are The upper and lower bounds; η represents the cut set; h represents the bandwidth; This indicates a fuzzy reduction.

[0117] As an example, η = 2.

[0118] fuzzy cumulative distribution function The cumulative probability interval [0,1] is divided into r-1 second subintervals; for each second subinterval, the fuzzy cumulative distribution function corresponding to the second subinterval is solved using the cubic spline interpolation method. inverse function The formula is as follows:

[0119]

[0120] Where: β k ,σ k ,ψ k and Γ k These are polynomial coefficients.

[0121] It is understandable that the β obtained by fitting within each second sub-interval k ,σ k ,ψ k and Γ k Therefore, each second sub-interval corresponds to a fuzzy cumulative distribution function. inverse function

[0122] Solve for the first subinterval {λ1,…,λ k ,…,λ r Generate the kth first subinterval within} The cumulative probability value y k The formula is as follows:

[0123]

[0124] Determine the cumulative probability value y k To determine which second subinterval it belongs to, the cumulative probability value y will be calculated. k Substitute the fuzzy cumulative distribution function of the corresponding second subinterval inverse function In the process, we obtain the variable x. i In the kth first subinterval Sample values ​​s ik The formula is as follows:

[0125]

[0126] In another exemplary embodiment of this application, in order to deeply integrate the ability of interval type II fuzzy sets to express uncertainty with the temporal reasoning ability of dynamic Bayesian networks, and obtain a fuzzy dynamic Bayesian network model that has good robustness and adaptability and high fault diagnosis accuracy in RCS system fault diagnosis and state assessment, the conditional probability table between the evidence layer nodes and the fault layer nodes can be iteratively calculated based on the fuzzy observation data values ​​to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed. Therefore, step 205 is replaced by steps 2051 to 2053:

[0127] Step 2051: Based on the fuzzy observation data values, the expectation-maximization algorithm is used to iteratively calculate the conditional probability table between the evidence layer nodes and the fault layer nodes to obtain the initial conditional probability table, as follows:

[0128] The expected value of each state of the fault layer node corresponding to the fuzzy observation data value S is calculated through the E-step of the expectation-maximization algorithm. The formula is as follows:

[0129]

[0130] In the formula: d is the number of iterations of the expectation-maximization algorithm; θ ( d) is the conditional probability table estimated after the d-th iteration.

[0131] Among them, each state of the fault layer node refers to either the occurrence or non-occurrence of the corresponding fault condition.

[0132] Through the M steps of the expectation-maximization algorithm, the maximum likelihood estimate of the expected value is calculated, and then a new conditional probability table θ is estimated. (d+1) This is the initial conditional probability table, and the formula is as follows:

[0133]

[0134] In the formula: θ is the estimated conditional probability table, referring to the table used in the (d+1)th calculation. ( d) Estimated θ (d+1) A candidate solution; P(S,Z|θ) is the joint probability distribution between the fault layer nodes and the evidence layer nodes.

[0135] Step 2052: Based on the initial conditional probability table, a set of candidate conditional probability tables is randomly generated. Based on the whale optimization algorithm and the set of candidate conditional probability tables, the optimal conditional probability table is obtained, as follows:

[0136] From θ (d+1) A set of candidate conditional probability tables is randomly generated. The whale optimization algorithm is applied to any one of the conditional probability tables θ in the candidate conditional probability table set. q After w+1 iterations, the optimized conditional probability table is obtained. Through continuous optimization, the optimal conditional probability table is selected from the candidate set as the final conditional probability table for the DBN model, as shown in the following formula:

[0137] A = 2a·φ-a (23);

[0138] H = 2·φ (24);

[0139]

[0140] In the formula: w is the number of iterations of the whale optimization algorithm; A and H are optimization coefficients; a is a convergence factor that decreases linearly from 2 to 0 during the iteration process; φ is a random value in the interval [0,1]. It is the optimal conditional probability table in the candidate set after the w-th iteration. It is the conditional probability table after w iterations of optimization; D is and The distance between them; b is a logarithmic spiral constant; l and p represent random values ​​in the intervals [-1, 1] and [0, 1], respectively.

[0141] As an example, b = 1.

[0142] Step 2053: Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, construct the fuzzy dynamic Bayesian network model. The fuzzy dynamic Bayesian network model is used to determine the fault diagnosis results of the reactor coolant system of a nuclear power plant.

[0143] As an example, such as Figure 3 As shown, the fault layer nodes Z in the fuzzy dynamic Bayesian network model include Z1-Z8, and the variables X corresponding to the evidence layer nodes include x1-x. 18 .

[0144] This application also provides an application scenario in which the above-described method for constructing a fault model of a nuclear power plant reactor coolant system is applied. Specifically, the method for constructing a fault model of a nuclear power plant reactor coolant system provided in this embodiment can be applied in scenarios involving fault diagnosis and condition assessment of nuclear power plant reactor coolant systems.

[0145] To further demonstrate the advantages of the fault diagnosis model (fuzzy DBN model) constructed in this application, the fault diagnosis accuracy of the fuzzy DBN model in this application was compared with the fault diagnosis accuracy of the traditional DBN model and a type-I fuzzy DBN model. Figure 4 As shown in the figure, under noise-free conditions, the diagnostic accuracy of all fault diagnosis models is above 90%. Under Gaussian noise with a mean of 0 and a standard deviation of 0.05, the fault diagnosis accuracy of the fuzzy DBN model in this application decreases from 96.79% to 95.93%, but is still better than other models. Under Gaussian noise with a mean of 0 and a standard deviation of 0.1-0.25, the fault diagnosis accuracy of the fuzzy DBN model in this application decreases by 6.45%, which is smaller than that of the traditional DBN model and the Type I fuzzy DBN model. Under Gaussian noise with a mean of 0 and a standard deviation of 0.3, the fault diagnosis accuracy of the fuzzy DBN model in this application is still 88.04%, which is 15.12% higher than that of the traditional DBN model. The above results show that under high uncertainty conditions, the fuzzy DBN model constructed in this application has a higher diagnostic accuracy than the traditional model.

[0146] Based on the same inventive concept, this application also provides a method for diagnosing faults in a nuclear power plant reactor coolant system, such as... Figure 5 As shown, this method is executed by a computer device, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 301 to 302. Wherein:

[0147] Step 301: Real-time acquisition of multi-parameter time-series data of the nuclear power plant reactor coolant system.

[0148] Step 302: Input the real-time collected multi-parameter time-series data into the fuzzy dynamic Bayesian network model to obtain the fault diagnosis results of the nuclear power plant reactor coolant system; wherein, the fuzzy dynamic Bayesian network model is determined based on the nuclear power plant reactor coolant system fault diagnosis model construction method described in any of the above embodiments.

[0149] In another exemplary embodiment of this application, in order to obtain the causal chain between the fault layer node and the evidence layer node when a nuclear power plant reactor coolant system malfunctions, the nuclear power plant reactor coolant system fault diagnosis method further includes steps 303 to 304. Wherein:

[0150] Step 303: Based on the fault diagnosis results of the nuclear power plant reactor coolant system and the fuzzy dynamic Bayesian network model, determine the fault layer node corresponding to the multi-parameter time-series data of the nuclear power plant reactor coolant system collected in real time.

[0151] Step 304: Based on the fault layer nodes corresponding to the nuclear power plant reactor coolant system and the fuzzy dynamic Bayesian network model, determine the evidence layer nodes corresponding to the multi-parameter time-series data of the nuclear power plant reactor coolant system collected in real time.

[0152] As an example, to demonstrate the interpretability provided by the fuzzy DBN model constructed in this application, a fault diagnosis case for a test sample under the Z2 operating condition (small breakage and water loss accident in the heat pipe section) is selected in this example. The evidence variables and fault diagnosis results in this fault diagnosis case are shown in Tables 3-5.

[0153] Table 3 Fault Diagnosis Cases - Time Window 0s

[0154]

[0155] When the time window for the test sample is within 1 second, variables x1, x2, x4, x8, and x 10 -x 18 The state of Z2 is low, the states of X3 and X9 are medium (normal state), and the states of X5, X6, and X7 are high. At this time, the posterior probability of Z2 is 26.25%, which is lower than the normal operating condition (Z1). When the fuzzy DBN model constructed in this application is used for fault diagnosis, after inputting the observed values ​​of the evidence layer variables of the current time slice, the greater the posterior probability of the fault layer variable, the more likely the fault type corresponding to the variable is the fault diagnosis result of the current time slice. Therefore, when the time window of the test sample is within 1 second, the fault diagnosis result is normal operating condition. The reason for this misdiagnosis is that within this time window, the fault characteristics of Z2 are not obvious, and the existing evidence is insufficient to accurately diagnose Z2, thus causing it to be classified as Z1.

[0156] Table 4 Fault Diagnosis Cases - Time Window 120s

[0157]

[0158] When the time window is 120 seconds, the state of x3 changes from medium to high, while x7 changes from high to medium. With this new evidence, the posterior probability of Z2 rises to 97.70%, while the posterior probability of Z1 falls to 2.30%. This change is mainly due to two reasons: the increase in x3 (pressure of steam generator B) indicates anomalies in heat exchange within the RCS, possibly due to a reduction in coolant; the decrease in x7 (average temperature of the RCS) indicates insufficient coolant to remove heat from the reactor core.

[0159] Table 5 Fault Diagnosis Cases - Time Window 240s

[0160]

[0161] When the time window was 240 seconds, the states of x7 and x9 changed from medium to low, and x5 and x6 changed from high to medium. The results showed that the posterior probability of Z2 further increased to 99.46%, significantly higher than the other seven operating conditions. This can be attributed to the following two points: the decrease in x7 (average temperature of the reactor coolant system) and x9 (temperature of cold pipe B) highlights the severe limitation of coolant flow, indicating that RCS is no longer able to effectively remove heat from the reactor core; the decrease in x5 (undersaturation margin temperature of A loop) and x6 (undersaturation margin temperature of B loop) indicates that the coolant temperature is close to or exceeds its saturation temperature, reducing its cooling capacity.

[0162] Based on the same inventive concept, this application also provides a fuzzy DBN inference algorithm for fault diagnosis of reactor coolant systems in nuclear power plants. This algorithm is executed by a computer device, specifically a terminal or server, or both. In this application embodiment, the algorithm is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps:

[0163] Input evidence layer nodes X = {x1,…,x} i ,…,x j ,…,x m}; Fault layer node Z; Sample size r for Latin hypercube sampling; Number of iterations d for the expectation-maximization algorithm; Number of iterations w for the whale optimization algorithm.

[0164] Step 1, use formulas (1)-(3) to calculate variable x i and x j Kendall correlation coefficient between them.

[0165] Step 2: Use formula (4) to learn the undirected acyclic graph structure in the DBN model.

[0166] Step 3, construct the shadow manifold using formulas (5)-(6). and

[0167] Step 4: Calculate the cross-mapping estimate using formulas (7)-(9).

[0168] Step 5, calculate using formula (10) and x iThe Pearson correlation coefficient ρ between (t) and T .

[0169] Step 6: Calculate the correlation coefficient after z-transformation using formulas (11) and (12). and The differences G between them are studied, and the structure in the DBN model is learned.

[0170] Step 7, for 1≤i≤m.

[0171] Step 8, x i The domain is divided into r subintervals.

[0172] Step 9: Define the probability density function in each subinterval using formulas (13)-(16).

[0173] Step 10: Define the fuzzy cumulative distribution function in each sub-interval using formula (17).

[0174] Step 11: Use formula (18) to approximate any function using multiple cubic polynomials in each subinterval.

[0175] Step 12: Use formula (19)-formula (20) to generate r Latin hypercube samples.

[0176] Step 13, End.

[0177] Step 14, define the Latin hypercube sample as S = {s1, ..., s2}. i ,…,s m} represents the observation data of the evidence layer nodes.

[0178] Step 15: Initialize the conditional probability table in the DBN model.

[0179] Step 16, for 1≤j≤d.

[0180] Step 17: Use formula (21) to calculate the expected value of the fault layer node Z corresponding to the Latin hypercube sample S.

[0181] Step 18: Use formula (22) to re-estimate a new conditional probability table.

[0182] Step 19, End.

[0183] Step 20, from θ ( In d), a candidate conditional probability table is randomly generated.

[0184] Step 21, for 1≤l≤w.

[0185] Step 22: Iterate the candidate conditional probability table using formulas (23)-(26).

[0186] Step 23, End.

[0187] Step 24: Select the best conditional probability table from the candidate set as the conditional probability table that satisfies the global optimal solution.

[0188] Step 25: Calculate the posterior probability of fault layer node Z.

[0189] Step 26: Output the fault diagnosis results.

[0190] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 6 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operating system and computer programs stored in the non-volatile storage media. The database stores historical operating data or real-time multi-parameter time-series data of the nuclear power plant reactor coolant system. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system, a fault diagnosis method for a nuclear power plant reactor coolant system, or a fuzzy DBN inference algorithm for fault diagnosis of a nuclear power plant reactor coolant system.

[0191] Those skilled in the art will understand that Figure 6 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0192] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0193] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0194] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0195] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0196] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0197] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0198] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system, characterized in that, The method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system includes: Acquire historical operating data of the reactor coolant system of a nuclear power plant; the historical operating data includes historical normal operation data and historical fault operation data; the historical normal operation data is multi-parameter time-series data under normal operating conditions; the historical fault operation data includes various fault conditions and multi-parameter time-series data corresponding to each fault condition; Based on the historical operating data, multiple evidence layer nodes and multiple fault layer nodes of the dynamic Bayesian network model are determined. Calculate the Kendall correlation coefficients between the evidence layer nodes, and generate an undirected acyclic graph structure of the dynamic Bayesian network model based on the Kendall correlation coefficients, the evidence layer nodes, the fault layer nodes, and the R-Teng Copula model; Calculating the first causal relationship between the nodes of the evidence layer includes defining the lag embedding vector of each evidence layer node and calculating the corresponding shadow manifold, calculating the Pearson correlation coefficient between the lag embedding vector of each evidence layer node and the corresponding cross-mapping estimate, and combining the z-transform to calculate the difference between the z-scores corresponding to the maximum and minimum sample values ​​of the Pearson correlation coefficient to determine the first causal relationship. Calculate the second causal relationship between the evidence layer node and the fault layer node; The directed acyclic graph structure of the dynamic Bayesian network model is determined based on the undirected acyclic graph structure, the first causal relationship, and the second causal relationship. For each evidence layer node, the historical operation data corresponding to the evidence layer node is evenly divided into multiple sub-intervals; for each sub-interval, Latin hypercube sampling is performed to obtain the sample value corresponding to the sub-interval; for each sample value corresponding to the sub-interval, based on the interval type II fuzzy set, the fuzzy observation data of the sample value corresponding to the sub-interval is calculated to obtain the fuzzy observation data of the sample value corresponding to the sub-interval; the fuzzy observation data of the sample values ​​corresponding to all sub-intervals are determined as the fuzzy observation data value corresponding to the evidence layer node to obtain the fuzzy observation data value corresponding to all the evidence layer nodes. Based on the fuzzy observation data values, the conditional probability table between the evidence layer nodes and the fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed. The fuzzy dynamic Bayesian network model is used to determine the fault diagnosis results of the reactor coolant system of a nuclear power plant.

2. The method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system according to claim 1, characterized in that, Based on the fuzzy observation data values, the conditional probability table between the evidence layer nodes and the fault layer nodes is iteratively calculated to obtain the optimal conditional probability table. Then, based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, a fuzzy dynamic Bayesian network model is constructed, specifically including: Based on the fuzzy observation data values, the expectation-maximization algorithm is used to iteratively calculate the conditional probability table between the evidence layer nodes and the fault layer nodes to obtain the initial conditional probability table. Based on the initial conditional probability table, a set of candidate conditional probability tables is randomly generated; Based on the whale optimization algorithm and the set of candidate conditional probability tables, the optimal conditional probability table is obtained; Based on the directed acyclic graph structure of the dynamic Bayesian network model and the optimal conditional probability table, the fuzzy dynamic Bayesian network model is constructed.

3. The method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system according to claim 1, characterized in that, Based on the historical operational data, multiple evidence layer nodes and multiple fault layer nodes are identified in the dynamic Bayesian network model, specifically including: Using the random forest algorithm, the operating conditions and several variables that affect the fault diagnosis of the reactor coolant system in nuclear power plants under each operating condition are selected from the historical operating data. Set the variables as evidence layer nodes and the operating conditions as fault layer nodes; In this system, one variable is set as one evidence layer node, and different variables are set as different evidence layer nodes. One working condition is set as one fault layer node, and different working conditions are set as different fault layer nodes.

4. A method for diagnosing faults in a nuclear power plant reactor coolant system, characterized in that, The fault diagnosis method for the nuclear power plant reactor coolant system includes: Real-time acquisition of multi-parameter time-series data of the reactor coolant system in nuclear power plants; The real-time collected multi-parameter time-series data is input into the fuzzy dynamic Bayesian network model to obtain the fault diagnosis results of the nuclear power plant reactor coolant system. The fuzzy dynamic Bayesian network model is determined based on the method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system as described in any one of claims 1-3.

5. The method for fault diagnosis of a nuclear power plant reactor coolant system according to claim 4, characterized in that, Also includes: Based on the fault diagnosis results of the nuclear power plant reactor coolant system and the fuzzy dynamic Bayesian network model, the fault layer nodes corresponding to the multi-parameter time-series data of the nuclear power plant reactor coolant system collected in real time are determined. Based on the fault layer nodes corresponding to the nuclear power plant reactor coolant system and the fuzzy dynamic Bayesian network model, the evidence layer nodes corresponding to the multi-parameter time-series data of the nuclear power plant reactor coolant system collected in real time are determined.

6. A computer device, comprising: A memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that the processor executes the computer program to implement the method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system as described in any one of claims 1-3 or the method for diagnosing a fault in a nuclear power plant reactor coolant system as described in any one of claims 4-5.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program implements the method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system as described in any one of claims 1-3, or the method for diagnosing a fault in a nuclear power plant reactor coolant system as described in any one of claims 4-5.

8. A computer program product, comprising a computer program, characterized in that, When executed by a processor, the computer program implements the method for constructing a fault diagnosis model for a nuclear power plant reactor coolant system as described in any one of claims 1-3, or the method for diagnosing a fault in a nuclear power plant reactor coolant system as described in any one of claims 4-5.

Images