Nuclear power plant diagnostic model development methods, diagnostic methods, systems, media and equipment

Abstract

This invention relates to a method, system, medium, and equipment for developing nuclear power plant diagnostic models, including: generating a nuclear power plant accident dataset; filtering the nuclear power plant accident dataset to obtain filtered feature parameters; clustering the feature parameters to obtain a clustered accident response parameter dataset; constructing a Bayesian network diagnostic model; learning probabilistic parameters based on the clustered accident response parameter dataset and the Bayesian network diagnostic model to form a conditional probability table; testing and optimizing the Bayesian network diagnostic model; determining whether the diagnostic accuracy of the final trained Bayesian network diagnostic model meets the requirements; if it does, completing model development; if not, adjusting and optimizing until the diagnostic accuracy meets the requirements. This invention solves the problems existing in the prior art, achieving accurate diagnosis from abnormal symptoms to fault causes, significantly improving the accuracy of system diagnosis and emergency response capabilities.

Description

Technical Field

[0001] This invention relates to the technical field of nuclear power plant fault diagnosis, and more specifically, to a method for developing a nuclear power plant diagnostic model, a diagnostic method, a system, a medium, and equipment. Background Technology

[0002] Nuclear power plant accident diagnosis, as a key technology for ensuring nuclear safety, has always focused on accuracy and real-time performance in industry research. Sensor malfunctions and signal interference are common problems in nuclear power plant accident scenarios. The uncertainty of multi-source data and the complex causal relationships make it difficult for traditional diagnostic methods to efficiently integrate dynamic information and meet the real-time diagnostic needs under complex operating conditions. Against this backdrop, there is an urgent need for a robust intelligent diagnostic technology to overcome the bottlenecks of traditional methods in uncertainty handling, multi-source information fusion, and slow computation speed.

[0003] Traditional solutions mainly fall into two categories. One is based on Bayesian inference (e.g., patent CN202510953526.X). This prediction method relies primarily on historical consumption data and fails to fully consider the impact of dynamic factors such as real-time equipment operating status and maintenance plan changes on spare parts demand, resulting in insufficient adaptability of the prediction model. Furthermore, this method is highly dependent on data quality and completeness; in the presence of missing data or noise interference, the prediction accuracy drops significantly, making it difficult to meet the real-time and reliability requirements of nuclear power plant spare parts management. The other is based on imitation learning (e.g., patent CN202510566496.7). Its imitation learning strategy relies entirely on historical operating data, lacking the ability to make autonomous decisions under unknown operating conditions and sudden situations, thus limiting the model's generalization ability. Additionally, this method uses a simple average of the outputs of multiple agents as the final control result, lacking an adaptive evaluation mechanism for the reliability of each agent's output under dynamic operating conditions, which may generate unsafe control commands under complex transient conditions. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a method for developing a diagnostic model for a nuclear power plant, a diagnostic method, a system, a medium, and equipment, in response to the problems existing in the prior art.

[0005] The technical solution adopted by this invention to solve its technical problem is: a method for developing a nuclear power plant accident diagnosis model, comprising: Generate a dataset of nuclear power plant accidents; The nuclear power plant accident dataset is filtered for characteristic operating parameters to obtain the filtered characteristic parameters. Time series clustering is performed on the filtered feature parameters to obtain a clustered accident response parameter dataset; Construct a Bayesian network diagnostic model; Based on the clustered accident response parameter dataset, and combined with the Bayesian network diagnostic model, probability parameters are learned to form a conditional probability table. The Bayesian network diagnostic model was tested and optimized multiple times. Determine whether the diagnostic accuracy of the final trained Bayesian network diagnostic model meets the requirements; If yes, then the development of the nuclear power plant accident diagnosis model will be completed; if not, then the parameters will continue to be adjusted to optimize the model until the diagnostic accuracy meets the requirements.

[0006] In the nuclear power plant accident diagnosis model development method of the present invention, the generation of nuclear power plant accident dataset includes: Obtain typical accident conditions and other accident conditions of nuclear power plants; The nuclear power plant system simulation model was used to simulate and calculate the typical accident conditions and other accident conditions to obtain the dynamic response characteristics of the characteristic operating parameters. An accident response parameter dataset containing time-series variation curves is generated based on the dynamic response characteristics of the aforementioned characteristic operating parameters; the accident response parameter dataset is the nuclear power plant accident dataset.

[0007] In the nuclear power plant accident diagnosis model development method of the present invention, the step of filtering the feature operating parameters of the nuclear power plant accident dataset to obtain the filtered feature parameters includes: Acquire expert experience and knowledge; Based on the expert experience and knowledge, parameters related to nuclear power plant accident diagnosis are selected from the nuclear power plant accident dataset to obtain the filtered feature parameters.

[0008] In the nuclear power plant accident diagnosis model development method of the present invention, the step of performing time series clustering on the filtered feature parameters to obtain a clustered accident response parameter dataset includes: The K-Means clustering algorithm is used to perform time series clustering on the filtered feature parameters to obtain the clustered accident response parameter dataset.

[0009] In the nuclear power plant accident diagnosis model development method of the present invention, the step of using the K-Means clustering algorithm to perform time series clustering on the filtered feature parameters to obtain the clustered accident response parameter dataset includes: The K-Means clustering algorithm is used to group the filtered feature parameters, and the discrete data of the feature parameters output according to a set interval are clustered. The steady-state data are grouped according to the set fluctuation range to form the clustered accident response parameter dataset.

[0010] In the nuclear power plant accident diagnosis model development method of the present invention, the construction of the Bayesian network diagnosis model includes: To obtain the physical characteristics of the thermal-hydraulic and system control of nuclear power plants; Gain access to expert engineering experience and fault analysis knowledge; The topology of the Bayesian network diagnostic model is constructed based on the physical characteristics of the nuclear power plant's thermal-hydraulic and system control, the expert's engineering experience, and fault analysis knowledge.

[0011] In the nuclear power plant accident diagnosis model development method described in this invention, the construction of the Bayesian network diagnosis model is a hierarchical serial reasoning model constructed using a hierarchical serial reasoning mode.

[0012] In the nuclear power plant accident diagnosis model development method described in this invention, the reasoning process of the Bayesian network diagnosis model is as follows: Using an accident type Bayesian network model, the category of the current accident is initially determined, and a prior probability distribution of the accident type is established. Based on the prior probability distribution of the accident type, the corresponding Bayesian network model for accident location is dynamically invoked to locate the accident and obtain the accident location information. Based on the determined accident type and location information, the corresponding accident occurrence loop Bayesian network model is invoked for precise localization.

[0013] In the nuclear power plant accident diagnosis model development method of the present invention, the step of combining the Bayesian network diagnosis model to learn probability parameters and form a conditional probability table includes: Based on the causal Bayesian network structure of the Bayesian network diagnostic model, the conditional probabilities between nodes are learned using the Bayesian estimation method to form the conditional probability table.

[0014] The present invention also provides a method for diagnosing nuclear power plant accidents, comprising: The Bayesian network diagnostic model is invoked; the Bayesian network diagnostic model is a model designed using the nuclear power plant accident diagnostic model development method according to any one of claims 1-9; Obtain real-time monitoring data on characteristic operations; The K-Means clustering algorithm is used to discretize the feature data to obtain grouped data; The grouped data is input into the Bayesian network diagnostic model for calculation to obtain the posterior probability distribution of the accident type.

[0015] In the nuclear power plant accident diagnosis method of the present invention, the step of inputting the grouped data into a Bayesian network diagnostic model for inference to obtain the posterior probability distribution of the accident type includes: The grouped data is input into the Bayesian network diagnostic model; Based on the grouped data and combined with the causal Bayesian network structure of the Bayesian network diagnostic model, the Bayesian estimation method is used to learn the parameters of the conditional probabilities between nodes to form a conditional probability table. Based on the conditional probability table and the causal Bayesian network structure, a variable elimination algorithm is used for diagnostic reasoning to obtain the posterior probability of the accident type.

[0016] The present invention also provides a nuclear power plant accident diagnosis system, comprising: The model invocation unit is used to invoke the Bayesian network diagnostic model; the Bayesian network diagnostic model is a model designed using the nuclear power plant accident diagnostic model development method described above. The real-time data acquisition unit is used to acquire real-time monitoring feature operation data; The data clustering unit is used to discretize the feature running data using the K-Means clustering algorithm to obtain grouped data; The accident type diagnosis unit is used to input the grouped data into the Bayesian network diagnosis model for calculation to obtain the posterior probability distribution of the accident type.

[0017] The present invention also provides a storage medium storing a computer program adapted for loading by a processor to perform the steps of the nuclear power plant accident diagnosis method as described above.

[0018] The present invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the nuclear power plant accident diagnosis method as described above by calling the computer program stored in the memory.

[0019] The nuclear power plant diagnostic model development method, diagnostic method, system, medium, and equipment of this invention have the following beneficial effects: They include: generating a nuclear power plant accident dataset; filtering the nuclear power plant accident dataset to obtain filtered feature parameters; clustering the feature parameters to obtain a clustered accident response parameter dataset; constructing a Bayesian network diagnostic model; learning probabilistic parameters based on the clustered accident response parameter dataset and the Bayesian network diagnostic model to form a conditional probability table; testing and optimizing the Bayesian network diagnostic model; determining whether the diagnostic accuracy of the final trained Bayesian network diagnostic model meets the requirements; if it does, completing the model development; if not, adjusting and optimizing until the diagnostic accuracy meets the requirements. This invention can solve the problems existing in the prior art, achieving accurate diagnosis from abnormal symptoms to fault causes, and significantly improving the accuracy of system diagnosis and emergency response capabilities. Attached Figure Description

[0020] The present invention will be further described below with reference to the accompanying drawings and embodiments. In the accompanying drawings: Figure 1 This is a flowchart illustrating the nuclear power plant accident diagnosis model development method provided in this embodiment of the invention. Figure 2 This is a schematic diagram of the reasoning process of the hierarchical serial reasoning model provided in the embodiments of the present invention; Figure 3 This is a flowchart illustrating the nuclear power plant accident diagnosis method provided in an embodiment of the present invention; Figure 4 This is a logic block diagram of the nuclear power plant accident diagnosis system provided in an embodiment of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] To address the problems existing in current technologies, this invention provides a method for developing a nuclear power plant accident diagnosis model. This method generates a nuclear power plant accident dataset, filters characteristic operating parameters, and performs time-series clustering to build a complete diagnostic data foundation. Simultaneously, for the constructed Bayesian network model, an iterative optimization mechanism is employed to continuously improve model performance. Ultimately, this provides a more effective Bayesian network diagnostic method (i.e., a nuclear power plant accident diagnosis method) for integrating multi-source uncertain information. This diagnostic method utilizes a probabilistic reasoning framework to reliably diagnose common problems such as sensor failures and signal interference, from abnormal symptoms to the cause of the failure. Furthermore, this diagnostic method focuses on the clustering processing of characteristic parameters and the optimized design of the Bayesian network structure, significantly improving diagnostic accuracy and system emergency response capabilities, providing strong technical support for nuclear safety systems.

[0023] In a preferred embodiment, such as Figure 1 As shown, the development method for the nuclear power plant accident diagnosis model includes the following steps: Step S10: Generate nuclear power plant accident dataset.

[0024] In some embodiments, generating a nuclear power plant accident dataset includes: acquiring typical accident conditions and other accident conditions of the nuclear power plant; using a nuclear power plant system simulation model to perform simulation calculations on the typical accident conditions and other accident conditions to obtain the dynamic response characteristics of characteristic operating parameters; generating an accident response parameter dataset containing time-series variation curves based on the dynamic response characteristics of the characteristic operating parameters; the accident response parameter dataset is a nuclear power plant accident dataset.

[0025] Specifically, using a high-precision nuclear power plant system simulation model and system analysis program, simulation calculations are performed on typical nuclear power plant accident conditions and other accident conditions (i.e., accident conditions that need to be considered, specifically determined by actual application, and not specifically limited in this invention). This captures the dynamic response characteristics of characteristic operating parameters such as reactor power, pressurizer water level and pressure, coolant temperature and flow rate, steam generator water level, and main feedwater flow rate during the accident process, generating an accident response parameter dataset containing time-series variation curves. This accident response parameter dataset containing time-series variation curves is the nuclear power plant accident dataset. Constructing this nuclear power plant accident dataset can provide a scientific basis for subsequent analysis.

[0026] Step S20: Filter the characteristic operating parameters of the nuclear power plant accident dataset to obtain the filtered characteristic parameters.

[0027] In some embodiments, the process of filtering the nuclear power plant accident dataset for characteristic operating parameters to obtain the filtered characteristic parameters includes: acquiring expert experience knowledge; and selecting parameters related to nuclear power plant accident diagnosis from the nuclear power plant accident dataset based on the expert experience knowledge to obtain the filtered characteristic parameters.

[0028] Specifically, by leveraging expert experience and knowledge, parameters most relevant to nuclear power plant accident diagnosis are selected from the accident response parameter dataset to form a simplified accident response parameter dataset (i.e., filtered feature parameters). This effectively reduces the input of irrelevant or redundant data and avoids the negative impact of data quality on the model's diagnostic accuracy.

[0029] Step S30: Perform time series clustering on the filtered feature parameters to obtain a clustered accident response parameter dataset.

[0030] In some embodiments, performing time-series clustering on the filtered feature parameters to obtain a clustered accident response parameter dataset includes: using the K-Means clustering algorithm to perform time-series clustering on the filtered feature parameters to obtain a clustered accident response parameter dataset. Preferably, using the K-Means clustering algorithm to perform time-series clustering on the filtered feature parameters to obtain a clustered accident response parameter dataset includes: using the K-Means clustering algorithm to group the filtered feature parameters, performing clustering processing on the discrete data of the feature parameters output according to a set interval; and grouping the steady-state data according to a set fluctuation range to form a clustered accident response parameter dataset. The set interval can be 10 seconds.

[0031] Specifically, the K-Means clustering algorithm was used to group the simplified accident response parameter dataset. The nuclear power plant system simulation model was clustered according to the discrete data of characteristic operating parameters output every 10 seconds. Due to the high sensitivity of Bayesian networks to data changes, even decimal-level differences will be identified as independent groups. Clustering can effectively reduce the amount of input data for the Bayesian network diagnostic model. Compared with traditional equal-width partitioning, equal-frequency partitioning, or empirical partitioning methods, clustering can accurately capture the natural distribution characteristics of the data and more effectively preserve the inherent details of the data, thus providing high-quality input for Bayesian network diagnosis. At the same time, for steady-state data, a fluctuation range of ±1-3% was set, and the data within this range were grouped separately to simulate the measurement error of the measuring instruments on the characteristic operating parameters. Finally, a clustered accident response parameter dataset was formed, providing standardized data input for the efficient operation of the subsequent Bayesian network diagnostic model. This invention employs the K-Means clustering algorithm to discretize and preprocess the time series of characteristic operating parameters of nuclear power plants, thereby improving the quality of input data for Bayesian networks. Furthermore, this invention specifically sets fluctuation ranges for steady-state data and groups them into single clusters, which can simulate instrument errors and enhance robustness.

[0032] The goal of the K-Means clustering algorithm is to minimize the sum of squared errors within clusters. Its core formula is as follows: In the formula, It is the number of clusters; It is the first The data point set of each cluster; It belongs to Data points; It is the first The center of a cluster; Representing data points With cluster center The square of the Euclidean distance between them.

[0033] Step S40: Construct a Bayesian network diagnostic model.

[0034] In this embodiment of the invention, constructing a Bayesian network diagnostic model includes: acquiring the physical characteristics of the thermal-hydraulic and system control systems of a nuclear power plant; acquiring expert engineering experience and fault analysis knowledge; and constructing the topology of the Bayesian network diagnostic model based on the physical characteristics of the nuclear power plant's thermal-hydraulic and system control systems, expert engineering experience, and fault analysis knowledge. Specifically, this constructed Bayesian network diagnostic model is a hierarchical serial inference model built using a hierarchical serial inference pattern. That is, the Bayesian network diagnostic model is a hierarchical serial inference model built based on expert experience and engineering experience.

[0035] In practical applications, based on the physical characteristics of nuclear power plant thermal-hydraulic systems and control systems, and combined with expert engineering experience and fault analysis knowledge, a topology for a Bayesian network diagnostic model is constructed. The details are as follows: By analyzing the physical correlations and causal transmission paths among various characteristic operating parameters, the directed arc connections between nodes are determined, forming a causal Bayesian network structure that conforms to the evolution logic of nuclear power plant accidents.

[0036] This causal Bayesian network structure employs a hierarchical sequential reasoning pattern. Its core idea is to decouple different diagnostic dimensions and construct the reasoning process as a hierarchical sequential reasoning model, forming a progressive diagnostic mechanism from coarse to fine granular. The specific reasoning process is as follows: Figure 2 As shown. Specifically, the reasoning process is mainly divided into three progressive stages: First (Phase 1): A preliminary determination of the current accident category is made using an accident type Bayesian network model, establishing a prior probability distribution for the accident type. This phase primarily relies on system-level macroscopic thermal parameters to establish the prior probability distribution for the accident type.

[0037] Secondly (second stage): Based on the prior probability distribution of the accident type, the corresponding Bayesian network model for accident location is dynamically invoked to perform location localization and obtain accident location information. Based on the accident type diagnosis results of the first stage, the Bayesian network model for accident location corresponding to that accident type is dynamically invoked, and the specific feature parameters of that type of accident are used for location localization, thereby gradually converging the diagnostic range.

[0038] Finally (Phase 1): Based on the determined accident type and location information, the corresponding accident occurrence loop Bayesian network model is invoked for precise localization. In this phase, based on the determined accident type and location information, the corresponding accident occurrence loop Bayesian network model is further invoked, and the final precise localization is achieved through the differential signals between each loop.

[0039] Step S50: Based on the clustered accident response parameter dataset, combine the Bayesian network diagnostic model to learn the probability parameters and form a conditional probability table.

[0040] The process of learning probability parameters and forming a conditional probability table by combining a Bayesian network diagnostic model includes: learning the parameters of the conditional probabilities between nodes using Bayesian estimation methods based on the causal Bayesian network structure of the Bayesian network diagnostic model, and forming a conditional probability table.

[0041] Specifically, in this step, based on the constructed clustered event response parameter dataset and the established causal Bayesian network structure, the conditional probabilities between nodes are learned using Bayesian estimation methods to quantify the degree of dependence between nodes and form a conditional probability table.

[0042] Step S60: Test and optimize the Bayesian network diagnostic model multiple times.

[0043] Step S70: Determine whether the diagnostic accuracy of the final trained Bayesian network diagnostic model meets the requirements; if yes, the development of the nuclear power plant accident diagnostic model is completed; if no, continue to adjust the parameters to optimize the model until the diagnostic accuracy meets the requirements.

[0044] Specifically, after completing the construction of the Bayesian network diagnostic model and forming the corresponding conditional probability table in step S40, the Bayesian network diagnostic model is tested and optimized multiple times until the diagnostic accuracy of the trained Bayesian network diagnostic model meets the requirements, and finally a complete nuclear power plant accident precision diagnosis system is formed.

[0045] This invention provides a method for developing a nuclear power plant accident diagnosis model. The method generates a nuclear power plant accident dataset, then selects valuable feature parameters; clusters the time series of the selected feature parameters; sets the topology of the Bayesian network diagnosis model and trains it; tests the performance of the Bayesian network diagnosis model; and finally, iteratively optimizes the model to improve the diagnostic effect.

[0046] Based on the Bayesian network diagnostic model constructed in this invention, this invention provides a method for diagnosing nuclear power plant accidents.

[0047] like Figure 3 As shown, in a preferred embodiment, the nuclear power plant accident diagnosis method includes the following steps: Step S31: Call the Bayesian network diagnostic model.

[0048] The Bayesian network diagnostic model is designed using the aforementioned nuclear power plant accident diagnostic model development method.

[0049] Step S32: Obtain real-time monitoring feature operation data.

[0050] Step S33: Use the K-Means clustering algorithm to discretize the feature running data to obtain grouped data.

[0051] Step S34: Input the grouped data into the Bayesian network diagnostic model for calculation to obtain the posterior probability distribution of the accident type.

[0052] In some embodiments, inputting grouped data into a Bayesian network diagnostic model for inference to obtain the posterior probability distribution of accident types includes: inputting grouped data into a Bayesian network diagnostic model; based on the grouped data and combined with the causal Bayesian network structure of the Bayesian network diagnostic model, using a Bayesian estimation method to learn the parameters of the conditional probabilities between nodes to form a conditional probability table; and using a variable elimination algorithm to perform diagnostic inference based on the conditional probability table and the causal Bayesian network structure to obtain the posterior probability of the accident type.

[0053] Specifically, based on the determined conditional probability table and causal Bayesian network structure, the variable elimination algorithm is used to achieve efficient diagnostic reasoning for various types of nuclear power plant accidents. After the characteristic operation data obtained by real-time monitoring using the K-Means clustering algorithm is discretized and preprocessed, it is then input as evidence variables into the Bayesian network diagnostic model to quickly calculate the posterior probability distribution of each potential accident type, thereby achieving accurate identification and location of nuclear power plant accident types.

[0054] refer to Figure 4 The present invention also provides a nuclear power plant accident diagnosis system.

[0055] like Figure 4 As shown, the nuclear power plant accident diagnosis system includes: Model invocation unit 401 is used to invoke the Bayesian network diagnostic model. This Bayesian network diagnostic model is a model designed using the aforementioned nuclear power plant accident diagnostic model development method.

[0056] The real-time data acquisition unit 402 is used to acquire real-time monitoring feature operation data.

[0057] Data clustering unit 403 is used to discretize the feature running data using the K-Means clustering algorithm to obtain grouped data.

[0058] The accident type diagnosis unit 404 is used to input grouped data into the Bayesian network diagnosis model for calculation to obtain the posterior probability distribution of the accident type.

[0059] This invention develops the aforementioned Bayesian network diagnostic model and uses this model to diagnose nuclear power plant accidents. This nuclear power plant accident diagnosis method can quickly and accurately identify the types of faults in advance, avoid reactor trips or reactor failures, or reduce the duration of unplanned downtime caused by accidents, thereby reducing power generation losses; it also reduces the risk of accident escalation, reduces equipment repair or replacement costs; reduces the manpower required for manual inspection, and lowers the on-site operation costs for maintenance personnel; and it can fully meet the real-time and reliability requirements of nuclear power plant spare parts management.

[0060] Specifically, the specific coordination and operation process between the various units in the nuclear power plant accident diagnosis system can be referred to the above-mentioned nuclear power plant accident diagnosis method, and will not be repeated here.

[0061] Furthermore, an electronic device of the present invention includes a memory and a processor; the memory is used to store a computer program; the processor is used to execute the computer program to implement the nuclear power plant accident diagnosis method as described above. Specifically, according to embodiments of the present invention, the processes described above with reference to the flowchart can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. In such embodiments, when the computer program is downloaded, installed, and executed by an electronic device, it performs the functions defined above in the methods of the embodiments of the present invention. The electronic device in the present invention can be a terminal such as a laptop, desktop computer, tablet computer, or smartphone, or it can be a server.

[0062] Furthermore, one type of storage medium of the present invention stores a computer program thereon, which, when executed by a processor, implements the nuclear power plant accident diagnosis method described above. Specifically, it should be noted that the storage medium described above in the present invention can be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In the present invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present invention, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, wherein computer-readable program code is carried. The transmitted data signal can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. The computer-readable signal medium can also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0063] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device.

[0064] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.

[0065] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0066] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in random access memory (RAM), main memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0067] The above embodiments are only for illustrating the technical concept and features of the present invention, and are intended to enable those skilled in the art to understand the content of the present invention and implement it accordingly. They do not limit the scope of protection of the present invention. All equivalent changes and modifications made within the scope of the claims of the present invention should fall within the scope of the claims of the present invention.

Claims

1. A method for developing a nuclear power plant accident diagnosis model, characterized in that, include: Generate a dataset of nuclear power plant accidents; The nuclear power plant accident dataset is filtered for characteristic operating parameters to obtain the filtered characteristic parameters. Time series clustering is performed on the filtered feature parameters to obtain a clustered accident response parameter dataset; Construct a Bayesian network diagnostic model; Based on the clustered accident response parameter dataset, and combined with the Bayesian network diagnostic model, probability parameters are learned to form a conditional probability table. The Bayesian network diagnostic model was tested and optimized multiple times. Determine whether the diagnostic accuracy of the final trained Bayesian network diagnostic model meets the requirements; If yes, then the development of the nuclear power plant accident diagnosis model will be completed; if not, then the parameters will continue to be adjusted to optimize the model until the diagnostic accuracy meets the requirements.

2. The method for developing a nuclear power plant accident diagnosis model according to claim 1, characterized in that, The generated nuclear power plant accident dataset includes: Obtain typical accident conditions and other accident conditions of nuclear power plants; The nuclear power plant system simulation model was used to simulate and calculate the typical accident conditions and other accident conditions to obtain the dynamic response characteristics of the characteristic operating parameters. An accident response parameter dataset containing time-series variation curves is generated based on the dynamic response characteristics of the aforementioned characteristic operating parameters; the accident response parameter dataset is the nuclear power plant accident dataset.

3. The method for developing a nuclear power plant accident diagnosis model according to claim 1, characterized in that, The process of filtering the nuclear power plant accident dataset for characteristic operating parameters to obtain the filtered characteristic parameters includes: Acquire expert experience and knowledge; Based on the expert experience and knowledge, parameters related to nuclear power plant accident diagnosis are selected from the nuclear power plant accident dataset to obtain the filtered feature parameters.

4. The method for developing a nuclear power plant accident diagnosis model according to claim 1, characterized in that, The step of performing time-series clustering on the filtered feature parameters to obtain a clustered accident response parameter dataset includes: The K-Means clustering algorithm is used to perform time series clustering on the filtered feature parameters to obtain the clustered accident response parameter dataset.

5. The method for developing a nuclear power plant accident diagnosis model according to claim 4, characterized in that, The K-Means clustering algorithm is used to perform time-series clustering on the filtered feature parameters to obtain the clustered accident response parameter dataset, which includes: The K-Means clustering algorithm is used to group the filtered feature parameters, and the discrete data of the feature parameters output according to a set interval are clustered. The steady-state data are grouped according to the set fluctuation range to form the clustered accident response parameter dataset.

6. The method for developing a nuclear power plant accident diagnosis model according to claim 1, characterized in that, The construction of the Bayesian network diagnostic model includes: To obtain the physical characteristics of the thermal-hydraulic and system control systems of nuclear power plants; Gain access to expert engineering experience and fault analysis knowledge; The topology of the Bayesian network diagnostic model is constructed based on the physical characteristics of the nuclear power plant's thermal-hydraulic and system control, the expert's engineering experience, and fault analysis knowledge.

7. The method for developing a nuclear power plant accident diagnosis model according to claim 1, characterized in that, The constructed Bayesian network diagnostic model is a hierarchical serial reasoning model constructed using a hierarchical serial reasoning pattern.

8. The method for developing a nuclear power plant accident diagnosis model according to claim 7, characterized in that, The reasoning process of the Bayesian network diagnostic model is as follows: Using an accident type Bayesian network model, the category of the current accident is initially determined, and a prior probability distribution of the accident type is established. Based on the prior probability distribution of the accident type, the corresponding Bayesian network model for accident location is dynamically invoked to locate the accident and obtain the accident location information. Based on the determined accident type and location information, the corresponding accident occurrence loop Bayesian network model is invoked for precise localization.

9. The method for developing a nuclear power plant accident diagnosis model according to claim 1, characterized in that, The step of learning probability parameters by combining the Bayesian network diagnostic model to form a conditional probability table includes: Based on the causal Bayesian network structure of the Bayesian network diagnostic model, the conditional probabilities between nodes are learned using the Bayesian estimation method to form the conditional probability table.

10. A method for diagnosing nuclear power plant accidents, characterized in that, include: The Bayesian network diagnostic model is invoked; the Bayesian network diagnostic model is a model designed using the nuclear power plant accident diagnostic model development method according to any one of claims 1-9; Obtain real-time monitoring data on key operational features; The K-Means clustering algorithm is used to discretize the feature data to obtain grouped data; The grouped data is input into the Bayesian network diagnostic model for calculation to obtain the posterior probability distribution of the accident type.

11. The nuclear power plant accident diagnosis method according to claim 10, characterized in that, The step of inputting the grouped data into a Bayesian network diagnostic model for inference to obtain the posterior probability distribution of the accident type includes: The grouped data is input into the Bayesian network diagnostic model; Based on the grouped data and combined with the causal Bayesian network structure of the Bayesian network diagnostic model, the Bayesian estimation method is used to learn the parameters of the conditional probabilities between nodes to form a conditional probability table. Based on the conditional probability table and the causal Bayesian network structure, a variable elimination algorithm is used for diagnostic reasoning to obtain the posterior probability of the accident type.

12. A nuclear power plant accident diagnosis system, characterized in that, include: The model invocation unit is used to invoke the Bayesian network diagnostic model; the Bayesian network diagnostic model is a model designed using the nuclear power plant accident diagnostic model development method according to any one of claims 1-9; The real-time data acquisition unit is used to acquire real-time monitoring feature operation data; The data clustering unit is used to discretize the feature running data using the K-Means clustering algorithm to obtain grouped data; The accident type diagnosis unit is used to input the grouped data into the Bayesian network diagnosis model for calculation to obtain the posterior probability distribution of the accident type.

13. A storage medium, characterized in that, The storage medium stores a computer program adapted for loading by a processor to perform the steps of the nuclear power plant accident diagnosis method as described in claim 10 or 11.

14. An electronic device, characterized in that, It includes a memory and a processor, wherein the memory stores a computer program, and the processor executes the steps of the nuclear power plant accident diagnosis method as described in claim 10 or 11 by calling the computer program stored in the memory.

Images