Nuclear power plant accident prediction method and device
By establishing accident analysis models and prediction models, predicting the accident status of nuclear power plants, the problem of the current slow feedback of nuclear power plants is solved, and the accident handling efficiency and safety of nuclear power plants are improved.
Patent Information
- Application Number
- CN202510125386.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-26
- Publication Date
- 2025-06-03
AI Technical Summary
After abnormal situations occur, the feedback from the current nuclear power plant is slow, which may lead to missing the optimal time for mitigation and affecting the safe operation of the nuclear power plant.
By establishing an accident analysis model, generating a sample database, and combining multi-layer neural network algorithms to establish an accident prediction model, using this model to predict the accident status of a nuclear power plant, providing a reference for decision-making by accident handlers.
It improves the efficiency of accident handling, prevents nuclear power units from developing towards more serious accidents, and enhances the safety of nuclear power plants.
Smart Images
Figure CN120087603A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of nuclear power plant accident handling, and particularly relates to a method and device for predicting nuclear power plant accidents. Background Art
[0002] A nuclear power plant is a very complex machine. Especially for existing pressurized water reactor nuclear power plants, in addition to the nuclear steam supply system and the thermal system, in order to ensure safety, a number of dedicated safety facilities are also provided. Coupled with the auxiliary systems and support systems of the above systems, the total number of systems in a nuclear power plant far exceeds one hundred. As long as one of the subsystems or even one of the components fails, the nuclear power plant will deviate from the normal operating conditions. Different events and accidents have different symptoms, and different operating procedures need to be adopted to ensure the safe shutdown and cooling of the reactor. Once an accident occurs in a nuclear power plant, the handling process often faces many difficulties such as tight time, great responsibility, and many changes. The current nuclear power plant adopts an accident handling method of "detection - response - handling", that is, when the nuclear power plant instruments detect abnormal conditions, the control system makes corresponding responses, and then the accident handlers take corresponding measures based on these responses. This results in a slow feedback between the occurrence of abnormalities and the taking of measures, which may miss the best time to mitigate the accident, and further cause the nuclear power unit to develop into a more serious accident, affecting the safe operation of the nuclear power plant.
[0003] In recent years, newly emerging technologies and new knowledge, such as artificial intelligence and big data, can mine and analyze massive amounts of data, master the operating laws of accident conditions, and predict the trend of accident conditions in advance. If relevant information on accident prediction can be provided to personnel at all levels of accident handling (such as nuclear power emergency commanders, emergency technical support personnel, reactor operators, etc.) during the accident handling process, it can provide decision-making references, bases, and technical support for accident handlers, enabling accident handlers to intervene in advance, prevent the nuclear power unit from developing into a more serious accident, and improve the safety of the nuclear power plant.
[0004] The existing patent CN118094379B discloses a method and system for diagnosing and predicting oil well faults through multi-factor fusion analysis. The method includes the following steps: S1: Collect production data, historical maintenance records, geological information, and surrounding environment data of the oil well; S2: Clean and standardize the data collected in S1, and extract key information through feature engineering; S3: Integrate and optimize the key information extracted in S2 through a preset hybrid algorithm framework; S4: Use a long short-term memory network to construct a fault diagnosis and prediction model; S5: Predict the probability and type of fault occurrence; S6: Regularly adjust and optimize the fusion analysis framework in S3 and the prediction model parameters in S4.
[0005] The existing patent CN118484763A discloses a fault prediction method based on an AI question-and-answer system, and the method includes the following steps: based on real-time sensor data and historical fault records, a hidden Markov model is used for state sequence analysis to identify the hidden transition paths of the device state, analyze the probability distribution of state transitions, and generate a state transition sequence and its probability distribution information.
[0006] In summary, neither of the above two existing patents solves the problem that the feedback between the occurrence of an anomaly and the taking of measures in the current nuclear power plant is slow, which may lead to missing the best time to mitigate the accident. Summary of the Invention
[0007] Based on the above technical problems, the present invention proposes a method and device for predicting nuclear power plant accidents, which solves the problem that the feedback between the occurrence of an anomaly and the taking of measures in the current nuclear power plant is slow, which may lead to missing the best time to mitigate the accident.
[0008] To achieve the above object, the present invention proposes a method for predicting nuclear power plant accidents.
[0009] A method for predicting nuclear power plant accidents includes:
[0010] Establish an accident analysis model for the target nuclear power plant;
[0011] Generate a sample database for the target nuclear power plant by using the accident analysis model;
[0012] Establish an accident prediction model for the target nuclear power plant;
[0013] Train the accident prediction model based on the sample database;
[0014] Predict the accident state of the target nuclear power plant by using the trained accident prediction model.
[0015] Further, establishing an accident analysis model for the target nuclear power plant includes:
[0016] Select an accident analysis program and use the accident analysis program to establish an accident analysis model for the target nuclear power plant.
[0017] Further, generating a sample database for the target nuclear power plant by using the accident analysis model includes:
[0018] Obtain the typical accident sequences of the target nuclear power plant;
[0019] Input the typical accident sequences into the accident analysis model to generate a sample database for the target nuclear power plant.
[0020] Further, obtaining the typical accident sequences of the target nuclear power plant includes:
[0021] Determine the typical accident sequences of the target nuclear power plant based on the FSAR accident analysis results and the accident handling procedure steps of the target nuclear power plant.
[0022] Further, establish an accident prediction model for the target nuclear power plant, including:
[0023] Establish an accident prediction model using a multi-layer neural network algorithm.
[0024] Further, train the accident prediction model based on a sample database, including:
[0025] Read the safety monitoring parameter data of the target nuclear power plant from the sample database. The safety monitoring parameter data includes one or more of the wide-range water level of the pressure vessel, the average temperature at the core outlet, the core subcooling margin, the primary loop temperature, the primary loop pressure, the pressure reduction rate of the pressurizer, the pressurizer water level, the pressurizer pressure, the steam generator water level, the steam generator pressure, the containment pressure, the spent fuel pool level, and the spent fuel pool temperature;
[0026] Perform data preprocessing on the safety monitoring parameter data and generate corresponding training data and test data;
[0027] Train the accident prediction model using the training data;
[0028] Test the effectiveness of the trained accident prediction model using the test data.
[0029] Further, it also includes:
[0030] Optimize the accident prediction model through preset model intervention conditions.
[0031] Further, use the trained accident prediction model to predict the accident state of the target nuclear power plant, including:
[0032] Use the trained accident prediction model to predict and output the safety monitoring parameter data of the target nuclear power plant in real time;
[0033] Generate corresponding prompt information for the output safety monitoring parameter data.
[0034] To achieve the above object, the present invention proposes a prediction device for nuclear power plant accidents.
[0035] A prediction device for nuclear power plant accidents, including:
[0036] A first establishment module for establishing an accident analysis model of the target nuclear power plant;
[0037] A generation module for generating a sample database of the target nuclear power plant using the accident analysis model;
[0038] The second establishment module is used to establish an accident prediction model for the target nuclear power plant;
[0039] The training module is used to train the accident prediction model based on the sample database;
[0040] The prediction module is used to predict the accident state of the target nuclear power plant by using the trained accident prediction model.
[0041] Furthermore, the first establishment module is used for:
[0042] Select an accident analysis program and use the accident analysis program to establish an accident analysis model for the target nuclear power plant.
[0043] Furthermore, the generation module is used for:
[0044] Obtain the typical accident sequences of the target nuclear power plant;
[0045] Input the typical accident sequences into the accident analysis model to generate a sample database for the target nuclear power plant.
[0046] Furthermore, obtaining the typical accident sequences of the target nuclear power plant includes:
[0047] Determine the typical accident sequences of the target nuclear power plant according to the FSAR accident analysis results and the accident handling procedure steps of the target nuclear power plant.
[0048] Furthermore, the second establishment module is used for:
[0049] Use the multi-layer neural network algorithm to establish an accident prediction model.
[0050] Furthermore, the training module is used for:
[0051] Read the safety monitoring parameter data of the target nuclear power plant from the sample database. The safety monitoring parameter data includes one or more of the wide-range water level of the pressure vessel, the average temperature at the core outlet, the core subcooling margin, the primary loop temperature, the primary loop pressure, the pressure reduction rate of the pressurizer, the pressurizer water level, the pressurizer pressure, the steam generator water level, the steam generator pressure, the containment pressure, the spent fuel pool level, and the spent fuel pool temperature;
[0052] Perform data preprocessing on the safety monitoring parameter data and generate corresponding training data and test data;
[0053] Train the accident prediction model by using the training data;
[0054] Test the effect of the trained accident prediction model by using the test data.
[0055] Furthermore, it further includes an optimization module for:
[0056] Optimize the accident prediction model according to the preset model intervention conditions.
[0057] Furthermore, the prediction module is used for:
[0058] Use the trained accident prediction model to predict and output the safety monitoring parameter data of the target nuclear power plant in real time;
[0059] Generate corresponding prompt information for the output safety monitoring parameter data.
[0060] Based on the above technical solutions, the present invention has at least the following beneficial effects:
[0061] 1. A prediction method for nuclear power plant accidents proposed by the present invention uses an accident analysis model to generate a sample database of the target nuclear power plant, provides training data for the prediction model, establishes an accident prediction model by combining intelligent algorithms, and uses the accident prediction model to predict the accident state of the target nuclear power plant, providing accident prediction information for personnel at all levels in the accident handling of pressurized water reactor nuclear power plants, thereby providing decision-making reference, basis, and technical support for accident handlers, enabling accident handlers to intervene in advance, effectively improving the accident handling efficiency, preventing the nuclear power unit from developing into a more serious accident, and improving the safety of the nuclear power plant.
[0062] 2. The present invention obtains the typical accident sequences of the target nuclear power plant, inputs the typical accident sequences into the accident analysis model to generate a sample database of the target nuclear power plant. This method can ensure that the established nuclear power plant model can accurately reflect the actual operation state of the target nuclear power plant, making the generated sample data have high authenticity and reliability, providing a reliable training basis for the accident prediction model, and thus significantly improving the prediction accuracy of the accident prediction model. BRIEF DESCRIPTION OF THE DRAWINGS
[0063] The accompanying drawings forming a part of this invention are used to provide a further understanding of the invention. The schematic embodiments and descriptions thereof of the invention are used to explain the invention and do not constitute an improper limitation of the invention. In the drawings:
[0064] Figure 1 is a flowchart of a prediction method for nuclear power plant accidents according to an embodiment of the present invention;
[0065] Figure 2 is a flowchart of training the accident prediction model based on the sample database in an embodiment of the present invention;
[0066] Figure 3 is a schematic diagram of a prediction device for nuclear power plant accidents according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS
[0067] It should be noted that, without conflict, the embodiments in the present invention and the features in the embodiments can be combined with each other. The present invention will be described in detail below with reference to the drawings and in combination with the embodiments.
[0068] The present invention will be further described in detail below with reference to specific embodiments, and these embodiments should not be construed as limiting the scope claimed by the present invention.
[0069] Embodiment
[0070] To solve the problem that the feedback between the occurrence of an anomaly and the taking of measures in existing nuclear power plants is relatively slow, which may lead to missing the best time to mitigate the accident, the present invention proposes a method and device for predicting nuclear power plant accidents.
[0071] To achieve the above object, the present invention proposes a method for predicting nuclear power plant accidents.
[0072] As Figure 1 shown in the flowchart of a method for predicting nuclear power plant accidents according to an embodiment of the present invention, the method includes the following steps:
[0073] S1. Establish an accident analysis model for the target nuclear power plant.
[0074] Since it is difficult to obtain the operation data under nuclear power plant accidents and the data volume is extremely small, which is insufficient to support the training of the accident prediction model, in this embodiment, computer simulation technology is used to simulate the operation of nuclear power plant accidents. Further, establishing an accident analysis model for the target nuclear power plant includes: selecting an accident analysis program and using the accident analysis program to establish an accident analysis model for the target nuclear power plant.
[0075] Taking a certain nuclear power plant as an example in this embodiment, the accident analysis program RELAP5 is selected as the calculation kernel to establish an accident analysis model for the nuclear power plant. The modeling scope includes but is not limited to the following systems: reactor core, primary coolant system, containment, spent fuel pool, engineered safety features system, secondary circuit, etc. After establishing the accident analysis model, typical accidents of the nuclear power plant (such as a break in the primary loop, a rupture of the steam generator transfer heat pipe, a rupture of the main steam pipe, etc.) are selected, and the established accident analysis model is debugged and subjected to steady-state and transient verification tests to ensure that the established nuclear power plant model can correctly reflect the actual state of the nuclear power plant. When conducting the debugging and verification tests, it is required that the deviation between the calculated steady-state data and the design value of the nuclear power plant does not exceed 1%, and the accident transients within the design basis are consistent with the trends in the safety analysis report to verify the correctness of the accident analysis model.
[0076] S2. Generate a sample database for the target nuclear power plant by using the accident analysis model.
[0077] Further, generating a sample database for the target nuclear power plant by using the accident analysis model includes:
[0078] S201, obtain the typical accident sequences of the target nuclear power plant.
[0079] Further, obtaining the typical accident sequences of the target nuclear power plant includes: determining the typical accident sequences of the target nuclear power plant based on the FSAR accident analysis results and the accident handling procedure steps of the target nuclear power plant. FSAR (Final Safety Analysis Report) is an analysis document of the nuclear power plant, which contains trend charts, parameter values, etc. of various typical accidents. The accident handling procedure is a procedural document based on which the nuclear power plant handles accidents, and it contains all the handling steps.
[0080] Specifically, based on the FSTR accident analysis results and the accident handling procedure steps, select the typical accident sequences, and by introducing model uncertainties, derive a series of similar accident sequences for each accident sequence. Model uncertainties mainly come from differences in system structure parameters and the severity of the initiating accident. For example, in the analysis of a primary loop break accident, a series of accident sequences can be obtained by changing the break size, break location, break occurrence time, the execution time of a certain key operation step, etc.
[0081] S202, input the typical accident sequences into the accident analysis model to generate a sample database of the target nuclear power plant.
[0082] Input the typical accident sequences into the accident analysis model, and use the accident analysis model to perform a large number of calculations on various accident scenarios to generate the required sample database for the training of the prediction model.
[0083] S3, establish an accident prediction model for the target nuclear power plant.
[0084] In this embodiment, a multi-layer neural network algorithm is used to develop a prediction model for predicting the accident status of the nuclear power plant in the next period of time. The multi-layer neural network structure is a neural network model containing multiple hidden layers. Since the mechanism functions of the neuron nodes in each hidden layer are non-linear, there is a non-linear representation between the input value and the output value of each neuron node. Therefore, the multi-layer neural network model has a strong non-linear mapping learning ability and is suitable for complex system problems. In this step, it is necessary to complete the definition of the network structure of the multi-layer neural network. After establishing the accident prediction model of the target nuclear power plant, train the prediction model with training data and verify the accuracy of the prediction model with test data.
[0085] S4, train the accident prediction model based on the sample database.
[0086] Further, as Figure 2 shown, training the accident prediction model based on the sample database includes:
[0087] S401. Read the safety monitoring parameter data of the target nuclear power plant from the sample database.
[0088] The safety monitoring parameter data in this embodiment refers to the parameter data that has a relatively large impact on the safety of the nuclear power plant, mainly including key parameters such as the core of the nuclear power plant, the main system, the steam generator, the containment, the spent fuel pool, the action signals of the dedicated safety equipment, the power status, and the main pump status that can be monitored.
[0089] Further, the safety monitoring parameter data includes one or more of the wide-range water level of the pressure vessel, the average temperature at the core outlet, the core subcooling margin, the primary loop temperature, the primary loop pressure, the pressure reduction rate of the pressurizer, the pressurizer water level, the pressurizer pressure, the steam generator water level, the steam generator pressure, the containment pressure, the spent fuel pool liquid level, and the spent fuel pool temperature.
[0090] S402. Perform data preprocessing on the safety monitoring parameter data and generate corresponding training data and test data.
[0091] The preprocessing process includes standardizing the safety monitoring parameter data, marking features and labels, and generating corresponding training data and test data. In this embodiment, the features correspond to the average value, median, variance, standard deviation, kurtosis coefficient, etc. of a group or a class of safety monitoring parameter data.
[0092] S403. Train the accident prediction model using the training data.
[0093] S404. Test the effect of the trained accident prediction model using the test data.
[0094] Considering that the development of the nuclear accident state not only has a strong trend but also has event shocks and various prediction scheme trade-offs, in another embodiment of the present invention, after performing step S4, a preset intervention condition is introduced to adjust the prediction result, integrating the problem background knowledge and prediction analysis. For example, the preset model intervention condition can be by changing the accident progress, changing the operation time of a certain device during accident handling, etc. If the model prediction effect does not meet the requirements after adjusting the prediction result by the preset intervention condition, methods such as optimizing the network structure and the optimizer can be further used to improve the model training efficiency and the model prediction accuracy, and improve the rationality and practicality of trend prediction. S5. Predict the accident state of the target nuclear power plant using the trained accident prediction model.
[0095] Further, predicting the accident state of the target nuclear power plant using the trained accident prediction model includes:
[0096] S501. Use the trained accident prediction model to predict and output the safety monitoring parameter data of the target nuclear power plant in real time.
[0097] S502, generate corresponding prompt information for the output safety monitoring parameter data.
[0098] As shown in Table 1, the corresponding prompt information generated by using the safety monitoring parameter data predicted and output by the accident prediction model is presented. Among them, Prompt Information 1 represents the tag number of the parameter, and the value is the limit value of the parameter value. When this limit value is reached, the corresponding Prompt Information 2 is generated.
[0099] Table 1 Accident Prediction Safety Monitoring Parameters and Prompt Information
[0100]
[0101] To achieve the above object, the present invention proposes a prediction device for nuclear power plant accidents.
[0102] As Figure 3 shown in
[0103] The first establishment module 31 is used to establish an accident analysis model for the target nuclear power plant.
[0104] Further, the first establishment module 31 is used for:
[0105] Select an accident analysis program and use the accident analysis program to establish an accident analysis model for the target nuclear power plant.
[0106] The generation module 32 is used to generate a sample database for the target nuclear power plant by using the accident analysis model.
[0107] Further, the generation module 32 is used for:
[0108] Obtain the typical accident sequences of the target nuclear power plant.
[0109] Input the typical accident sequences into the accident analysis model to generate a sample database for the target nuclear power plant.
[0110] Further, obtaining the typical accident sequences of the target nuclear power plant includes:
[0111] Determine the typical accident sequences of the target nuclear power plant according to the FSAR accident analysis results and accident handling procedure steps of the target nuclear power plant.
[0112] The second establishment module 33 is used to establish an accident prediction model for the target nuclear power plant.
[0113] Further, the second establishment module 33 is used for:
[0114] An accident prediction model is established using a multi-layer neural network algorithm.
[0115] A training module 34, configured to train the accident prediction model based on a sample database.
[0116] Further, the training module 34 is configured to:
[0117] Read safety monitoring parameter data of a target nuclear power plant from the sample database, where the safety monitoring parameter data includes one or more of wide-range water level of a pressure vessel, average temperature at the core outlet, core subcooling margin, primary loop temperature, primary loop pressure, pressurizer cooling rate, pressurizer water level, pressurizer pressure, steam generator water level, steam generator pressure, containment pressure, spent fuel pool level, and spent fuel pool temperature.
[0118] Perform data preprocessing on the safety monitoring parameter data, and generate corresponding training data and test data.
[0119] Train the accident prediction model using the training data.
[0120] Test the effect of the trained accident prediction model using the test data.
[0121] Further, it further includes an optimization module 36, configured to:
[0122] Optimize the accident prediction model through preset model intervention conditions and / or optimizers.
[0123] A prediction module 35, configured to predict the accident state of a target nuclear power plant using the trained accident prediction model.
[0124] Further, the prediction module 35 is configured to:
[0125] Use the trained accident prediction model to predict and output in real time the safety monitoring parameter data of the target nuclear power plant.
[0126] Generate corresponding prompt information for the output safety monitoring parameter data.
[0127] It should be understood that an accident prediction device for a nuclear power plant is consistent with the description of an accident prediction method embodiment corresponding thereto, so this embodiment will not be elaborated herein.
[0128] In summary, as can be seen from the above description, the above embodiments of the present invention achieve the following technical effects:
[0129] 1. A method and device for predicting nuclear power plant accidents proposed by the present invention generate a sample database of a target nuclear power plant using an accident analysis model, provide training data for a prediction model, establish an accident prediction model by combining intelligent algorithms, and use the accident prediction model to predict the accident state of the target nuclear power plant, providing accident prediction information for personnel at all levels of the pressurized water reactor nuclear power plant accident handling, thereby providing decision-making reference, basis, and technical support for accident handlers, enabling accident handlers to intervene in advance, effectively improving the accident handling efficiency, preventing the nuclear power unit from developing into a more serious accident, and improving the safety of the nuclear power plant.
[0130] 2. The present invention generates a sample database of a target nuclear power plant by obtaining typical accident sequences of the target nuclear power plant and inputting the typical accident sequences into an accident analysis model. This method can ensure that the established nuclear power plant model can accurately reflect the actual operating state of the target nuclear power plant, making the generated sample data have high authenticity and reliability, providing a reliable training basis for the accident prediction model, and thus significantly improving the prediction accuracy of the accident prediction model.
[0131] The above are only the preferred embodiments of the present invention and are not used to limit the present invention. For those skilled in the art, the present invention can have various changes and modifications. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.
[0132] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the existence of additional identical elements in the process, method, article or device comprising the element.
[0133] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, which can be specifically implemented in any computer-readable medium for use by an instruction execution system, device or equipment (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, device or equipment), or used in combination with these instruction execution systems, devices or equipment.
[0134] It should be understood that various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like.
[0135] It should be noted that in the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, without contradiction, those skilled in the art can combine and combine the different embodiments or examples described in this specification and the features of different embodiments or examples.
Claims
1. A method for predicting nuclear power plant accidents, characterized in that: include: Establish accident analysis models for target nuclear power plants; Generating a sample database of the target nuclear power plant using the accident analysis model; Establishing an accident prediction model for the target nuclear power plant; Training the accident prediction model based on the sample database; The trained accident prediction model is used to predict the accident status of the target nuclear power plant.
2. The method according to claim 1, characterized in that Establish an accident analysis model for the target nuclear power plant, including: An accident analysis program is selected, and an accident analysis model of the target nuclear power plant is established using the accident analysis program.
3. The method according to claim 1, characterized in that Generating a sample database of the target nuclear power plant using the accident analysis model includes: Obtaining a typical accident sequence of the target nuclear power plant; The typical accident sequence is input into the accident analysis model to generate a sample database of the target nuclear power plant.
4. The method according to claim 3, characterized in that Obtain a typical accident sequence of the target nuclear power plant, including: According to the FSAR accident analysis results and accident handling procedure steps of the target nuclear power plant, a typical accident sequence of the target nuclear power plant is determined.
5. The method according to any one of claims 1 to 4, characterized in that: Establishing an accident prediction model for the target nuclear power plant, including: An accident prediction model is established using a multi-layer neural network algorithm.
6. The method according to claim 5, characterized in that Training the accident prediction model based on the sample database includes: Reading the safety monitoring parameter data of the target nuclear power plant from the sample database, the safety monitoring parameter data including one or more of the wide range water level of the pressure vessel, the average temperature of the core outlet, the core subcooling margin, the primary circuit temperature, the primary circuit pressure, the pressurizer cooling rate, the pressurizer water level, the pressurizer pressure, the steam generator water level, the steam generator pressure, the containment pressure, the spent pool liquid level, and the spent pool temperature; Performing data preprocessing on the safety monitoring parameter data and generating corresponding training data and test data; Using the training data to train the accident prediction model; The test data is used to test the effect of the trained accident prediction model.
7. The method according to claim 6, characterized in that Also includes: The accident prediction model is optimized through preset model intervention conditions.
8. The method according to claim 1, characterized in that: The accident state of the target nuclear power plant is predicted using the trained accident prediction model, including: Using the trained accident prediction model to predict and output the safety monitoring parameter data of the target nuclear power plant in real time; Generate corresponding prompt information for the output safety monitoring parameter data.
9. A prediction device for nuclear power plant accidents, characterized in that: include: The first building module is used to build an accident analysis model of the target nuclear power plant; A generation module, used to generate a sample database of the target nuclear power plant using the accident analysis model; A second establishing module is used to establish an accident prediction model for the target nuclear power plant; A training module, used for training the accident prediction model based on the sample database; The prediction module is used to predict the accident status of the target nuclear power plant using the trained accident prediction model.
10. The device according to claim 9, characterized in that The first establishing module is used to: An accident analysis program is selected, and an accident analysis model of the target nuclear power plant is established using the accident analysis program.
11. The device according to claim 9, characterized in that The generating module is used for: Obtaining a typical accident sequence of the target nuclear power plant; The typical accident sequence is input into the accident analysis model to generate a sample database of the target nuclear power plant.
12. The device according to claim 11, characterized in that Obtain a typical accident sequence of the target nuclear power plant, including: According to the FSAR accident analysis results and accident handling procedure steps of the target nuclear power plant, a typical accident sequence of the target nuclear power plant is determined.
13. The device according to any one of claims 9 to 12, characterized in that The second building module is used to: An accident prediction model is established using a multi-layer neural network algorithm.
14. The device according to claim 13, characterized in that The training module is used to: Reading the safety monitoring parameter data of the target nuclear power plant from the sample database, the safety monitoring parameter data including one or more of the wide range water level of the pressure vessel, the average temperature of the core outlet, the core subcooling margin, the primary circuit temperature, the primary circuit pressure, the pressurizer cooling rate, the pressurizer water level, the pressurizer pressure, the steam generator water level, the steam generator pressure, the containment pressure, the spent pool liquid level, and the spent pool temperature; Performing data preprocessing on the safety monitoring parameter data and generating corresponding training data and test data; Using the training data to train the accident prediction model; The test data is used to test the effect of the trained accident prediction model.
15. The device according to claim 14, characterized in that Also includes optimization modules for: The accident prediction model is optimized through preset model intervention conditions.
16. The device according to claim 9, characterized in that The prediction module is used to: Using the trained accident prediction model to predict and output the safety monitoring parameter data of the target nuclear power plant in real time; Generate corresponding prompt information for the output safety monitoring parameter data.