High-efficiency nuclear reactor safety margin assessment method and system, terminal equipment and computer program product

Abstract

The invention is suitable for the technical field of nuclear reactor safety of a nuclear power plant, and provides an efficient nuclear reactor safety margin evaluation method and system, terminal equipment and a computer program product, and the method comprises the following steps: constructing a training data set based on a sample data set of uncertainty parameters and time sequence data of key safety parameters; training the initial agent model based on the training data set to obtain a target agent model; the sequence overrun probability of the key safety parameters is efficiently calculated through the target agent model; and calculating the shutdown risk of the nuclear reactor based on the sequence overrun probability and taking the shutdown risk as a safety margin assessment result of the nuclear reactor. According to the method provided by the invention, the sequence overrun probability of the key safety parameters is efficiently calculated, and the shutdown risk of the nuclear reactor is calculated based on the sequence overrun probability and is used as the safety margin evaluation result of the nuclear reactor, so that the safety margin of the nuclear reactor is efficiently, quickly and accurately evaluated.

Description

Technical Field

[0001] This application belongs to the technical field of nuclear reactor safety analysis in nuclear power plants, and particularly relates to an efficient method, system, terminal device, and computer program product for evaluating the safety margin of a nuclear reactor. Background Art

[0002] In the context of vigorously developing "dual carbon" and clean energy, nuclear safety is the cornerstone of the sustainable development of nuclear energy. In the flexible operation and accident response of nuclear power plants, ensuring the safety and reliability of reactors, identifying potential risks, and predicting and warning in advance are crucial. In recent years, the Risk-Informed Safety Margin Characterization (RISMC) method, as an advanced nuclear safety assessment tool, has gradually become a key means to cope with complex operating environments and accident scenarios.

[0003] Currently, RISMC research mainly focuses on reactor design and severe accident scenarios, such as reactor power increase, nuclear power plant life extension, loss-of-coolant accident, etc., aiming to more realistically evaluate safety margins and related safety measures. However, in the application process, the RISMC method still faces challenges such as high computational resource consumption, complex calculation processes, and long safety margin assessment cycles, and it is difficult to meet the needs of nuclear power plants for rapid risk assessment and strategy decision-making during the operation stage.

[0004] Therefore, there is an urgent need for an efficient method for evaluating the safety margin of a nuclear reactor to improve the efficiency of margin assessment while ensuring prediction accuracy. Summary of the Invention

[0005] In view of this, the embodiments of this application provide an efficient method, system, terminal device, and computer program product for evaluating the safety margin of a nuclear reactor to achieve efficient, rapid, and accurate evaluation of the safety margin of a nuclear reactor.

[0006] The first aspect of the embodiments of this application provides an efficient method for evaluating the safety margin of a nuclear reactor, including: Constructing a training dataset based on a sample dataset of uncertainty parameters and time series data of key safety parameters; Training an initial surrogate model based on the training dataset to obtain a target surrogate model; Efficiently calculating the sequence exceedance probability of the key safety parameters using the target surrogate model; Calculating the shutdown risk of the nuclear reactor based on the sequence exceedance probability and using it as the evaluation result of the safety margin of the nuclear reactor.

[0007] In one implementation of the first aspect, the efficiently calculating the sequence exceedance probability of the key safety parameters using the target surrogate model includes: Sample the uncertainty parameters to obtain a sample set, where the sample set includes multiple samples; Calculate the key safety parameter values corresponding to each sample in the sample dataset through the target surrogate model; Based on the key safety parameter values, form a probability distribution of the key safety parameters, and calculate the sequence exceedance probability of the safety parameters.

[0008] In one implementation of the first aspect, calculating the reactor trip risk of the nuclear reactor based on the sequence exceedance probability and using it as the evaluation result of the safety margin of the nuclear reactor includes: Calculate the reactor trip risk SDF based on the following formula:

[0009] Where, is the probability of occurrence of abnormal event i, represents the kth sequence under abnormal event i, is the dynamic evolution sequence The probability of occurrence, is the probability that abnormal event i causes any one or more key safety parameters To exceed the safety threshold range, Is the key safety parameter The probability of being higher than the safety threshold range; Use the reactor trip risk as the evaluation result of the nuclear reactor safety margin.

[0010] In one implementation of the first aspect, after training the initial surrogate model with the training dataset to obtain the target surrogate model, it further includes: Based on one or more indicators among the mean square error, mean absolute error, mean absolute percentage error, coefficient of determination, and error exceedance rate of the feature selection of the key safety parameters, evaluate the applicability of the target surrogate model to the evaluation of the nuclear reactor safety margin; If the applicability of the target surrogate model to the evaluation of the nuclear reactor safety margin is less than the preset applicability threshold, further iteratively optimize the target surrogate model until the applicability of the target surrogate model to the evaluation of the nuclear reactor safety margin is not less than the preset applicability threshold.

[0011] In one implementation of the first aspect, constructing the training dataset based on the sample dataset of the uncertainty parameters and the time series data of the key safety parameters further includes: Sample the selected uncertainty parameters to generate an uncertainty sample set, where the uncertainty sample set includes multiple groups of sampling samples, and each group of sampling samples includes a value of each of the uncertainty parameters sampled at the same moment; Based on the self-programming simulation driver, batch embed each set of sampling samples in the uncertainty sample set into the input card corresponding to the target simulation model; Batch run the input card to obtain large-scale simulation results; Extract time series extreme value information from the large-scale simulation results, where the time series extreme value information includes time series extreme values of at least one key safety parameter, and one piece of time series extreme value information corresponds to one set of the sampling samples; Associate and store each set of sampling samples in the multiple sets of sampling samples with the corresponding time series extreme value information of the simulation output to construct a training data set.

[0012] Select system variables whose influence on the behavior and safety margin of the nuclear reactor system is greater than a preset influence threshold as uncertainty parameters; Select intervention control variables as uncertainty parameters.

[0013] In one implementation manner of the first aspect, the method further includes: Select a target task scenario based on the occurrence frequency and potential consequences of the event sequence; Establish an initial simulation model for the target task scenario; Simulate the target event sequence based on the initial simulation model to generate high-fidelity simulation data; Verify the accuracy of the initial simulation model based on the high-fidelity simulation data; If the accuracy of the initial simulation model is lower than the preset accuracy, return to the step of simulating the target event sequence based on the initial simulation model to generate high-fidelity simulation data; If the accuracy of the initial simulation model is not lower than the preset accuracy, use the initial simulation model as the target simulation model.

[0014] The second aspect of the embodiments of the present application provides an efficient nuclear reactor safety margin evaluation system, including: A training data set module for constructing a training data set based on a sample data set of uncertainty parameters and time series data of key safety parameters; An agent model training module for training an initial agent model based on the training data set to obtain a target agent model; A sequence overrun probability module for efficiently calculating the sequence overrun probability of the key safety parameters using the target agent model; A safety margin calculation module for calculating the shutdown risk of the nuclear reactor based on the sequence overrun probability and using it as the safety margin evaluation result of the nuclear reactor.

[0015] A third aspect of the embodiments of the present application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the efficient nuclear reactor safety margin evaluation method described in the first aspect are implemented.

[0016] A fourth aspect of the embodiments of the present application provides a computer program product, including a computer program. When the computer program is run, the efficient nuclear reactor safety margin evaluation method described in the first aspect is executed.

[0017] The beneficial effects of the first aspect of the embodiments of the present application are as follows: By constructing a training data set based on the sample data set of uncertainty parameters and the time series data of key safety parameters, and training an initial surrogate model based on the training data set to obtain a target surrogate model, then using the target surrogate model to efficiently calculate the sequence overrun probability of the key safety parameters, and finally calculating the reactor trip risk of the nuclear reactor based on the sequence overrun probability and using it as the safety margin evaluation result of the nuclear reactor. By efficiently calculating the sequence overrun probability of the key safety parameters and calculating the reactor trip risk of the nuclear reactor based on the sequence overrun probability as the safety margin evaluation result of the nuclear reactor, an efficient, fast, and accurate evaluation of the safety margin of the nuclear reactor is realized.

[0018] It can be understood that the beneficial effects of the above second aspect to fourth aspect can refer to the relevant descriptions in the above first aspect and will not be elaborated here. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0020] Figure 1 is a schematic diagram of the implementation process of the efficient nuclear reactor safety margin evaluation method provided by the embodiments of the present application; Figure 2 is a schematic diagram of the implementation process of the efficient nuclear reactor safety margin evaluation method provided by the embodiments of the present application; Figure 3 is a sample size change curve graph of the overrun probability and its Wilson confidence interval provided by the embodiments of the present application; Figure 4 is a schematic diagram of the implementation process of the efficient nuclear reactor safety margin evaluation method provided by the embodiments of the present application; Figure 5 is a change curve of the model training loss function provided by the embodiments of the present application; Figure 6 It is a prediction result diagram of the Transformer target proxy model provided by an embodiment of the present application; Figure 7 It is a prediction result diagram of the support vector machine target proxy model provided by an embodiment of the present application; Figure 8 It is a prediction result diagram of the multi-layer perceptron target proxy model provided by an embodiment of the present application; Figure 9 It is the distribution of key safety parameters when the power drops to 60% and the valve opening drops to 40% at 50% feed water flow rate provided by an embodiment of the present application; Figure 10 It is the distribution of key safety parameters when the power drops to 70% and the valve opening drops to 40% at 50% feed water flow rate provided by an embodiment of the present application; Figure 11 It is the distribution of key safety parameters when the power drops to 90% and the valve opening drops to 40% at 70% feed water flow rate provided by an embodiment of the present application; Figure 12 It is the distribution of key safety parameters when the power drops to 80% and the valve opening drops to 40% at 70% feed water flow rate provided by an embodiment of the present application; Figure 13 It is the reactor trip event tree at 70% feed water flow rate provided by an embodiment of the present application; Figure 14 It is the reactor trip event tree at 50% feed water flow rate provided by an embodiment of the present application; Figure 15 It is a schematic structural diagram of an efficient nuclear reactor safety margin assessment system provided by an embodiment of the present application; Figure 16 It is a schematic diagram of a computer device provided by an embodiment of the present application; Figure 17 It is a schematic diagram of a computer program product provided by an embodiment of the present application. Detailed implementation manners

[0021] In the following description, specific details such as specific system structures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present application. However, those skilled in the art should clearly understand that the present application can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present application.

[0022] It should be understood that, as used in the specification of the present application and the appended claims, the term "comprising" indicates the presence of the described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or their groups.

[0023] It should also be understood that the term "and / or" used in the specification of the present application and the appended claims refers to any combination and all possible combinations of one or more of the associated listed items, and includes these combinations.

[0024] As used in the specification of the present application and the appended claims, the term "if" can be interpreted as "when" or "once" or "in response to determining" or "in response to detecting" according to the context. Similarly, the phrase "if determined" or "if [the described condition or event] is detected" can be interpreted as meaning "once determined" or "in response to determining" or "once [the described condition or event] is detected" or "in response to detecting [the described condition or event]" according to the context.

[0025] In addition, in the description of the specification of the present application and the appended claims, the terms "first", "second", "third", etc. are only used for distinguishing descriptions and cannot be understood as indicating or implying relative importance.

[0026] The reference to "one embodiment" or "some embodiments" or the like described in the specification of the present application means that a specific feature, structure or characteristic described in connection with that embodiment is included in one or more embodiments of the present application. Thus, the statements "in one embodiment", "in some embodiments", "in other some embodiments", "in still other embodiments", etc. that appear in different places in this specification do not necessarily all refer to the same embodiment, but mean "one or more but not all embodiments", unless otherwise specifically emphasized in other ways. The terms "comprising", "including", "having" and their variants all mean "including but not limited to", unless otherwise specifically emphasized in other ways.

[0027] The present application provides a method for efficiently evaluating the safety margin of a nuclear reactor, which is used to achieve an efficient and rapid evaluation of the safety margin of the nuclear reactor. The method provided by the present application constructs a training data set based on a sample data set of uncertainty parameters and time series data of key safety parameters, trains an initial surrogate model based on the training data set to obtain a target surrogate model, then efficiently calculates the sequence overrun probability of the key safety parameters based on the target surrogate model, and finally calculates the shutdown risk of the nuclear reactor based on the sequence overrun probability and uses it as the evaluation result of the safety margin of the nuclear reactor. By efficiently calculating the sequence overrun probability of the key safety parameters and calculating the shutdown risk of the nuclear reactor based on the sequence overrun probability as the evaluation result of the safety margin of the nuclear reactor, an efficient, rapid and accurate evaluation of the safety margin of the nuclear reactor is achieved.

[0028] The method for efficiently evaluating the safety margin of a nuclear reactor provided by the embodiments of the present application can be applied to computer devices such as mobile phones, tablet computers, wearable devices, augmented reality (AR) / virtual reality (VR) devices, laptop computers, ultra-mobile personal computers (UMPCs), netbooks, personal digital assistants (PDAs), desktop computers, notebooks, palmtop computers, and cloud servers. The embodiments of the present application do not impose any restrictions on the specific types of computer devices.

[0029] In applications, various models in the method provided by the present application can be trained by machine learning. The method provided by the embodiments of the present application can be integrated into industrial application software related to nuclear power plants to facilitate industrial information and data processing, industrial data processing related to nuclear power plants, and industrial data analysis. The method provided by the present application improves the prediction accuracy of key safety margins through a data-driven surrogate model under complex operating conditions, thereby ensuring flexible, safe and reliable operation.

[0030] Expected prospects: 1) Provide operators with fast and quantifiable risk guidance decision support, and improve the operator's situation awareness ability. Effectively identify the operation path after anomalies, reduce unnecessary conservatism, and reduce dependence on expert judgment.

[0031] 2) Realize dynamic deduction of event sequences, intelligently predict and propose diversified disposal paths and new risk insights, and support the formation of optimal strategies.

[0032] As Figure 1 shown, the embodiments of the present application provide a method for efficiently evaluating the safety margin of a nuclear reactor, including: Step S1, construct a training dataset based on the sample dataset of uncertainty parameters and the time series data of key safety parameters.

[0033] Step S2, train an initial surrogate model based on the training dataset to obtain a target surrogate model.

[0034] In applications. After constructing the training dataset, select a suitable model architecture (such as a Transformer-based model, LSTM, SVM) according to task requirements, computational costs, and model applicability. To improve prediction accuracy, various strategies can be adopted to improve feature extraction and model adaptability, including increasing the network depth, optimizing the attention mechanism, and carefully selecting a suitable loss function (such as mean squared error, mean absolute error, cross-entropy) according to specific prediction tasks.

[0035] Data partitioning is crucial for accurately evaluating the generalization ability of the model. Common strategies include simple holdout validation, k-fold cross-validation, and leave-one-out cross-validation. Holdout validation is computationally efficient and suitable for large datasets, such as the commonly used 7:3 training set and validation set split, while k-fold cross-validation can provide a stable and robust estimate of the model on different data subsets but has a higher computational cost. The choice of data partitioning method should balance the requirements of accuracy and computational resources.

[0036] Optimization algorithms play a key role in ensuring efficient and stable model training. Gradient-based optimizers, such as stochastic gradient descent (SGD) or adaptive algorithms, such as Adam. SGD is simple and effective but may converge slowly or oscillate; adaptive optimizers usually offer faster convergence speeds and better stability, so they are more popular in complex deep learning tasks. To avoid overfitting during training, various regularization and generalization techniques can be adopted, such as L2 regularization, early stopping, and cross-validation. These methods help limit model complexity, prevent over-adjustment of parameters, and ensure stable model performance under various operating conditions.

[0037] Finally, determine the optimal combination of model parameters through hyperparameter optimization (such as grid search, random search, Bayesian optimization).

[0038] In applications, verify the effectiveness and robustness of the target surrogate model by using an independent test dataset to confirm its applicability to nuclear safety margin assessment.

[0039] Step S3, efficiently calculate the sequence exceedance probability of the key safety parameters using the target surrogate model.

[0040] Step S4, calculate the shutdown risk of the nuclear reactor based on the sequence exceedance probability and use it as the evaluation result of the safety margin of the nuclear reactor.

[0041] As shown Figure 2 In one embodiment, in step S3, using the target surrogate model to efficiently calculate the sequence overrun probability of the key safety parameters includes: Step S31: Sampling the uncertainty parameters to obtain a sample set, where the sample set includes multiple samples.

[0042] Step S32: Calculating the key safety parameter values corresponding to each sample in the sample dataset through the target surrogate model.

[0043] Step S33: Forming a probability distribution of the key safety parameters based on the key safety parameter values and calculating the sequence overrun probability of the safety parameters.

[0044] In one embodiment, step S4: Calculating the reactor shutdown risk based on the sequence overrun probability and using it as the safety margin evaluation result of the nuclear reactor includes: Step S41: Calculating the reactor shutdown risk SDF based on the following formula:

[0045] Wherein, is the probability of occurrence of abnormal event i; represents the kth sequence under abnormal event i, is the probability of occurrence of the dynamic evolution sequence ; is the probability that abnormal event i causes any one or more key safety parameters to exceed the safety threshold range, is the probability that the key safety parameter is higher than the safety threshold range.

[0046] In application, is also called the sequence overrun probability ( ), which is the joint probability that one or more key safety parameters exceed the safety threshold range.

[0047] In one embodiment, event i is a partial loss of feedwater event, is the probability that the feedwater loss reaches a certain specific level, . Wherein is the probability that the reactor power drops to a certain specific value, where is the probability that the main steam valve drops to a certain specific value, that is, this dynamic evolution sequence mainly includes valve opening adjustment and reactor power adjustment, and the sequence probability is the product of the probabilities of these two sub-events; It includes the following items: the signal for reactor shutdown when the water levels of steam generators 1 and 2 drop, the signal for reactor shutdown when the pressurizer water level changes, and the signal for reactor shutdown triggered by the pressurizer pressure. If any one of the above four items occurs, it can be considered that the limit is exceeded.

[0048] In the application, to ensure the estimated value has a sufficient confidence level, each simulation is regarded as a Bernoulli trial: Let the total sample size be n, and the number of failures (i.e., exceeding the limit) be k. Therefore, the point estimate of the probability of exceeding the limit is p = k / n.

[0049] Assume that the overall distribution of the probability of exceeding the limit approximately follows a normal distribution. For interval estimation of 𝑝 at a given confidence level (e.g., 95%), the Wilson confidence interval calculation based on the inversion of the Score test is adopted. The formula for its upper bound is: ; The formula for the lower bound is: ; where z is the value at the α / 2 quantile of the standard normal distribution (e.g., at a 95% confidence level, z ≈ 1.96).

[0050] Define the interval width as , the initial sample size for training the target surrogate model is 1000, and 1000 are added for each iterative training. To avoid a low single - interval width caused by accidental sampling fluctuations, this application sets that when the results of five consecutive iterative trainings of the surrogate model are all lower than 1%, it is considered that the sample size is sufficient. Figure 3 shows the convergence trend of the model output after increasing the sampling times. When the training sample size of the surrogate model exceeds 40000, the value of the probability of exceeding the limit tends to be stable, and the width of the confidence interval is low enough to meet the credibility requirements.

[0051] Step S42, regard the reactor shutdown risk as the evaluation result of the nuclear reactor safety margin.

[0052] In the application, this application focuses on the impact of the operations before the shutdown of the nuclear reactor in non - accident conditions. Therefore, the calculation of the nuclear reactor safety margin is transformed into the calculation of the shutdown risk SDF described above.

[0053] In the application, in combination with the safety margin calculation method, the limit values of key safety parameters are statistically analyzed. After evaluating their over-limit probabilities, a safety parameter distribution map is further drawn to visually display the changes in safety margins under different conditions. Finally, the calculation results are presented through visualization means, including the dynamic event sequence of key safety parameters, the risk limit surface, and the probability distribution curve of key safety parameters, so as to facilitate the operator to intuitively understand the system state and provide data support for optimization decisions.

[0054] In one embodiment, after training the initial surrogate model based on the training data set to obtain the target surrogate model in step S3, the method further includes: Step S401, based on one or more of the mean squared error, mean absolute error, mean absolute percentage error, coefficient of determination, and error over-limit rate of feature selection of key safety parameters, evaluate the applicability of the target surrogate model to the evaluation of the safety margin of the nuclear reactor.

[0055] In the application, after obtaining the target surrogate model, it is necessary to comprehensively evaluate the prediction performance of the target surrogate model using multiple indicators to ensure that it can meet the application requirements of the safety margin analysis of nuclear power plants. Considering the specific characteristics of parameter prediction in nuclear power plant safety analysis, the selection of evaluation indicators needs to be determined in combination with the specific prediction task type. When the target surrogate model is applied to the absolute value prediction of key safety parameters, the mean squared error (MSE, Mean Squared Error) and mean absolute error (MAE, Mean Absolute Error) are preferably used to evaluate the error magnitude, so as to intuitively reflect the average level of prediction error. If the MAE is less than 1% of the magnitude of the original data, it indicates that the target surrogate model has better performance. When comparing the prediction effects of parameters with different dimensions, the mean absolute percentage error (MAPE, Mean Absolute Percentage Error) is generally used to eliminate the influence of dimension differences. When there are a large number of values close to 0 in the data, the calculation result of MAPE may be too high. At this time, the coefficient of determination ( is used to reflect the fitting effect of the target surrogate model on the overall change trend of the parameter, the closer it is to 1, the stronger the explanatory ability of the model to the parameter change trend. Generally, it is required that > 0.9 to ensure the reliability of the model.

[0056] The calculation formulas of the above indicators are as follows:

[0057]

[0058]

[0059]

[0060] where n is the sample size, is the actual value, is the predicted value, is the average value of the actual values.

[0061] Step S402: If the applicability of the target surrogate model for evaluating the safety margin of the nuclear reactor is less than the preset applicability threshold, further iteratively optimize the target surrogate model until the applicability of the target surrogate model for evaluating the safety margin of the nuclear reactor is not less than the preset applicability threshold.

[0062] In application, the generalization ability of the target surrogate model is further tested in combination with independent test data. If the performance of the target surrogate model does not meet the predetermined generalization standard, the training samples can be increased, the model structure can be optimized, or the hyperparameters can be adjusted, and then the target surrogate model is further iteratively optimized until the prediction accuracy and generalization performance of the model meet the requirements of the actual application scenario. The above steps enable the target surrogate model to achieve the necessary prediction accuracy and reliability required to support the decision-making and safety margin evaluation of nuclear power plant risk guidance.

[0063] As Figure 4 shown, in one embodiment, step S1 of constructing a training dataset based on a sample dataset of uncertainty parameters and time series data of key safety parameters includes: Step S11: Sample the selected uncertainty parameters to generate an uncertainty sample set, where the uncertainty sample set includes multiple groups of sampling samples, and each group of sampling samples includes a value of each of the uncertainty parameters sampled at the same moment.

[0064] In application, an advanced sampling method is used to generate data reflecting different parameter changes. For example, Latin Hypercube Sampling (LHS) can cover the input space more evenly compared to simple random sampling or Monte Carlo sampling, thereby improving the training efficiency and enhancing the robustness of the resulting model. In addition, after sampling, all variables are normalized to a consistent scale.

[0065] Step S12: Based on a self-programmed simulation driver, batch-embed each group of sampling samples in the uncertainty sample set into the input card corresponding to the target simulation model.

[0066] In application, the target simulation model is an optimized nuclear reactor thermal-hydraulic simulation program.

[0067] Step S13: Batch-run the input card to obtain a large-scale simulation result.

[0068] In an application, a target simulation model is used to perform simulation calculations on a sample data set to obtain time series data of key safety parameters, with particular attention paid to the extreme values in the time series data, such as the peak fuel cladding temperature, the low value of the steam generator water level, and the high pressure of the pressurizer, etc.

[0069] Step S14, extract time series extreme value information from the large-scale simulation results, where the time series extreme value information includes time series extremes of at least one key safety parameter, and one piece of time series extreme value information corresponds to a group of the sampling samples.

[0070] Step S15, associatively store each group of sampling samples in the multiple groups of sampling samples with the corresponding time series extreme value information output by the simulation to construct a training data set.

[0071] In an application, a noise reduction technique is applied to improve the data quality of the training data set and the stability of model training. Therefore, by normalizing all data to a comparable scale, the training efficiency of the subsequent surrogate model is improved.

[0072] In one embodiment, before step S11 of sampling based on uncertainty parameters to obtain a sample data set of the uncertainty parameters, the method further includes: Step S01, select system variables whose influence on the behavior and safety margin of the nuclear reactor system is greater than a preset influence threshold as uncertainty parameters.

[0073] In an application, through sensitivity analysis, identify system variables that have a significant impact on the behavior and safety margin of the nuclear reactor system as uncertainty parameters.

[0074] Step S02, select intervention control variables as uncertainty parameters.

[0075] In an application, intervention control variables include operator actions and control variables, reflecting intervention options that can significantly change the accident process (for example, the timing of feedwater recovery, the adjustment of reactor power).

[0076] After selecting the uncertainty parameters, represent the uncertainty of the uncertainty parameters with a probability distribution according to engineering constraints, a reliability database, or a technical report. It can be understood that the above process of selecting uncertainty parameters ensures that the data set contains both system variables with an impact and actual operator actions.

[0077] In one embodiment, the method further includes: Step S51, select a target task scenario based on the occurrence frequency and potential consequences of the event sequence.

[0078] In applications, the target task scenarios are selected based on the occurrence frequency and potential consequences of event sequences, specifying the particular scenarios during the operation of nuclear power plants or nuclear facilities that require decision support, especially focusing on risk management under complex conditions, with emphasis on the following two types of scenarios: risk-important scenarios and operation-important scenarios.

[0079] Among them, risk-important scenarios refer to events or sequences of events with high risk contributions. According to the calculation results of importance indicators, they are ranked by risk importance, such as Fussel-Vesely importance, risk increase importance (RAW), and risk reduction importance (RRW); operation-important scenarios refer to scenarios that require higher requirements for the main control room operators' correct judgment and execution actions and have greater consequences. Timely and effective responses from the operators will not directly lead to unplanned reactor shutdowns. Therefore, comprehensive decision support with risk guidance is required to provide risk warnings and action time windows before operator intervention.

[0080] Step S52, establish an initial simulation model for the target task scenario.

[0081] In applications, the target task scenario is modeled through a thermal-hydraulic (T-H) model. For example, the target task scenario is modeled by using a T-H simulation program (such as RELAP5) to obtain an initial simulation model.

[0082] Step S53, simulate the target event sequence based on the initial simulation model to generate high-fidelity simulation data.

[0083] In applications, the simulation environment of the initial simulation model is carefully planned to accurately reflect the responses of the nuclear power plant (NPP) systems and equipment, which involves high-precision simulations of key system parameters such as reactor power level, coolant flow rate, system pressure, and temperature. In addition, for each selected operating condition, simulations are carried out by setting appropriate boundary conditions to balance the calculation accuracy and model granularity while avoiding excessive computational costs.

[0084] Step S54, verify the accuracy of the initial simulation model based on the high-fidelity simulation data; In applications, for the target task scenario, the T-H simulation program simulates specific event sequences, including the reduction of feedwater flow rate, the adjustment of the main steam pipe valve opening of the steam generator, and the change of reactor power level, to observe how operator intervention affects the water level and system stability. During the simulation, key safety parameters are recorded to capture the transient behavior of the system. Subsequently, the results of the T-H simulation are compared with experimental data, historical records, the calculated T-H simulation results in the nuclear power plant safety analysis report, or theoretical expectations to verify the accuracy of the model.

[0085] Step S55, if the accuracy of the initial simulation model is lower than the preset accuracy, return to step S53, which is the step of simulating the target event sequence based on the initial simulation model to generate high-fidelity simulation data.

[0086] In applications, if the accuracy of the initial simulation model is lower than the preset accuracy, return to step S53 to continuously optimize the initial simulation model.

[0087] Step S56, if the accuracy of the initial simulation model is not lower than the preset accuracy, use the initial simulation model as the target simulation model.

[0088] In applications, if the accuracy of the initial simulation model is not lower than the preset accuracy, use the initial simulation model as the target simulation model for subsequent use.

[0089] In applications, the key safety parameter distribution, the risks of each intervention path, and the limit surface / risk profile can be obtained based on the target surrogate model. The key safety parameter distribution is the probability distribution of the key safety parameters calculated by the target surrogate model, which is used to evaluate ; the risks of each intervention path are the probabilities of adverse results caused by different operation strategies quantified by the target surrogate model, enabling the operator to weigh the risk levels of each path according to the operation objectives; the target surrogate model can construct a limit surface or a risk profile to illustrate how certain parameter limits (e.g., maximum pressurizer pressure, minimum water level) interact with each intervention scenario. It highlights the boundary between the safe area and the unsafe area, thus enabling more direct and risk-based decisions.

[0090] The embodiment of the present application also selects the rupture of the feedwater branch pipe in the secondary loop system of a nuclear power plant (i.e., the partial loss of feedwater accident) as an example of the target task scenario, and conducts a safety margin assessment through the above method provided by the present application. The specific process of the example is as follows.

[0091] According to the operating procedures of the nuclear power plant, the degree of partial loss of water is not sufficient to directly cause the reactor to trip, but it may affect its continuous power operation and pose potential risks. Therefore, this article simulates the dynamic evolution of the accident through a thermal-hydraulic (T-H) transient analysis and evaluates different operator intervention strategies to optimize the operation response, avoid unplanned reactor shutdowns, and improve operation safety and flexibility. The research uses the RELAP5 thermal-hydraulic simulation program for nuclear reactor systems to construct a thermal-hydraulic simulation model of a typical 900MW dual-loop pressurized water reactor, including key equipment such as the reactor pressure vessel, core, hot and cold leg pipes, main pumps, steam generators, and pressurizers. The main design parameters are shown in Table 1.

[0092] Table 1 Design values of the main parameters of the nuclear power plant

[0093] Case Scenario Description In the scenario of partial loss of feedwater in the secondary loop involved in this example, a leak occurs in the feedwater branch pipe of one steam generator, resulting in a decrease in feedwater flow rate, and then causing the water level of this steam generator to continuously drop. In response to this situation, the operator can take measures such as reducing the core power and adjusting the opening of the main steam pipeline valve to maintain the water level of the steam generator and the system stability. During the simulation process, the RELAP5 program is used to analyze the evolution trends of key safety parameters such as the water level of the steam generator, the pressure of the pressurizer, and the reactor power with the change of feedwater flow rate, and further explore the effects of different intervention strategies.

[0094] The main events and time nodes are as follows: At t = 1000 s, a leak occurs in the feedwater branch pipe of Steam Generator 1, resulting in partial loss of water in the secondary loop system, a drop in the water level of the steam generator, and affecting steam production and heat exchange capacity.

[0095] At t = 1060 s: The operator starts to take countermeasures, including reducing the reactor power and adjusting the opening of the main steam pipeline valve of the steam generator, to mitigate the impact of the accident. When the pressure of the pressurizer exceeds the set limit value, the safety valve and relief valve are normally started to effectively reduce the system pressure and prevent overpressure failure.

[0096] For the sake of simplifying the calculation, the following assumptions are made: 1. The average value of the temperature feedback coefficients of the reactor fuel elements and moderators is adopted.

[0097] 2. The condenser maintains a constant vacuum to ensure the normal operation of the steam turbine.

[0098] 3. The reactor operates at partial power and the automatic reactor shutdown is not triggered.

[0099] 4. The main steam control valve, steam discharge valve, and pressurizer safety valve operate normally.

[0100] The reactor may be automatically shut down if the key safety parameters exceed the safety thresholds. Referring to the accident analysis report of feedwater loss, there are 3 types of parameter signals that may lead to shutdown: 1. The low water level shutdown signal (<1.24 m) of the steam generator water level (SG1lev, SG2lev).

[0101] 2. The overpressure shutdown signal (>16.3 Mpa) of the pressurizer pressure (PRZpre).

[0102] 3. The low water level shutdown signal (<0.5 m) of the pressurizer water level (PRZlev).

[0103] Target Agent Model Training and Verification After the thermal-hydraulic model is constructed, data required to generate the training target surrogate model is generated. This study focuses on analyzing the intelligent intervention strategies under feedwater flow rate changes and different accident sequences, and mainly considers the impacts of the following uncertainty parameters on accident consequences: 1. The feedwater loss flow rate level in the secondary loop.

[0104] 2. The value of the operator manually adjusting the reactor power.

[0105] 3. The opening degree of the steam valve manually controlled by the operator.

[0106] 4. The recovery time of the feedwater flow rate.

[0107] To construct the dataset required for the training target surrogate model, this study first models a partial loss-of-feedwater accident (LOFW) in the secondary loop system of a nuclear power plant and defines key operating events and corresponding operator intervention measures. The operator's intervention strategies mainly include adjusting the reactor power level and the opening degree of the main steam outlet valve. To fully cover possible intervention paths, this study discretizes these operating variables, generating a total of 32 different intervention paths. In addition, different feedwater flow rate levels and their recovery times also affect the accident evolution. Therefore, under each intervention path, the Latin hypercube sampling (LHS) method is used to randomly sample these two parameters to expand the input variable space and enhance the generalization ability of the model. Finally, 50 samples are extracted for each intervention path, and 50 samples are separately extracted for the accident path without any intervention measures, generating a total of 1650 data samples. Table 2 details the sampling methods for operating events and intervention measures.

[0108] Table 2 Operating Events and Intervention Measures

[0109] After data sampling is completed, the RELAP5 thermal-hydraulic simulation program is used to calculate all samples to obtain time-series data of key safety parameters (such as steam generator water level, pressurizer pressure, etc.). In order to train the target surrogate model, it is necessary to extract the maximum and minimum values of these key safety parameters to determine whether an over-limit situation occurs and further use them for safety margin calculation. During the training process of the target surrogate model, feedwater flow rate, feedwater recovery time, and operator intervention measures are selected as input parameters, and the key safety parameters, namely the water level of Steam Generator 1 (SG1 Lev), the water level of Steam Generator 2 (SG2 Lev), the pressure of the pressurizer (PRZ Pre), and the water level of the pressurizer (PRZ Lev), are used as output parameters to ensure that the model can accurately predict the safety margin level under different working conditions. Table 3 lists the input and output parameters of the target surrogate model.

[0110] Table 3 Input and Output Parameters of the Target Surrogate Model

[0111] Subsequently, the collected data is divided into a training set (70%), a test set (15%), and a validation set (15%) proportionally to ensure that the model has good learning ability, generalization ability, and stable performance evaluation. During the training process, the mean squared error (MSE) is used as the loss function, and the Adam optimizer is used to update the model weights. At the same time, the Dropout mechanism is introduced to reduce the risk of overfitting. Figure 5 The curve of the change of the loss function for model training shows the change trend of the model loss function. During the training process, the loss value gradually converges after 60 epochs and drops below 1%, indicating that the model has achieved a relatively good fitting effect. Since the input scenarios and parameter dimensions involved in this study are relatively small, the model training efficiency is high, and the training can be completed within 150 epochs, with relatively low overall computational resource consumption (about 1 minute), meeting the requirements of high-efficiency computing.

[0112] After training is completed, the performance of the target surrogate model based on Transformer is verified. MSE, MAE, MAPE, and are selected as evaluation indicators. Considering that the parameters involved in the case are 4-dimensional and the parameter interval span is large, directly calculating MSE and MAE will result in insignificant dimensional errors. Therefore, the calculation data in this case is normalized. In addition, since there are a large number of values of 0 in the steam generator water level data, this will cause the result to be too large when calculating MAPE and cannot actually reflect the accuracy of the model. Finally, the coefficient of determination is set in this case. > 0.95 is used as a criterion to ensure that the target proxy model has high prediction accuracy and generalization ability, and can meet the requirements of actual engineering applications. Figures 6 to 8 They are the regression performances of the Transformer model, Support Vector Machine (SVM), and Multi-Layer Perceptron (MLP), respectively. As can be seen from the figure: The Transformer model performs better than traditional regression models (SVM and MLP) in regression tasks. This is because the Transformer can effectively capture global patterns in complex data through the self-attention mechanism, making the predicted values closer to the actual values. Table 4 further verifies the advantage of the Transformer in prediction accuracy. The comparison results show that compared with MLP and SVM, the MSE decreases by about 89.5% and 90.5%, and the MAE decreases by about 71.5% and 69.2% after combining the Transformer algorithm. It is increased to 0.982, and the accuracy and stability of the regression calculation are greatly improved.

[0113] Table 4 Comparison of performance indicators of different regression models

[0114] Samples are taken for the water supply recovery time of different intervention paths, and the corresponding key output parameter values are calculated by the regression model (target proxy model), and the probability of exceeding the threshold is statistically analyzed. Based on the comprehensive analysis of these paths, several typical key paths are extracted at two water supply flow levels in this paper, and a reactor trip event tree is further constructed to comprehensively evaluate the safety margin of the system under various scenarios. Figures 9 to 12 It further reveals the kernel density estimation (KDE) distribution function of the key safety parameters during reactor trip under different intervention measures. Figure 9 It is the distribution of key safety parameters when the power drops to 60% and the valve opening drops to 40% at 50% water supply flow rate; Figure 10 It is the distribution of key safety parameters when the power drops to 70% and the valve opening drops to 40% at 50% water supply flow rate; Figure 11 It is the distribution of key safety parameters when the power drops to 90% and the valve opening drops to 40% at 70% water supply flow rate; Figure 12 It is the distribution of key safety parameters when the power drops to 80% and the valve opening drops to 40% at 70% water supply flow rate. Obviously, when the water supply flow rate is 70%, reasonably adjusting the reactor power and precisely controlling the valve opening can effectively avoid unplanned reactor trips, while when the water supply flow rate is 50%, even with corresponding intervention measures, the reactor trip signal will still be triggered in some scenarios. Figure 13 andFigure 14 The reactor trip event tree shows the sequence exceedance probabilities under different operating conditions, Figure 13 which is the reactor trip event tree at 70% feedwater flow rate; Figure 14 which is the reactor trip event tree at 50% feedwater flow rate, where success indicates successful avoidance of reactor trip and failure indicates emergency reactor trip. Combining the probabilistic safety analysis data of a specific power plant (the probability of feedwater loss event, the probability of the header event), the reactor trip risk is further obtained according to the reactor trip risk calculation formula. In the example, event i is the partial feedwater loss event, which is the probability of feedwater loss to a specific level; which represents the kth sequence under the partial feedwater loss event, which is the dynamic evolution sequence probability. In a specific case . Among them which is the probability of the reactor power dropping to a specific value, among which which is the probability of the main steam valve dropping to a specific value, that is, this dynamic evolution sequence mainly includes valve opening adjustment and reactor power adjustment, and the sequence probability is the product of the probabilities of these two sub-events; which is the probability that the abnormal event i causes any one or more key safety parameters to exceed the safety threshold range, which is the probability that the key safety parameter is higher than the safety threshold range. In the example includes the following items: the water levels of steam generators 1 and 2 drop to the reactor trip signal, the pressurizer water level changes to the reactor trip signal, and the pressurizer pressure triggers the reactor trip signal. If any one of the above four items occurs, it can be considered as exceeding the limit. The applicability of the model and method proposed in this application in multi-scenario evaluation is verified.

[0115] It should be understood that the magnitudes of the sequence numbers of the steps in the above embodiments do not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation to the implementation process of the embodiments of this application.

[0116] The embodiments of this application also provide an efficient nuclear reactor safety margin evaluation system for performing the steps in the embodiments of the above-mentioned efficient nuclear reactor safety margin evaluation method. The efficient nuclear reactor safety margin evaluation system can be a virtual device (Virtual Appliance) in a computer device, run by the processor of the computer device, or the computer device itself.

[0117] As Figure 15 shown, the efficient nuclear reactor safety margin evaluation system 50 provided by the embodiments of this application includes: A training dataset module 501, configured to construct a training dataset based on a sample dataset of uncertainty parameters and time series data of key safety parameters; An agent model training module 502, configured to train an initial agent model based on the training dataset to obtain a target agent model; A sequence overrun probability module 503, configured to efficiently calculate the sequence overrun probability of the key safety parameters by using the target agent model; A safety margin calculation module 504, configured to calculate the reactor trip risk of a nuclear reactor based on the sequence overrun probability and use it as the evaluation result of the safety margin of the nuclear reactor.

[0118] In one embodiment, the sequence overrun probability module 503 is configured to: Sample the uncertainty parameters to obtain a sample set, where the sample set includes multiple samples; Calculate the key safety parameter values corresponding to each sample in the sample dataset through the target agent model; Form a probability distribution of the key safety parameters based on the key safety parameter values and calculate the sequence overrun probability of the safety parameters.

[0119] In one embodiment, the safety margin calculation module 504 is configured to: Calculate the reactor trip risk SDF based on the following formula:

[0120] Where, is the probability of occurrence of abnormal event i is the dynamic evolution sequence is the probability of occurrence, and it is the probability that abnormal event i causes any one or more key safety parameters to exceed the safety threshold range, is the key safety parameter is the probability of being higher than the safety threshold range.

[0121] Use the reactor trip risk as the evaluation result of the nuclear reactor safety margin.

[0122] In one embodiment, the high-efficiency nuclear reactor safety margin evaluation system 50 further includes: An agent model verification module, configured to: Evaluate the applicability of the target agent model for nuclear reactor safety margin evaluation based on one or more of the mean square error, mean absolute error, mean absolute percentage error, coefficient of determination, and error overrun rate of feature selection of key safety parameters; If the applicability of the target surrogate model for evaluating the safety margin of the nuclear reactor is less than the preset applicability threshold, the target surrogate model is further iteratively optimized until the applicability of the target surrogate model for evaluating the safety margin of the nuclear reactor is not less than the preset applicability threshold.

[0123] In one embodiment, the training dataset module 501 is configured to: Sample the selected uncertainty parameters to generate an uncertainty sample set, where the uncertainty sample set includes multiple groups of sampling samples, and each group of sampling samples includes a value of each of the uncertainty parameters sampled at the same moment; Based on a self-programmed simulation driver, batch embed each group of sampling samples in the uncertainty sample set into the input card corresponding to the target simulation model; Batch run the input card to obtain a large-scale simulation result; Extract time series extreme value information from the large-scale simulation result, where the time series extreme value information includes time series extreme values of at least one key safety parameter, and one piece of time series extreme value information corresponds to one group of sampling samples; Associate and store each group of sampling samples in the multiple groups of sampling samples with the corresponding time series extreme value information of the simulation output to construct a training dataset.

[0124] In one embodiment, the high-efficiency nuclear reactor safety margin evaluation system 50 further includes: An uncertainty parameter screening module, configured to: Select system variables whose influence on the behavior and safety margin of the nuclear reactor system is greater than the preset influence threshold as uncertainty parameters; Select intervention control variables as uncertainty parameters.

[0125] In one embodiment, the high-efficiency nuclear reactor safety margin evaluation system 50 further includes: A target simulation model module, configured to: Select a target task scenario based on the occurrence frequency and potential consequences of the event sequence; Establish an initial simulation model for the target task scenario; Simulate the target event sequence based on the initial simulation model to generate high-fidelity simulation data; Verify the accuracy of the initial simulation model based on the high-fidelity simulation data; If the accuracy of the initial simulation model is lower than the preset accuracy, return to the step of simulating the target event sequence based on the initial simulation model to generate high-fidelity simulation data; If the accuracy of the initial simulation model is not lower than the preset accuracy, the initial simulation model is used as the target simulation model.

[0126] In applications, each module in the high-efficiency nuclear reactor safety margin evaluation system can be a software program module, can also be implemented by different logic circuits integrated in a processor, or can also be implemented by multiple distributed processors.

[0127] Figure 16 Schematic diagram of the structure of a computer device provided by an embodiment of the present application. As Figure 16 shown, the computer device 6 of this embodiment includes: at least one processor 60 ( Figure 16 only one is shown in the figure), a processor, a memory 61, and a computer program 62 stored in the memory 61 and operable on the at least one processor 60. When the processor 60 executes the computer program 62, the steps in any of the above-mentioned embodiments of the high-efficiency nuclear reactor safety margin evaluation method are implemented.

[0128] The computer device may include, but is not limited to, a processor 60 and a memory 61. Those skilled in the art can understand that Figure 16 merely examples of the computer device 6, and do not constitute a limitation on the computer device 6. It may include more or fewer components than shown in the figure, or combine certain components, or different components. For example, it may also include input / output devices, network access devices, etc.

[0129] The processor 60 may be a central processing unit (CPU), and the processor 60 may also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc.

[0130] The memory 61 may be an internal storage unit of the computer device 6 in some embodiments, such as a hard disk or memory of the computer device 6. The memory 61 may also be an external storage device of the computer device 6 in other embodiments, such as a plug-in hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the computer device 6. Further, the memory 61 may also include both the internal storage unit and the external storage device of the computer device 6. The memory 61 is used to store an operating system, application programs, a BootLoader, data, and other programs, such as the program code of the computer program. The memory 61 may also be used to temporarily store data that has been output or will be output.

[0131] It should be noted that for the content such as information interaction and execution process between the above-mentioned device / units, since it is based on the same concept as the method embodiment of the present application, for its specific functions and the technical effects brought, reference can be specifically made to the method embodiment part, and details will not be repeated here.

[0132] Those skilled in the art can clearly understand that for the convenience and brevity of description, only the above division of each functional unit and module is used as an example. In actual applications, the above functions can be allocated to different functional units and modules according to needs, that is, the internal structure of the device is divided into different functional units or modules to complete all or part of the functions described above. Each functional unit and module in the embodiment can be integrated into a processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above integrated unit can be implemented in the form of hardware or in the form of a software functional unit. In addition, the specific names of each functional unit and module are only for the convenience of mutual distinction and do not limit the protection scope of the present application. The specific working process of the units and modules in the above system can refer to the corresponding process in the foregoing method embodiment, and details will not be repeated here.

[0133] The embodiment of the present application also provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments can be implemented.

[0134] As Figure 17 shown, the embodiment of the present application provides a computer program product 7, including a computer program 62. When the computer program 62 is run, the steps in the above-mentioned method embodiments for evaluating the safety margin of each high-efficiency nuclear reactor are executed.

[0135] When the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, to implement all or part of the processes in the above-described embodiment methods of this application, a computer program can be used to instruct the relevant hardware to complete. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-described various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file, or some intermediate form, etc. The computer-readable medium can at least include: any entity or device that can carry the computer program code to the device / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium. For example, a USB flash drive, a mobile hard disk, a magnetic disk, or an optical disc, etc. In some jurisdictions, according to legislation and patent practice, the computer-readable medium cannot be an electrical carrier signal and a telecommunication signal.

[0136] In the above embodiments, the descriptions of the various embodiments have their own emphases. For the parts not detailed or recorded in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0137] In the embodiments provided in this application, it should be understood that the disclosed computer devices and methods can be implemented in other ways. For example, the computer device embodiments described above are only illustrative. For example, the division of the modules or units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed couplings or direct couplings or communication connections to each other can be through some interfaces, and the indirect couplings or communication connections of the devices or units can be in an electrical, mechanical, or other form.

[0138] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should all be included within the protection scope of the present application.

Claims

1. An efficient method for evaluating the safety margin of a nuclear reactor, characterized in that, Including: Constructing a training dataset based on a sample dataset of uncertainty parameters and time series data of key safety parameters; Training an initial surrogate model based on the training dataset to obtain a target surrogate model; Efficiently calculating the sequence overrun probability of the key safety parameters using the target surrogate model; Calculating the reactor trip risk of the nuclear reactor based on the sequence overrun probability and using it as the evaluation result of the safety margin of the nuclear reactor.

2. The high-efficiency nuclear reactor safety margin evaluation method according to claim 1, wherein The efficiently calculating the sequence overrun probability of the key safety parameters using the target surrogate model includes: Sampling the uncertainty parameters to obtain a sample set, where the sample set includes multiple samples; Calculating the value of the key safety parameter corresponding to each sample in the sample dataset through the target surrogate model; Forming a probability distribution of the key safety parameters based on the key safety parameter values and calculating the sequence overrun probability of the safety parameters.

3. The high-efficiency nuclear reactor safety margin evaluation method according to claim 2, characterized in that Calculating the reactor trip risk of the nuclear reactor based on the sequence overrun probability and using it as the evaluation result of the safety margin of the nuclear reactor includes: Calculating the reactor trip risk SDF based on the following formula: Among them, is the probability of the occurrence of abnormal event i, represents the k-th sequence under abnormal event i, is the dynamic evolution sequence the probability of occurrence, is the probability that abnormal event i causes any one or more key safety parameters to exceed the safety threshold range, is the key safety parameter the probability of being higher than the safety threshold range; Using the reactor trip risk as the evaluation result of the nuclear reactor safety margin.

4. The high-efficiency nuclear reactor safety margin evaluation method according to any one of claims 1 to 3, characterized in that After training the initial surrogate model based on the training dataset to obtain the target surrogate model, it further includes: Evaluating the applicability of the target surrogate model for nuclear reactor safety margin assessment based on one or more of the mean squared error, mean absolute error, mean absolute percentage error, coefficient of determination, and error overrun rate of feature selection of key safety parameters; If the applicability of the target surrogate model for nuclear reactor safety margin assessment is less than a preset applicability threshold, further iteratively optimizing the target surrogate model until the applicability of the target surrogate model for nuclear reactor safety margin assessment is not less than the preset applicability threshold.

5. The efficient nuclear reactor safety margin evaluation method according to any one of claims 1 to 3, characterized in that, The constructing a training dataset based on a sample dataset of uncertainty parameters and time series data of key safety parameters includes: Sampling the selected uncertainty parameters to generate an uncertainty sample set, where the uncertainty sample set includes multiple groups of sampling samples, and each group of sampling samples includes a value of each of the uncertainty parameters sampled at the same moment; Based on a self-programmed simulation driver, batch-embedding each group of sampling samples in the uncertainty sample set into the input card corresponding to the target simulation model; Batch-running the input card to obtain a large-scale simulation result; Extracting time series extreme value information from the large-scale simulation result, where the time series extreme value information includes at least the time series extreme values of one key safety parameter, and one piece of time series extreme value information corresponds to one group of sampling samples; Associatively storing each group of sampling samples in the multiple groups of sampling samples with the corresponding time series extreme value information of the simulation output to construct a training dataset.

6. The high-efficiency nuclear reactor safety margin evaluation method according to claim 5, characterized in that, Before sampling based on the uncertainty parameters to obtain the sample dataset of the uncertainty parameters, it further includes: Selecting system variables whose influence on the behavior and safety margin of the nuclear reactor system is greater than a preset influence threshold as uncertainty parameters; Selecting intervention control variables as uncertainty parameters.

7. The high-efficiency nuclear reactor safety margin evaluation method according to claim 5, wherein The method further includes: Select a target task scenario based on the occurrence frequency and potential consequences of the event sequence; Establish an initial simulation model for the target task scenario; Simulate the target event sequence based on the initial simulation model to generate high-fidelity simulation data; Verify the accuracy of the initial simulation model based on the high-fidelity simulation data; If the accuracy of the initial simulation model is lower than the preset accuracy, return to the step of simulating the target event sequence based on the initial simulation model to generate high-fidelity simulation data; If the accuracy of the initial simulation model is not lower than the preset accuracy, use the initial simulation model as the target simulation model.

8. An efficient nuclear reactor safety margin assessment system, characterized in that, Including: A training dataset module for constructing a training dataset based on a sample dataset of uncertainty parameters and time series data of key safety parameters; An agent model training module for training an initial agent model based on the training dataset to obtain a target agent model; A sequence overrun probability module for efficiently calculating the sequence overrun probability of the key safety parameters using the target agent model; A safety margin calculation module for calculating the reactor trip risk of the nuclear reactor based on the sequence overrun probability and using it as the evaluation result of the safety margin of the nuclear reactor.

9. A computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the efficient nuclear reactor safety margin evaluation method according to any one of claims 1 to 7.

10. A computer program product, characterized in that, Including a computer program, when the computer program is run, the efficient nuclear reactor safety margin evaluation method according to any one of claims 1 to 7 is executed.

Images