Nuclear reactor analysis coupling calculation acceleration method and system

Abstract

The embodiment of the invention provides a nuclear reactor analysis coupling calculation acceleration method and system, and relates to the technical field of nuclear reactor system simulation analysis. According to the method, an input card and a grid model of a system analysis program and a containment analysis program of a nuclear reactor can be firstly established, then an alternative model is established, and the alternative model is a neural network model obtained by training sample transfer parameters. When coupling calculation is executed, real-time transmission parameters transmitted to a containment analysis program by a system analysis program can be obtained, the real-time transmission parameters are input into a substitution model to obtain a substitution result output by the substitution model, and then coupling calculation result information is generated according to the substitution result. According to the method, a fast calculation neural network model can be used for replacing a slow calculation containment analysis program, the influence of the calculation efficiency of the containment analysis program on the total coupling calculation efficiency is reduced, and the problem that the coupling calculation efficiency is low in the nuclear reactor simulation analysis process is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of nuclear reactor system simulation analysis, and particularly relates to a nuclear reactor analysis coupling calculation acceleration method and system. BACKGROUND

[0002] The simulation analysis of a nuclear reactor system refers to a system analysis method for simulating and analyzing possible reactor abnormalities in a specific period by collecting historical monitoring data and real-time monitoring data of the nuclear reactor, so as to take preventive measures in advance according to the analysis results. In the simulation analysis process of the nuclear reactor system, a calculation program needs to be used to perform calculation on related numerical values of the nuclear reactor system. The calculation program can be divided into a fast analysis program and a high-precision analysis program according to the simulation accuracy and speed.

[0003] Among them, the fast analysis program is represented by a reactor system analysis program, which uses a relatively simple grid division method such as one-dimensional grid to simulate the overall state of the nuclear reactor. It can be seen that the fast analysis program has fast calculation speed, but the calculation accuracy is lacking. The high-precision analysis program is represented by a containment analysis program, which can use a fine three-dimensional grid to simulate the state of a main component of the reactor, so the high-precision analysis program has slower calculation speed but higher calculation accuracy.

[0004] In order to improve the simulation accuracy of the overall reactor system, the fast analysis program and the high-precision analysis program of the reactor can be coupled for calculation. However, since the overall calculation speed of the coupled calculation depends on the slowest analysis program, when the high-precision analysis program participates in the coupled calculation, the coupled calculation efficiency will be reduced. SUMMARY

[0005] Therefore, the embodiments of the present application provide a nuclear reactor analysis coupling calculation acceleration method and system to solve the problem of low coupled calculation efficiency in the simulation analysis process of the nuclear reactor.

[0006] According to a first aspect of the present application, a nuclear reactor analysis coupling calculation acceleration method is provided, which comprises: establishing coupling calculation model data, the coupling calculation model data comprising input cards and grid models of a system analysis program and a containment analysis program of a nuclear reactor; constructing a substitute model according to the coupling calculation model data, the substitute model being a neural network model trained using sample transmission parameters; the input parameters of the substitute model comprising parameters transmitted by the system analysis program to the containment analysis program in the coupling calculation process; and the output parameters of the substitute model comprising parameters transmitted by the containment analysis program to the system analysis program in the coupling calculation process; Obtain real-time transmission parameters during the coupling calculation process, the real-time transmission parameters including the parameters transmitted by the system analysis program to the containment analysis program within the current coupling calculation cycle; The real-time transmission parameters are input into the substitution model to obtain the substitution result output by the substitution model; Based on the substitution results, couple calculation result information is generated.

[0007] In some embodiments, the method further includes: The containment analysis program is periodically invoked to perform coupled calculations based on the real-time transmitted parameters in order to obtain verification results. The accuracy error is calculated by comparing the verification results and the alternative results. If the accuracy error is greater than a preset error threshold, the alternative model is iteratively trained based on the accuracy error; If the accuracy error is less than or equal to a preset error threshold, the step of generating coupled calculation result information based on the substitution result is executed.

[0008] In some embodiments, constructing an alternative model based on the coupled computational model data includes: Acquire analytical target information of the nuclear reactor, including the physical problems to be studied during the nuclear reactor analysis process; The neural network model structure of the alternative model is set according to the analysis target information, and the neural network model structure of the alternative model is adapted to the physical problem to be studied in the nuclear reactor analysis process; The sample transfer parameters are obtained, including sample parameters and sample labels. The sample parameters include parameters passed from the system analysis program to the containment analysis program during the coupling calculation process. The sample labels are set according to the parameters passed from the containment analysis program to the system analysis program during the coupling calculation process. The alternative model is trained iteratively using the sample transfer parameters.

[0009] In some embodiments, obtaining the sample transfer parameters includes: Monitor the first parameter passed from the system analysis program to the containment analysis program during the coupling calculation process; The sample parameters are generated based on the first parameter; Record the second parameter that the containment analysis program passes to the system analysis program in response to the first parameter; Based on the second parameter, set sample labels for the sample parameters to generate associated data pairs; The sample transfer parameters are generated by combining multiple of the associated data pairs.

[0010] In some embodiments, iteratively training the alternative model using the sample transfer parameters includes: The sample parameters in the sample transfer parameters are input into the alternative model to obtain the training results output by the alternative model; The training loss is calculated by comparing the training results with the sample labels. If the training loss is greater than a preset loss threshold, the model parameters of the alternative model are modified according to the training loss. If the training loss is less than or equal to a preset loss threshold, output the model parameters of the alternative model.

[0011] In some embodiments, establishing coupled computational model data includes: Acquire modeling data for the nuclear reactor, including reactor type information, loss-of-coolant accident operating condition information, and structural parameter information; the structural parameter information includes system structural parameters and containment structure parameters. Set simulation conditions, including initial conditions and boundary conditions; Set the input card according to the modeling data and the simulation conditions; A nuclear reactor geometric model is constructed based on the structural parameter information. The nuclear reactor geometric model includes a system model and a containment model. The nuclear reactor geometric model is meshed to obtain the mesh model; the mesh model includes multiple mesh nodes, and the mesh nodes are assigned node numbers. Determine the interaction information between the system analysis program and the containment analysis program, the interaction information including the coupling surface and the data interaction interface; Based on the interaction information, iterative solution parameters are set for the mesh model.

[0012] In some embodiments, obtaining real-time transmission parameters during the coupled computation process includes: The coupling program is invoked, which controls the coupled calculation process of the system analysis program and the containment analysis program; The coupling program reads the original calculation parameters of the system analysis program and the containment analysis program. Perform dimensional and unit checks on the original calculation parameters to generate the real-time transmission parameters; Store the real-time transmission parameters.

[0013] In some embodiments, the real-time transmission parameters are input into the substitution model to obtain the substitution result output by the substitution model, including: The coupling program records the parameter calculation time of the containment analysis program. The parameter calculation time is the time difference between the moment when the first raw calculation parameter output by the system analysis program is sent to the containment analysis program and the moment when the containment analysis program outputs the second raw calculation parameter for the first raw calculation parameter. Obtain a preset duration threshold, which is set based on the calculation duration of the alternative model statistically analyzed within multiple historical coupling calculation cycles; If the parameter calculation time exceeds the preset time threshold, the real-time transmission parameter is input into the alternative model through the coupling procedure; If the calculation time of the parameter is less than or equal to the preset time threshold, the real-time transmission parameter is input into the containment analysis program through the coupling program.

[0014] In some embodiments, the real-time transmission parameters are input into the substitution model to obtain the substitution result output by the substitution model, including: Obtain the current computation stage, which is either the training stage or the acceleration stage; The parameter transfer target is determined based on the current computation stage; wherein, if the current computation stage is the training stage, the parameter transfer target is the containment analysis program and the alternative model; if the current computation stage is the acceleration stage, the parameter transfer target is the alternative model. The real-time transmission parameters are sent to the parameter transmission target.

[0015] According to a second aspect of this application, a nuclear reactor analysis coupled computation acceleration system is provided, the system comprising: The coupling condition establishment module is used to establish coupling calculation model data, which includes the input cards and mesh models of the nuclear reactor system analysis program and containment analysis program; An alternative model construction module is used to construct an alternative model based on the coupled computation model data. The alternative model is a neural network model trained using sample transfer parameters. The input parameters of the alternative model include the parameters passed to the containment analysis program by the system analysis program during the coupled computation process. The output parameters of the alternative model include the parameters passed to the system analysis program by the containment analysis program during the coupled computation process. The parameter acquisition module is used to acquire real-time transmission parameters during the coupling calculation process. The real-time transmission parameters include the parameters transmitted by the system analysis program to the containment analysis program within the current coupling calculation cycle. The substitution calculation module is used to input the real-time transmission parameters into the substitution model to obtain the substitution result output by the substitution model; The result output module is used to generate coupled calculation result information based on the substitution result.

[0016] According to a third aspect of this application, a computer device is provided, including a storage medium, a processor, and a computer program stored on the storage medium and executable on the processor, wherein the processor executes the program to implement the above-described nuclear reactor analysis coupled calculation acceleration method.

[0017] According to a fourth aspect of this application, a storage medium is provided that stores a computer program thereon, which, when executed by a processor, implements the above-described nuclear reactor analysis coupled calculation acceleration method.

[0018] By employing the above technical solutions, embodiments of this application provide a method and system for accelerating coupled computation in nuclear reactor analysis. The method first establishes input cards and mesh models for the system analysis program and containment analysis program of the nuclear reactor, and then constructs a substitute model, wherein the substitute model is a neural network model trained using sample transfer parameters. During coupled computation, real-time transfer parameters passed from the system analysis program to the containment analysis program can be obtained and input into the substitute model to obtain the substitute results output by the substitute model. Then, coupled computation result information is generated based on the substitute results. This method can utilize a fast-computing neural network model to replace the slow-computing containment analysis program, reducing the impact of the containment analysis program's computational efficiency on the overall efficiency of coupled computation, and solving the problem of low coupled computation efficiency in nuclear reactor simulation analysis.

[0019] The above description is only an overview of the technical solution of this application. In order to better understand the technical means of this application and to implement it in accordance with the contents of the specification, and to make the above and other objects, features and advantages of this application more obvious and understandable, the following are specific embodiments of this application. Attached Figure Description

[0020] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments of this application and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a schematic diagram of the nuclear reactor analysis coupled calculation acceleration method provided in the embodiments of this application; Figure 2 This is a schematic diagram of the iterative training process of the alternative model provided in the embodiments of this application; Figure 3 This is a schematic diagram of the coupled calculation process provided in the embodiments of this application; Figure 4 This is a schematic diagram of the real-time parameter transmission process provided in the embodiments of this application; Figure 5 This is a schematic diagram of the coupling calculation process at different stages provided in the embodiments of this application; Figure 6 This is a schematic diagram of the structure of a nuclear reactor analysis coupled computation acceleration system provided in an embodiment of this application. Detailed Implementation

[0021] The present application will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the embodiments of the present application can be combined with each other.

[0022] In the embodiments of this application, a nuclear reactor refers to a device that starts and controls a self-sustaining nuclear chain reaction, which may include application reactors for power generation, propulsion, medical and other fields, as well as research reactors for scientific research.

[0023] In some embodiments, the nuclear reactor may include a core, a cooling system, a pressure vessel, a circulation system, and a containment vessel. Several key components together constitute the nuclear reactor system. The reactor core may include nuclear fuel, a moderator, and control rods to generate energy for a nuclear-controlled nuclear reaction. The cooling system works in conjunction with the circulation system to remove the enormous heat generated by the core, preventing a core meltdown, and to transfer heat to energy-consuming equipment such as generators. The pressure vessel can contain the core, control rods, and coolant, and operates under high temperature and pressure.

[0024] The containment vessel is a reinforced concrete dome-shaped structure that houses the entire nuclear reactor. It serves as a safety barrier, containing radioactive materials and preventing their release into the environment in the event of an accident. For example, in a pressurized water reactor, the containment vessel can be made of prestressed concrete and lined with thick steel plates to ensure extremely high strength and airtightness. The containment vessel must be designed to withstand immense internal pressure, impacts from projectiles, earthquakes, and other extreme external events.

[0025] Therefore, simulation analysis of the system or key components is necessary during the design of a nuclear reactor system. Nuclear reactor system simulation analysis refers to a system analysis method that uses historical and real-time monitoring data of the nuclear reactor to simulate and analyze potential reactor anomalies within a specific period, in order to take preventative measures based on the analysis results.

[0026] In the simulation and analysis of nuclear reactor systems, computational programs are required to perform calculations on relevant numerical values ​​of the nuclear reactor system. These programs can be categorized into fast analysis programs and high-precision analysis programs based on their simulation accuracy and speed. Fast analysis programs, represented by reactor system analysis programs, use relatively simple mesh generation methods, such as one-dimensional meshes, to simulate the overall state of the nuclear reactor. It is evident that fast analysis programs are fast in computation but lack computational accuracy. High-precision analysis programs, represented by containment analysis programs, can use fine three-dimensional meshes to simulate the state of a major component of the reactor. Therefore, high-precision analysis programs are slower in computation but offer higher computational accuracy.

[0027] To improve the overall simulation accuracy of a reactor system, fast and high-precision analysis programs can be coupled. Coupled calculations refer to the calculations involving the interactions between two or more systems. For example, in the simulation analysis of a nuclear reactor system, coupled calculations can include the coupling between a system analysis program and a containment analysis program. That is, as the simulation progresses, the system analysis program transmits all or part of its analysis results to the containment analysis program, allowing the containment analysis program to perform further analysis based on the parameters transmitted from the system analysis program. Similarly, the containment analysis program can also transmit all or part of its analysis results to the system analysis program, enabling the system analysis program to perform further analysis based on the parameters transmitted from the containment analysis program.

[0028] However, the overall computational speed of coupled calculations depends on the slowest analysis program. For example, because the containment analysis program uses a fine 3D mesh to simulate the state of a major component of a nuclear reactor, its computational speed is relatively slower than that of the reactor system analysis program. Therefore, as a high-precision analysis program, the containment analysis program, when involved in coupled calculations, will lead to a decrease in the overall efficiency of coupled calculations.

[0029] With the development of high-performance computing and the widespread application of deep learning in various fields, it has become possible to accelerate the computational portion of high-precision computation programs in coupled computing by utilizing neural network algorithms with strong simulation capabilities and high computational speed. Neural networks are essentially mathematical functions with strong fitting capabilities, and they can learn the mathematical relationships hidden in the training dataset for a given input and output. Therefore, based on the coupled computation process, parameter values ​​passed between different programs during the coupled computation can be automatically extracted as the training dataset for the neural network. After training, the fast-computing neural network can replace the slower high-precision computation program, thereby achieving the effect of accelerating computation.

[0030] To address the low efficiency of coupled computation in nuclear reactor simulation and analysis, this application provides a method for accelerating coupled computation in nuclear reactor analysis in some embodiments. This method utilizes neural networks to replace high-precision computation programs. Based on the coupled computation process, a neural network with strong simulation capabilities and high computational speed is used to replace the slow high-precision computation program, achieving accelerated coupled computation while maintaining accuracy. This realizes accelerated coupled computation between the reactor system analysis program and the containment analysis program based on neural networks.

[0031] The method described herein can be applied to the simulation of reactor thermal-hydraulic phenomena during a loss of coolant accident (LOCA) in a pressurized water reactor. Therefore, in some embodiments of this application, the coupled calculation of the reactor thermal-hydraulic phenomena simulation analysis process during a loss of coolant accident in a pressurized water reactor is used as an example. It should be understood that the ideas and specific steps of the method described herein are also applicable to the coupled calculation process of any other program.

[0032] The method can be applied to electronic devices with data processing capabilities. These electronic devices include, but are not limited to, computers, servers, mobile terminals, smart wearable devices, and industrial control machines. For ease of description, this application embodiment uses an electronic device as the execution subject of the method. It should be understood that the method can also be applied to other types of execution subjects, which are not illustrated in this application embodiment. Figure 1 As shown, the method includes: S101. Establish coupled computational model data.

[0033] Before performing coupled calculations, it is necessary to deploy content related to coupled calculations in the electronic equipment, that is, to establish coupled calculation model data. This coupled calculation model data includes the input cards and mesh models of the nuclear reactor's system analysis program and containment analysis program.

[0034] The Reactor System Analysis Program (RSAP) is a computer program used to simulate the thermo-hydraulic behavior of nuclear reactor systems under normal and accident conditions. RSAP can be used to simulate the overall operating state of a nuclear reactor using relatively simple mesh generation and physical models. By deploying and running RSAP within electronic equipment, functional modules for performing system analyses can be formed. These modules can interact with other modules to meet more diverse analytical needs.

[0035] A containment analysis program is a computer program used to simulate the structural integrity and thermo-hydraulic behavior of a nuclear reactor containment under normal operating and accident conditions. The containment analysis program can simulate the state of a major component of a nuclear reactor using a fine-grained three-dimensional mesh, achieving high computational accuracy. Similarly, by deploying and running the containment analysis program in electronic equipment, functional modules for performing containment analysis can be formed within the electronic equipment.

[0036] In some embodiments, the coupled computational model data can be established based on the modeling data. That is, based on the reactor type and loss-of-coolant accident conditions under study, input cards and mesh models for the system analysis program and containment analysis program are established to prepare for coupled computation. Therefore, when establishing the coupled computational model data, the modeling data of the nuclear reactor can be obtained first.

[0037] The modeling data includes reactor type information, loss-of-coolant accident (LOC) condition information, and structural parameter information. Reactor type information can be obtained through design requirements and practical considerations, while LOC accident condition information can be obtained through actual monitoring or simulation analysis. Structural parameter information includes system structural parameters and containment structural parameters. Structural parameter information can be defined during the design process with reference to actual design requirements, or it can be obtained through actual structural measurements of the nuclear reactor used as the simulation object.

[0038] After acquiring the modeling data for the nuclear reactor, simulation conditions can be set, including initial conditions and boundary conditions. For example, initial conditions may include the initial temperature and initial pressure of the nuclear reactor; boundary conditions may include the inlet and outlet conditions of the pressure vessel, thermal boundary conditions, etc.

[0039] Next, the input card is set up based on the modeling data and simulation conditions. For example, after setting the initial and boundary conditions, material properties can be defined, that is, the material properties of the containment vessel and its internal components can be specified in the input card. Then, a specialized tool or script is used to integrate all the above information into the input card, and the input card is automatically generated after ensuring that the format is correct.

[0040] After obtaining the modeling data for the nuclear reactor, a geometric model of the nuclear reactor can be constructed based on the structural parameter information. To meet the analytical needs of system analysis programs and containment analysis programs, the nuclear reactor geometric model can include both a system model and a containment model. For example, a geometric model of the containment can be established based on the geometric structural parameters of the containment. The geometric model of the containment can include the size, shape, and layout of the internal compartments.

[0041] Next, the nuclear reactor geometric model is meshed to obtain a mesh model. Different meshing methods and mesh precision can be used to generate different mesh models for different nuclear reactor geometric models. Specifically, for the system geometric model used by the system analysis program, relatively simple meshing methods such as one-dimensional or two-dimensional meshes can be used to obtain one-dimensional or two-dimensional mesh models, and the resulting mesh model contains fewer meshes. However, for the containment geometric model used by the containment analysis program, more refined meshing methods such as three-dimensional meshes are required to obtain a three-dimensional mesh model, and the three-dimensional mesh model contains a larger number of meshes.

[0042] A mesh model can include multiple mesh nodes, and these nodes are assigned node numbers. For example, when meshing, a finite element analysis application or a dedicated meshing tool can be used based on the geometry of the containment structure. During meshing, nodes in the mesh can be renumbered to minimize computational resources. Then, an appropriate element type, such as tetrahedral or hexahedral elements, can be selected to accommodate different geometries.

[0043] After obtaining the mesh model, the interaction information between the system analysis program and the containment analysis program can be determined. This interaction information includes the coupling surface and the data interaction interface. Based on this interaction information, iterative solution parameters are set for the mesh model.

[0044] For example, to prepare for subsequent coupled calculations, after obtaining the mesh model through mesh generation, the coupling surface between the system analysis program and the containment analysis program can be defined, such as the location of small breaches in the pressure vessel. Then, an interface program is written or existing coupling tools are used to set up a data interaction interface to ensure real-time data transfer between the two programs. Finally, iterative solution parameters, such as convergence criteria and time steps, are set in the input card.

[0045] S102. Construct an alternative model based on the data from the coupled computational model.

[0046] After establishing the coupled computation model data, an alternative model can be constructed based on the coupled computation model data. This alternative model is a neural network model trained using sample transfer parameters, used to replace the high-precision analysis program in the coupled computation process. It enables parameter interaction between the high-precision analysis program and the fast analysis program, thereby obtaining the coupled computation results.

[0047] Therefore, the input parameters of the alternative model can include the parameters passed to the containment analysis program by the system analysis program during the coupling calculation. The output parameters of the alternative model can also include the parameters passed to the system analysis program by the containment analysis program during the coupling calculation. It should be noted that, in this embodiment, the output parameters including the parameters passed to the system analysis program by the containment analysis program during the coupling calculation mean that the parameters output by the alternative model and the parameters output by the containment analysis program in the coupling calculation have the same function, format, and similar parameter values; it does not mean that the alternative model directly outputs the original parameters output by the containment analysis program. In practical applications, the alternative model and the containment analysis program can participate in the coupling calculation process as two independent functional modules.

[0048] In order to build alternative models, such as Figure 2 As shown, in some embodiments, when constructing an alternative model based on coupled computational model data, the analysis target information of the nuclear reactor can be obtained first. This analysis target information includes the physical problems to be studied during the nuclear reactor analysis. The analysis target information can be obtained from documents such as research or design manuals, and is used to indicate the physical problems to be studied. For example, the analysis target information can specify general transient coupling for the nuclear reactor; then the physical problems to be studied include general transient coupling problems.

[0049] After obtaining the analysis target information of the nuclear reactor, a neural network model structure for an alternative model can be set according to the analysis target information. The neural network model structure of the alternative model is adapted to the physical problem to be studied during the nuclear reactor analysis process. That is, the model structure of the neural network model should have a structure adapted to the physical problem to be studied to ensure the convergence and speed of training. For example, for common transient coupling calculation problems, a recurrent neural network (RNN) containing temporal relationships can be used to extract the temporal relationships in the data.

[0050] Next, sample transfer parameters are obtained. These parameters can be generated by collecting the parameters transferred between the system analysis program and the containment analysis program during the coupled calculation process. Specifically, the sample transfer parameters include sample parameters and sample tags. The sample parameters include the parameters transferred from the system analysis program to the containment analysis program during the coupled calculation process. The sample tags are set according to the parameters transferred from the containment analysis program to the system analysis program during the coupled calculation process.

[0051] In some embodiments, to obtain sample transfer parameters, a parameter monitoring tool can be used to monitor the first parameter passed from the system analyzer to the containment analyzer during the coupling calculation process, and generate sample parameters based on the first parameter. Then, the second parameter passed from the containment analyzer to the system analyzer in response to the first parameter is recorded. Based on the second parameter, sample labels are set for the sample parameters to generate associated data pairs. Finally, by combining multiple associated data pairs, sample transfer parameters are generated.

[0052] For example, during coupled computation, the system analysis program can first pass parameter A to the containment analysis program. The electronic device can then obtain parameter A through a parameter monitoring tool and perform preprocessing such as format conversion, noise reduction, and outlier handling to obtain sample parameters. After inputting parameter A into the containment analysis program, the program can perform collaborative analysis calculations based on parameter A to generate parameter B, which is then passed to the system analysis program. At this point, the electronic device can obtain parameter B through the parameter monitoring tool and, after necessary preprocessing, use the preprocessed parameter B as a parameter label for parameter A, establishing a correlation between parameter A and parameter B to generate associated data pairs. By combining multiple sets of associated data pairs, sample transfer parameters are generated.

[0053] After obtaining the sample transfer parameters, model training can be performed, that is, iteratively training the surrogate model using the sample transfer parameters. In some embodiments, when performing iterative training of the surrogate model using the sample transfer parameters, the sample parameters from the sample transfer parameters can first be input into the surrogate model to obtain the training results output by the surrogate model. Then, the training loss is calculated by comparing the training results with the sample labels.

[0054] Then, a preset loss threshold is obtained, and the training loss is compared with the preset loss threshold. If the training loss is greater than the preset loss threshold, it means that the output accuracy of the current replacement model has not reached the preset accuracy requirement, and the replacement model has not yet converged. Therefore, the model parameters of the replacement model can be modified according to the training loss, and iterative training can continue based on the replacement model with modified model parameters. When the training loss is less than or equal to the preset loss threshold, it means that the output accuracy of the previous replacement model has reached the preset accuracy requirement, and the replacement model has been trained to converge. At this time, the model parameters of the replacement model can be output.

[0055] As can be seen, by constructing an alternative model based on the coupled computation model data, a neural network model adapted to the coupled computation can be prepared. This alternative model can be used in subsequent coupled computation processes to replace the containment analysis program in performing coupled computations with the system analysis program. Therefore, the main input of the neural network is the parameters passed from the system program to the containment analysis program during the coupled computation process, which may include some heap-type parameters depending on the generalization requirements of the coupled computation. The main output of the neural network is the parameters passed from the containment analysis program to the system program during the coupled computation process.

[0056] S103. Obtain the real-time transmission parameters during the coupled calculation process.

[0057] After constructing an alternative model based on the coupled computation model data, the alternative model can be used for partial analysis and calculation during the coupled computation process. Therefore, real-time transmission parameters during the coupled computation process can be obtained. These real-time transmission parameters include the parameters transmitted from the system analysis program to the containment analysis program within the current coupled computation cycle.

[0058] like Figure 3 As shown, in some embodiments, a coupling program can be prepared. The coupling program can control the coupled calculation process of the system analysis program and the containment analysis program, and call the system analysis program and the containment analysis program to perform coupled calculations. Therefore, when executing the real-time transfer parameters in the coupled calculation process, the coupling program can be called first, and the original calculation parameters of the system analysis program and the containment analysis program can be read through the coupling program. Then, dimensional checks and unit checks are performed on the original calculation parameters to generate real-time transfer parameters and store the real-time transfer parameters.

[0059] For example, a coupling program can control the computational flow of both the system analysis program and the containment analysis program. Therefore, through the coupling program, parameters from both programs can be read, and after checking their dimensions and units, passed to the other program. The coupling program can also store the parameters passed between the system analysis program and the containment analysis program and pass them to a neural network for training.

[0060] S104. Input the real-time transmitted parameters into the substitution model to obtain the substitution result output by the substitution model.

[0061] By acquiring real-time transmission parameters during the coupled computation process, electronic devices can input these parameters into a replacement model. The replacement model then uses a neural network to extract features and output results based on the input parameters, thereby obtaining the replacement result output by the replacement model.

[0062] In some embodiments, the electronic device can perform coupled computation by invoking a coupling program. During the coupled computation, parameters passed between the system analysis program and the containment analysis program can be transferred to the neural network or to the containment analysis program. Specifically, when executing the real-time input of parameters into the substitution model to obtain the substitution result output by the substitution model, the parameter computation time of the containment analysis program can be recorded through the coupling program. The parameter computation time is the time difference between the moment the first original computational parameters output by the system analysis program are sent to the containment analysis program and the moment the containment analysis program outputs the second original computational parameters based on the first original computational parameters.

[0063] For example, when the system analysis program inputs the first raw calculation parameter (parameter C) into the containment analysis program at time T1, the coupling program can record this input time T1 and monitor the parameter output events of the containment analysis program in real time. When the containment analysis program outputs the second raw calculation parameter (parameter D) for the first raw calculation parameter at time T2, the coupling program can record this output time T2. By calculating the time difference between the input time T1 and the output time T2, the parameter calculation duration ΔT of the containment analysis program can be obtained, i.e., ΔT = T2 - T1.

[0064] After recording the calculation time of the containment analysis program parameters, the electronic device can also obtain a preset time threshold by executing a coupling program. This preset time threshold is set based on the calculation time of the alternative models statistically analyzed over multiple historical coupling calculation periods. For example, the preset time threshold can be the average of the calculation times of the alternative models statistically analyzed over N historical coupling calculation periods. That is, by executing the coupling program, the calculation times of the alternative models over N historical coupling calculation periods can be obtained, namely ΔT1, ΔT2, ..., ΔTN. Then, the average of these N calculation times is calculated to obtain the preset time threshold, i.e., T' = (ΔT1 + ΔT2 + ... + ΔTN) / N.

[0065] By comparing the parameter calculation time with the preset time threshold, if the parameter calculation time is longer than the preset time threshold, it indicates that the current containment analysis program is slower than the analysis speed of the alternative model. Therefore, the parameters can be input into the alternative model in real time through the coupling program, so that the alternative model can be used to replace the containment analysis program for coupled calculation, thereby improving the efficiency of coupled calculation.

[0066] If the parameter calculation time is less than or equal to the preset time threshold, it means that the current containment analysis program is faster than or the same as the analysis speed of the alternative model. Therefore, by using the coupling program to input the parameters into the containment analysis program in real time, the alternative model can be used instead of the containment analysis program for coupling calculation, thereby improving the accuracy of the coupling calculation.

[0067] S105. Generate coupled calculation result information based on the substitution result.

[0068] After obtaining the substitution results output by the substitution model, coupled calculation result information can be generated based on the substitution results. For example, when generating coupled calculation result information based on the substitution results, the substitution results can be sent to the system analysis program, allowing the system analysis program to continue performing collaborative analysis calculations based on the substitution results. Subsequent analysis results obtained by the system analysis program during the collaborative analysis calculation process can still be used as real-time transmission parameters input to the substitution model to obtain corresponding substitution results, until the entire coupled calculation process is completed and the final coupled calculation result is obtained. The coupled calculation result information is then used to output the coupled calculation result.

[0069] In some embodiments, after the neural network model achieves its training objective by transmitting parameter data using batches of samples, data is transferred between the system analysis program and the neural network during subsequent coupled computations via a coupling program. At this point, the coupling program can periodically call the containment analysis program to perform coupled computations to verify the accuracy of the neural network's output. Therefore, the electronic device can periodically call the containment analysis program to perform coupled computations based on real-time transmitted parameters to obtain verification results, and calculate the accuracy error by comparing the verification results with the alternative results.

[0070] Then, the accuracy error is compared with a preset error threshold. If the accuracy error is greater than the preset error threshold, it indicates that the difference between the substitution result output by the substitution model and the analysis result output by the containment analysis program is too large, and the current substitution model cannot meet the coupling calculation requirements. Therefore, the substitution model can continue to be iteratively trained based on the accuracy error until the accuracy error is less than or equal to the preset error threshold. When the accuracy error is less than or equal to the preset error threshold, it indicates that the current substitution model can meet the coupling calculation requirements. Therefore, the coupling calculation result can be output by executing the step of generating coupling calculation result information based on the substitution result.

[0071] For example, the nuclear reactor analysis coupling calculation acceleration method described in the above embodiments can be applied to accurately simulate the transient conditions of a LOCA accident in a reactor. When simulating the transient conditions of a LOCA accident in a reactor, input cards and mesh models for the system analysis program and containment analysis program can be established according to the studied reactor type and loss-of-coolant accident conditions to prepare for coupling calculations. A neural network model adapted to the coupling calculations is then prepared, which will subsequently replace the containment analysis program in coupling calculations with the system analysis program. Simultaneously, a coupling program is prepared that calls the system analysis program and the containment analysis program for coupling calculations. During the coupling calculation process, the parameters passed between the system analysis program and the containment analysis program are transmitted to the neural network, which is then trained in parallel. After the neural network reaches its training objective using batch data, the coupling program controls the data transmission between the system analysis program and the neural network during subsequent coupling calculations.

[0072] By applying the technical solutions of the above embodiments, the nuclear reactor analysis coupling calculation acceleration method described in the embodiments can use a fast-computing neural network to replace the slow-computing containment analysis program, reducing the impact of the containment analysis program's computational efficiency on the overall coupling calculation efficiency. Furthermore, after the neural network training is complete, the slower containment analysis program still performs coupling calculations with the system analysis program without prioritizing high computational speed. The results are then used for neural network calculation verification, which can alleviate the uncertainty in the accuracy of the neural network calculation results.

[0073] In some embodiments, as a refinement and extension of the specific implementation of the above embodiments, and to fully illustrate the specific implementation process of this embodiment, some embodiments of this application also provide a method for accelerating coupled calculations in nuclear reactor analysis. The difference between this method and the above embodiments lies in that, in the step of inputting real-time transferred parameters into the substitution model to obtain the substitution result output by the substitution model, different parameter transfer targets can be selected according to different calculation stages, such as... Figure 4 As shown, the method includes: S201. Obtain the current calculation stage; S202. Determine the parameter transfer target based on the current calculation stage; S203. Send the real-time transmission parameters to the parameter transmission target.

[0074] When inputting real-time passed parameters into the substitution model, the current computation stage can be determined first, which may be either a training stage or an acceleration stage. The current computation stage can be determined based on whether the substitution model has been trained to convergence. For example, the training results output by the substitution model can be obtained by inputting sample passed parameters into the substitution model. Then, the training loss is calculated by comparing the training results with the sample labels. The training loss is then compared with a preset loss threshold. If the training loss is greater than the preset loss threshold, the current computation stage is determined to be the training stage; if the training loss is less than or equal to the preset loss threshold, the current computation stage is determined to be the acceleration stage.

[0075] Next, the parameter transfer target is determined based on the current computation stage, and the real-time transferred parameters are sent to the parameter transfer target. Specifically, if the current computation stage is the training stage, the parameter transfer target is determined to be the containment analysis program and the alternative model. Therefore, the real-time transferred parameters are sent to both the containment analysis program and the alternative model simultaneously, allowing the electronic device to train the alternative model synchronously during the coupled computation. If the current computation stage is the acceleration stage, the parameter transfer target is determined to be the alternative model. Therefore, the real-time transferred parameters are sent to the alternative model, allowing the electronic device to use the alternative model instead of the containment analysis program for coupled computation during the coupled computation.

[0076] For example, such as Figure 5 As shown, when executing the accelerated computation method of coupled nuclear reactor system analysis program and containment analysis program, the neural network can be trained in parallel. That is, during the normal computation process of electronic equipment calling the coupled program, the reactor system analysis program and the containment analysis program interact with each other through the coupled program. Simultaneously, the coupled program passes interaction parameters to the neural network for training. At this stage, the computation speed depends on both the reactor system analysis program and the containment analysis program. After the neural network training is complete, the neural network can be called for accelerated computation. The reactor system analysis program and the neural network interact with each other through the coupled program. Simultaneously, the containment analysis program can be used to verify the computation results of the neural network, ensuring the accuracy of the neural network's calculations. At this stage, the computation speed depends on both the reactor system analysis program and the neural network.

[0077] By applying the technical solutions of the above embodiments, the nuclear reactor analysis coupling calculation acceleration method described in the above embodiments can be based on the coupling calculation solution approach. During the neural network training phase, the user only needs to perform coupling calculations according to the solution approach. In the coupling calculation process, the user initially uses the coupling program normally, following the same process as other coupling programs. Simultaneously, the coupling program automatically passes the parameters passed during the coupling process for parallel neural network training. During neural network training, the user does not need to organize training data. Furthermore, after the neural network training is completed, the replacement model can automatically replace the containment analysis program in calculating the coupling parameters, improving the efficiency of coupling calculations. Asynchronous calculations can also be performed through the containment analysis program to verify the accuracy of the neural network calculation results.

[0078] In some embodiments, as a specific implementation of the nuclear reactor analysis coupled computation acceleration method in the above embodiments, some embodiments of this application also provide a nuclear reactor analysis coupled computation acceleration system, such as... Figure 6 As shown, the system includes: The coupling condition establishment module is used to establish coupling calculation model data, which includes the input cards and mesh models of the nuclear reactor system analysis program and containment analysis program; An alternative model construction module is used to construct an alternative model based on coupled computation model data. The alternative model is a neural network model trained using sample transfer parameters. The input parameters of the alternative model include parameters passed to the containment analysis program by the system analysis program during the coupled computation process. The output parameters of the alternative model include parameters passed to the system analysis program by the containment analysis program during the coupled computation process. The parameter acquisition module is used to acquire real-time transmission parameters during the coupling calculation process. The real-time transmission parameters include the parameters that the system analysis program transmits to the containment analysis program within the current coupling calculation cycle. The substitution calculation module is used to input real-time transmitted parameters into the substitution model to obtain the substitution results output by the substitution model. The results output module is used to generate coupled calculation result information based on the substitution results.

[0079] By applying the technical solutions of the above embodiments, the nuclear reactor analysis coupling calculation acceleration system provided in the above embodiments can establish the input cards and mesh models of the system analysis program and containment analysis program of the nuclear reactor through the coupling condition establishment module, and then construct the alternative model by the alternative model construction module. The alternative model is a neural network model trained using sample transfer parameters. During coupling calculation, the transfer parameter acquisition module can acquire the real-time transfer parameters transferred from the system analysis program to the containment analysis program. The alternative calculation module then inputs the real-time transfer parameters into the alternative model to obtain the alternative results output by the alternative model. Finally, the result output module generates coupling calculation result information based on the alternative results. The system can utilize a fast-computing neural network model to replace the slow-computing containment analysis program, reducing the impact of the containment analysis program's computational efficiency on the overall efficiency of coupling calculation and solving the problem of low coupling calculation efficiency in nuclear reactor simulation analysis.

[0080] It should be noted that other corresponding descriptions of the functional units involved in the nuclear reactor analysis coupled computation acceleration system provided in the embodiments of this application can be found in the corresponding descriptions in the nuclear reactor analysis coupled computation acceleration method provided in the above embodiments, and will not be repeated here.

[0081] This application also provides a computer device, specifically a personal computer, server, network device, etc. The computer device includes a bus, processor, memory, and communication interface, and may also include input / output interfaces and a display device. The processor of the computer device provides computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system, computer programs, and a database. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The database of the computer device stores location information. The network interface of the computer device is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements the steps in the various method embodiments.

[0082] Those skilled in the art will understand that the structure of the computer device described above is only a partial structure related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. A specific computer device may include more or fewer components, or combine certain components, or have different component arrangements.

[0083] In one embodiment, a computer-readable storage medium is also provided, which may be non-volatile or volatile, having stored thereon a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0084] In one embodiment, a computer program product is also provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0085] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties.

[0086] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the methods described above.

[0087] Any references to memory, database, or other media used in the embodiments provided in this application may include at least one of non-volatile and volatile memory. Non-volatile memory may include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc.

[0088] Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM can take many forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM).

[0089] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, distributed databases based on blockchain. The processors involved in the embodiments provided in this application may be, but are not limited to, general-purpose processors, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc.

[0090] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0091] The above embodiments merely illustrate several implementation methods of this application, and their descriptions are relatively specific and detailed, but they should not be construed as limiting the scope of this application's patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for accelerating coupled computational analysis of nuclear reactors, characterized in that, The method includes: Establish coupled computational model data, which includes the input cards and mesh models of the nuclear reactor system analysis program and containment analysis program; An alternative model is constructed based on the coupled computation model data. The alternative model is a neural network model trained using sample transfer parameters. The input parameters of the alternative model include the parameters passed to the containment analysis program by the system analysis program during the coupled computation process. The output parameters of the alternative model include the parameters passed to the system analysis program by the containment analysis program during the coupled computation process. Obtain real-time transmission parameters during the coupling calculation process, the real-time transmission parameters including the parameters transmitted by the system analysis program to the containment analysis program within the current coupling calculation cycle; The real-time transmission parameters are input into the substitution model to obtain the substitution result output by the substitution model; Based on the substitution results, couple calculation result information is generated.

2. The method according to claim 1, characterized in that, The method further includes: The containment analysis program is periodically invoked to perform coupled calculations based on the real-time transmitted parameters in order to obtain verification results. The accuracy error is calculated by comparing the verification results and the alternative results. If the accuracy error is greater than a preset error threshold, the alternative model is iteratively trained based on the accuracy error; If the accuracy error is less than or equal to a preset error threshold, the step of generating coupled calculation result information based on the substitution result is executed.

3. The method according to claim 1, characterized in that, Constructing an alternative model based on the coupled computational model data includes: Acquire analytical target information of the nuclear reactor, including the physical problems to be studied during the nuclear reactor analysis process; The neural network model structure of the alternative model is set according to the analysis target information, and the neural network model structure of the alternative model is adapted to the physical problem to be studied in the nuclear reactor analysis process; The sample transfer parameters are obtained, including sample parameters and sample labels. The sample parameters include parameters passed from the system analysis program to the containment analysis program during the coupling calculation process. The sample labels are set according to the parameters passed from the containment analysis program to the system analysis program during the coupling calculation process. The alternative model is trained iteratively using the sample transfer parameters.

4. The method according to claim 3, characterized in that, Obtaining the sample transfer parameters includes: Monitor the first parameter passed from the system analysis program to the containment analysis program during the coupling calculation process; The sample parameters are generated based on the first parameter; Record the second parameter that the containment analysis program passes to the system analysis program in response to the first parameter; Based on the second parameter, set sample labels for the sample parameters to generate associated data pairs; The sample transfer parameters are generated by combining multiple of the associated data pairs.

5. The method according to claim 3, characterized in that, Iteratively training the alternative model using the sample transfer parameters includes: The sample parameters in the sample transfer parameters are input into the alternative model to obtain the training results output by the alternative model; The training loss is calculated by comparing the training results with the sample labels. If the training loss is greater than a preset loss threshold, the model parameters of the alternative model are modified according to the training loss. If the training loss is less than or equal to a preset loss threshold, output the model parameters of the alternative model.

6. The method according to claim 1, characterized in that, Establish the coupled computational model data, including: Acquire modeling data for the nuclear reactor, including reactor type information, loss-of-coolant accident operating condition information, and structural parameter information; the structural parameter information includes system structural parameters and containment structure parameters. Set simulation conditions, including initial conditions and boundary conditions; Set the input card according to the modeling data and the simulation conditions; A nuclear reactor geometric model is constructed based on the structural parameter information. The nuclear reactor geometric model includes a system model and a containment model. The nuclear reactor geometric model is meshed to obtain the mesh model; the mesh model includes multiple mesh nodes, and the mesh nodes are assigned node numbers. Determine the interaction information between the system analysis program and the containment analysis program, the interaction information including the coupling surface and the data interaction interface; Based on the interaction information, iterative solution parameters are set for the mesh model.

7. The method according to claim 1, characterized in that, Obtain real-time transmitted parameters during the coupled computation process, including: The coupling program is invoked, which controls the coupled calculation process of the system analysis program and the containment analysis program; The coupling program reads the original calculation parameters of the system analysis program and the containment analysis program. Perform dimensional and unit checks on the original calculation parameters to generate the real-time transmission parameters; Store the real-time transmission parameters.

8. The method according to claim 7, characterized in that, The real-time transmission parameters are input into the substitution model to obtain the substitution result output by the substitution model, including: The coupling program records the parameter calculation time of the containment analysis program. The parameter calculation time is the time difference between the moment when the first raw calculation parameter output by the system analysis program is sent to the containment analysis program and the moment when the containment analysis program outputs the second raw calculation parameter for the first raw calculation parameter. Obtain a preset duration threshold, which is set based on the calculation duration of the alternative model statistically analyzed within multiple historical coupling calculation cycles; If the parameter calculation time exceeds the preset time threshold, the real-time transmission parameter is input into the alternative model through the coupling procedure; If the calculation time of the parameter is less than or equal to the preset time threshold, the real-time transmission parameter is input into the containment analysis program through the coupling program.

9. The method according to claim 1, characterized in that, The real-time transmission parameters are input into the substitution model to obtain the substitution result output by the substitution model, including: Obtain the current computation stage, which is either the training stage or the acceleration stage; The parameter transfer target is determined based on the current computation stage; wherein, if the current computation stage is the training stage, the parameter transfer target is the containment analysis program and the alternative model; if the current computation stage is the acceleration stage, the parameter transfer target is the alternative model. The real-time transmission parameters are sent to the parameter transmission target.

10. A nuclear reactor analysis coupled computation acceleration system, characterized in that, The system includes: The coupling condition establishment module is used to establish coupling calculation model data, which includes the input cards and mesh models of the nuclear reactor system analysis program and containment analysis program; An alternative model construction module is used to construct an alternative model based on the coupled computation model data. The alternative model is a neural network model trained using sample transfer parameters. The input parameters of the alternative model include the parameters passed to the containment analysis program by the system analysis program during the coupled computation process. The output parameters of the alternative model include the parameters passed to the system analysis program by the containment analysis program during the coupled computation process. The parameter acquisition module is used to acquire real-time transmission parameters during the coupling calculation process. The real-time transmission parameters include the parameters transmitted by the system analysis program to the containment analysis program within the current coupling calculation cycle. The substitution calculation module is used to input the real-time transmission parameters into the substitution model to obtain the substitution result output by the substitution model; The result output module is used to generate coupled calculation result information based on the substitution result.

Images