CN115859783A - Multi-fidelity network construction method and device for nuclear reactor simulation test - Google Patents
Multi-fidelity network construction method and device for nuclear reactor simulation test Download PDFInfo
- Publication number
- CN115859783A CN115859783A CN202211447961.8A CN202211447961A CN115859783A CN 115859783 A CN115859783 A CN 115859783A CN 202211447961 A CN202211447961 A CN 202211447961A CN 115859783 A CN115859783 A CN 115859783A
- Authority
- CN
- China
- Prior art keywords
- fidelity
- network
- data
- trained
- fidelity network
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000004088 simulation Methods 0.000 title claims abstract description 89
- 238000012360 testing method Methods 0.000 title claims abstract description 53
- 238000010276 construction Methods 0.000 title claims abstract description 17
- 238000012549 training Methods 0.000 claims abstract description 69
- 238000000034 method Methods 0.000 claims abstract description 40
- 238000004590 computer program Methods 0.000 claims description 42
- 238000003860 storage Methods 0.000 claims description 11
- 230000008878 coupling Effects 0.000 abstract description 4
- 238000010168 coupling process Methods 0.000 abstract description 4
- 238000005859 coupling reaction Methods 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 14
- 238000013528 artificial neural network Methods 0.000 description 13
- 238000010586 diagram Methods 0.000 description 7
- 238000004364 calculation method Methods 0.000 description 5
- 238000005259 measurement Methods 0.000 description 5
- 230000008859 change Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000009257 reactivity Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 3
- 238000013461 design Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 230000004913 activation Effects 0.000 description 2
- 230000003042 antagnostic effect Effects 0.000 description 2
- 230000006399 behavior Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000009792 diffusion process Methods 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 239000000446 fuel Substances 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 238000005457 optimization Methods 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- OKTJSMMVPCPJKN-UHFFFAOYSA-N Carbon Chemical compound [C] OKTJSMMVPCPJKN-UHFFFAOYSA-N 0.000 description 1
- 241000282414 Homo sapiens Species 0.000 description 1
- 238000000342 Monte Carlo simulation Methods 0.000 description 1
- 229910052768 actinide Inorganic materials 0.000 description 1
- 150000001255 actinides Chemical class 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 238000005253 cladding Methods 0.000 description 1
- 239000002826 coolant Substances 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000004992 fission Effects 0.000 description 1
- 239000012530 fluid Substances 0.000 description 1
- 229910021389 graphene Inorganic materials 0.000 description 1
- 238000000265 homogenisation Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 231100000572 poisoning Toxicity 0.000 description 1
- 230000000607 poisoning effect Effects 0.000 description 1
- 238000011158 quantitative evaluation Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 241000894007 species Species 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 229910052724 xenon Inorganic materials 0.000 description 1
- FHNFHKCVQCLJFQ-UHFFFAOYSA-N xenon atom Chemical compound [Xe] FHNFHKCVQCLJFQ-UHFFFAOYSA-N 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Mathematical Physics (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Geometry (AREA)
- Medical Informatics (AREA)
- Computer Hardware Design (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Monitoring And Testing Of Nuclear Reactors (AREA)
Abstract
The application relates to a multi-fidelity network construction method and device for nuclear reactor simulation testing. The method comprises the following steps: obtaining a first fidelity network from first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor; training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network; combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor. By adopting the method, the final simulation result can be output according to the coupling between the data with different fidelity degrees, so that the simulation efficiency is improved while the simulation precision is ensured.
Description
Technical Field
The present application relates to the field of nuclear power plant reactor core design and operation technologies, and in particular, to a multi-fidelity network construction method and apparatus, a computer device, a storage medium, and a computer program product for nuclear reactor simulation testing.
Background
During the design and operation of a nuclear reactor, simulations of actual or assumed operating conditions of the reactor are required to verify the safety of the designed or operated reactor. On one hand, aiming at the designed reactor, quantitative evaluation needs to be carried out on various operation boundaries and operation consequences of the reactor under various assumed accident conditions so as to ensure the safety of the designed reactor; on the other hand, in the operating reactor, various simulation calculations are required to be carried out, so that the design calculated parameters are consistent with the actual operating parameters within a certain error range, and the consistency of the designed reactor and the actual reactor is further ensured, thereby ensuring that the operating reactor has enough safety boundaries under various accident conditions. In addition, parameters directly related to the safety of reactor operation (e.g., fuel rod cladding temperature, etc.) cannot be measured directly, and must be derived from measurable reactor fluid parameters (e.g., temperature, pressure, etc.) or neutron detector readings (e.g., indicative of fission reactivity) in conjunction with a simulation model of the reactor. These models are approximate abstractions of the true reactor process, and can only partially model in-reactor neutron behavior as well as burnup behavior. Equivalent homogenization assumptions and neutron diffusion approximations are employed as in the nuclear design software package PCM to achieve the simulation of a three-dimensional core. The dot-pile equation, although not spatially distributed, is commonly used for inverse monitoring based on reactivity of measured power changes and xenon poisoning. These models are collectively called a state transition model, and represent a change process in the core state due to the control action transition to the next time state. The state transition model is used for not only the state distribution of the non-measurable variable at the current moment, but also the prediction of the core state at the subsequent moment, and unknown errors exist.
At present, a great number of theoretical reactor research results are available, including high fidelity models (such as Monte Carlo method, transport theory or diffusion approximation) and low fidelity models (such as point reactor dynamics), but the models are obtained by data modeling with single fidelity, and have great defects in the use process. The computational efficiency of the mathematical equation model with higher fidelity is difficult to meet the real-time requirement, and the accuracy requirement is difficult to meet by introducing excessive simplifying assumptions with lower fidelity.
Therefore, the current nuclear reactor simulation cannot improve the simulation efficiency while ensuring the accuracy.
Disclosure of Invention
In view of the foregoing, it is desirable to provide a multi-fidelity network construction method, apparatus, computer device, computer readable storage medium and computer program product for nuclear reactor simulation testing, which can improve simulation efficiency while ensuring simulation accuracy.
In a first aspect, the present application provides a method for constructing a multi-fidelity network for nuclear reactor simulation testing. The method comprises the following steps:
obtaining a first fidelity network from the first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from the second fidelity data of the sample nuclear reactor;
training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network;
combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In one embodiment, prior to obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor, the method further comprises:
acquiring a plurality of datasets of a sample nuclear reactor;
determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data;
and taking data except the first fidelity data in the plurality of data sets as second fidelity data.
In one embodiment, obtaining at least one second fidelity network from second fidelity data of a sample nuclear reactor comprises:
grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade;
acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of the second fidelity networks is the same as the number of the fidelity grades;
training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network, comprising:
and respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
In one embodiment, combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network comprises:
respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network;
and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
In one embodiment, the training of the multi-fidelity network with the first fidelity data to obtain the trained multi-fidelity network includes:
the multi-fidelity network is used as a generator network, and a corresponding discriminator network is obtained according to the generator network;
constructing a generation countermeasure network according to the generator network and the discriminator network;
training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network;
and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
In one embodiment, the method further comprises:
acquiring a control parameter and a first time state parameter of a target nuclear reactor;
inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor;
and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
In a second aspect, the application further provides a multi-fidelity network building apparatus for nuclear reactor simulation testing. The device comprises:
an acquisition module to acquire a first fidelity network from first fidelity data of a sample nuclear reactor and to acquire at least one second fidelity network from second fidelity data of the sample nuclear reactor;
the training module is used for training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network;
the combination module is used for combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In a third aspect, the present application also provides a computer device. The computer device comprises a memory storing a computer program and a processor implementing the following steps when executing the computer program:
obtaining a first fidelity network from first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor;
training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network;
combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of:
obtaining a first fidelity network from first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor;
training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network;
combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In a fifth aspect, the present application further provides a computer program product. The computer program product comprising a computer program which when executed by a processor performs the steps of:
obtaining a first fidelity network from first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor;
training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network;
combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
The multi-fidelity network construction method, apparatus, computer device, storage medium, and computer program product for nuclear reactor simulation testing described above obtain a first fidelity network from first fidelity data of a sample nuclear reactor, and obtain at least one second fidelity network from second fidelity data of the sample nuclear reactor; training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network; combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor. Therefore, when the target nuclear reactor is simulated through the multi-fidelity network, the second fidelity network can obtain a reference simulation result according to the input parameters, the input parameters and the simulation result are input into the first fidelity network together, the first fidelity network can output a final simulation result according to coupling among different fidelity data, and therefore simulation accuracy is guaranteed while simulation efficiency is improved.
Drawings
FIG. 1 is a schematic flow diagram of a method for constructing a multi-fidelity network for nuclear reactor simulation testing, according to an embodiment;
FIG. 2 is a diagram illustrating a second fidelity network in accordance with an embodiment;
FIG. 3 is a schematic diagram of a multi-fidelity network in one embodiment;
FIG. 4 is a schematic flow chart of generation of confrontation network training in another embodiment;
FIG. 5 is a block diagram of an embodiment of a multi-fidelity network building apparatus for nuclear reactor simulation testing;
FIG. 6 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more clearly understood, the present application is further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
In an embodiment, as shown in fig. 1, a multi-fidelity network construction method for nuclear reactor simulation testing is provided, and this embodiment is illustrated by applying the method to a computer device, and it is understood that the computer device may be a terminal or a server. The terminal may be, but is not limited to, various industrial computers. The server may be implemented as a stand-alone server or as a server cluster comprised of multiple servers. In this embodiment, the method includes the steps of:
The first fidelity data refers to high fidelity data and can be generated through high fidelity software; the second fidelity data is low fidelity data having a lower data precision than the first fidelity data and can be generated quickly by low fidelity software. The first fidelity data and the second fidelity data both comprise a plurality of control parameters and state parameters of input and output of the sample nuclear reactor, and specifically comprise: 1) Reactivity parameters, such as control rod position, etc.; 2) Power parameters, such as power level, power distribution; 3) Nuclear density parameters such as nuclear density (including fissile species, minor actinides, light nuclei, etc.) over the axial height of each assembly; 4) Macroscopic or microscopic reaction cross-section; 5) Thermal parameters such as coolant temperature, pressure, flow rate, etc., temperature of the fuel or material, etc. The reactivity parameter and the power parameter can be used as input parameters, and other state parameters can be used as output parameters.
Optionally, a suitable neural network is selected as the first fidelity network according to the simulation requirements of the target nuclear reactor and the data type of the first fidelity data. Similarly, another one or more suitable neural networks are selected as the second fidelity network, respectively, based on the simulation requirements of the targeted nuclear reactor and the data type of the second fidelity data.
And 104, training the at least one second fidelity network by adopting the second fidelity data to obtain at least one trained second fidelity network.
Optionally, all the second fidelity data may be directly used to train a second fidelity network, so as to obtain a trained second fidelity network. Or the second fidelity data can be accurately distinguished, each group of data in the second fidelity data is subjected to accuracy grading according to the state characteristics of the sample nuclear reactor, all the data are divided into a plurality of fidelity grades according to the accuracy grading, second fidelity networks with the same number as the fidelity grades are prepared, and each second fidelity network is trained by adopting data with one fidelity grade. For example, the second fidelity data is divided into two parts according to the precision, the data with higher precision is used as the medium fidelity data, the data with lower precision is used as the low fidelity data, and two second fidelity networks are prepared: and the low-fidelity network is trained by adopting the low-fidelity data to obtain the trained low-fidelity network.
Optionally, the output end of each trained second fidelity network is connected to the input end of the first fidelity network; and taking the input end of the first fidelity network and the input end of each trained second fidelity network as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network. When the multi-fidelity network is trained, input data are firstly input into each trained second fidelity network, each trained second fidelity network outputs respective simulation results, the simulation results are used as reference data and input data are input into the first fidelity network together, the first fidelity network combines the reference data to process the input data and output simulation data, the simulation data are compared with label data corresponding to the input data, and weight parameters of the first fidelity network are adjusted. And when the comparison result of the simulation data output by the first fidelity network and the label data corresponding to the input data meets the simulation requirement, finishing training to obtain the trained multi-fidelity network. In the whole training process, the trained second fidelity networks are not adjusted, namely, the first fidelity data are only used for training the first fidelity network.
In a feasible implementation mode, the control parameters and the current state parameters of the target nuclear reactor are obtained, then the control parameters and the current state parameters are input into a trained multi-fidelity network, the multi-fidelity network can output predicted state parameters, the predicted state parameters can represent changes of the state parameters of the target nuclear reactor under the influence of the control parameters, and the target nuclear reactor can be subjected to simulation test based on the predicted state parameters, so that the optimal control mode and control parameters of the target nuclear reactor are determined.
In the multi-fidelity network construction method for the nuclear reactor simulation test, a first fidelity network is obtained according to first fidelity data of a sample nuclear reactor, and at least one second fidelity network is obtained according to second fidelity data of the sample nuclear reactor; training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network; combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor. Therefore, when the target nuclear reactor is simulated through the multi-fidelity network, the second fidelity network can obtain a reference simulation result according to the input parameters, the input parameters and the simulation result are input into the first fidelity network together, the first fidelity network can output a final simulation result according to coupling among different fidelity data, and therefore simulation accuracy is guaranteed while simulation efficiency is improved.
In one embodiment, prior to obtaining the at least one second fidelity network from the second fidelity data of the sample nuclear reactor, further comprising: acquiring a plurality of datasets of a sample nuclear reactor; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data.
Alternatively, the first and second fidelity data for the sample nuclear reactor may be generated by a plurality of different fidelity software, which are extrapolated based on mathematical physics equations. Actual measurement data under partial operating conditions can be manually acquired, the actual measurement data can be stored as highest-level fidelity data or low-fidelity data, and the actual measurement data are determined based on the essence of data acquisition and data acquisition conditions, so that the data need to be subjected to fidelity division, the high-fidelity data are used as first fidelity data, and the rest of data are used as second fidelity data. A uniform data format is required between different fidelity data.
Specifically, the state space parameter of the sample nuclear reactor is defined as
High fidelity data sets are generated using high fidelity software. Because the high-fidelity simulation software has higher calculation precision but lower calculation efficiency. For example, a typical reactor state point calculation is performed on the order of minutes or hours. High fidelity data is relatively scarce. Assume that the input to the high fidelity simulation software is @>
The output is->
Obtaining high-fidelity input and output pairs:
since the process of the reactor can be seen as a markov process in nature, its input state space and output state space can be consistent in nature. Similarly, the input and output pairs of the medium fidelity are constructed:
and low fidelity input-output pairs: />
In this embodiment, a plurality of data sets of a sample nuclear reactor are acquired; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data. Data of different fidelity can be obtained.
In one embodiment, obtaining at least one second fidelity network from second fidelity data of a sample nuclear reactor comprises: grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of second fidelity networks is the same as the number of fidelity levels.
Further, training at least one second fidelity network by using second fidelity data to obtain at least one trained second fidelity network, including: and respectively adopting subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
Optionally, taking the example that the second fidelity network includes a medium fidelity network and a low fidelity network, for the two second fidelity networks, a standard neural network is selected, and fitting training is performed on input and output, so as to achieve real fidelityNow that
The network construction of (1). As shown in fig. 2, the number (or depth) m of layers of the neural network, the number (or width) n of nodes of each layer, the activation function ReLU or leakyReLu, the learning rate, the optimization algorithm Adam or SGD, and the like, may be set according to different problems for the hyper-parameters of the neural network training, or may finally determine the relevant parameters according to the hyper-parameter optimization algorithm HPO. The methods of constructing and training the neural network and the like can be selected according to the data type, the training requirement and the like, and are not described in detail herein. An optimized neural network model can be easily implemented to characterize low fidelity data or intrinsic relationships of the low fidelity data using a number of deep learning open source platforms, including paddlepaddley, pytorch, tenserflow, etc.
In this embodiment, the second fidelity data is subjected to level division to obtain at least one fidelity level and the subdata corresponding to each fidelity level; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of the second fidelity networks is the same as the number of the fidelity grades; and respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network. A plurality of trained second fidelity networks can be obtained.
In one embodiment, combining the at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network comprises: respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network; and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
Optionally, the first fidelity network is a high fidelity network, and the second fidelity network includes a medium fidelity network and a low fidelity network, for example, becauseA large amount of data exist in a low-fidelity network, the high-fidelity data volume is relatively small, and large overfitting errors are brought by directly training a high-fidelity data source. Thus, the data x is input with high fidelity high Put into a low fidelity network, a medium fidelity network and the like which are trained well to obtain the output under the several non-high fidelity network grades
And &>
In fact, is greater or less>
Or>
With true y high The difference in (2) characterizes the error of the low fidelity network extending to high fidelity data.
Further, through the learning training of the first fidelity network, the related error can be corrected. Therefore, a first fidelity network is constructed, and as shown in fig. 3, the output of the low fidelity network and the output network of the medium fidelity network are mapped to one of the inputs of the high fidelity network and combined to obtain a multi-fidelity network. Wherein the depth and width m of the multi-fidelity network high And n high The activation function, the training hyper-parameter, and the like, need to be determined according to the state parameters of the specific nuclear reactor, and are not described herein again.
In this embodiment, the output end of each trained second fidelity network is connected to one of the input ends of the first fidelity network; and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network. The multi-fidelity network can combine the output results of different fidelity networks to obtain a final simulation result, and a second fidelity network part in the multi-fidelity network is obtained through a large amount of low-fidelity data training, so that the calculation efficiency and precision of the training can be greatly improved, and the real-time requirement can be met.
In one embodiment, training the multi-fidelity network with the first fidelity data to obtain a trained multi-fidelity network comprises: the multi-fidelity network is used as a generator network, and a corresponding discriminator network is obtained according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
Optionally, taking the first fidelity network as a high fidelity network, and the second fidelity network comprising a medium fidelity network and a low fidelity network as an example, as shown in fig. 3, based on the output of the generator network
And y high The traditional neural network training is directly utilized>
And y high Such as the L1loss function: />
Training can be performed directly.
Based on the above, adding artificially well-designed noise to a real sample to synthesize new input data (or called countermeasure data) will cause the prediction error of the whole producer network. That is, the deep neural networks are vulnerable to defeat data and can be easily confused by the deep generator networks. The prediction error rate of the deep neural network on the antagonistic sample is very high, and under the condition that human beings think that the difference between the original sample and the antagonistic sample can hardly be distinguished (the deviation is very small), the original function of the deep neural network is failed and can not be used as the basis for predicting the subsequent state of the reactor. In particular, all measurement parameters are accompanied by random errors which can be eliminated in the actual measurement process of the nuclear power plant, and some potential error disturbance can cause catastrophic error results of deep neural network model prediction. Improving the robustness of the neural network (the ability to resist challenge samples) is intuitively important for the development of reactor prediction models. Therefore, a discriminator is added based on the idea of counterstudy. By using the method of confrontation training, the capability of the neural network in confronting sample cheating is improved. The basic idea of the confrontation training is to continuously generate and learn confrontation samples in the network training process. So that the reactor status label generated by the generator G (X) can fool the arbiter into agreement with the true reactor status label. I.e., the output of the discriminator D (Y) is the likelihood of being from a genuine tag. And the network structure, the learning rate and other hyper-parameters of the discriminator are adjusted according to a specific actual scene.
Further, as shown in FIG. 4, for generator G (X), it is necessary to continually fool discriminator D so that log (D (G (X))) is maximized, i.e., so that
And y high As close as possible. For the discriminator D, continuous learning is needed to prevent cheating by the generator, at this time, aiming at the real input y high The determination that needs to be maximized is true correctly, while it is true for false inputs>
The decision to achieve the maximum is false. So that log (D (Y)) + log (1-D (G (X))) is maximized. The specific training process is to train the discriminator D first and then the generator G until the discriminator D and the generator G reach a Nash equilibrium.
In the embodiment, a multi-fidelity network is used as a generator network, and a corresponding arbiter network is obtained according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network. The trained multi-fidelity network can be obtained, and the trained multi-fidelity network can be combined with output results of different fidelity networks to obtain a simulation result corresponding to input data.
The trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor pair, namely, the target nuclear reactor can carry out simulation test through the trained multi-fidelity network so as to realize simulation of state change of the target reactor. Specifically, in one embodiment, the step of performing simulation testing on the target nuclear reactor through the trained multi-fidelity network comprises: acquiring a control parameter and a first time state parameter of a target nuclear reactor; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
Optionally, taking the first fidelity network as a high fidelity network, and the second fidelity network including a medium fidelity network and a low fidelity network as an example, the multi-fidelity network obtained by combining the low fidelity network, the medium fidelity network, and the high fidelity network may implement the change of the reactor state
The simulation of (2). As shown in FIG. 3, in the practical application process, x is input into the multi-fidelity network high ，x high The control parameters and the first time state parameters of the target nuclear reactor are input data x processed by a low fidelity network and a medium fidelity network high Respectively output->
And &>
Then x is put high Input into the high fidelity network and will also->
And &>
Inputting high fidelity network, outputting reactor state simulation change
Including a second time-of-day status parameter of the target nuclear reactor. Wherein the relationship between the first time and the second time depends on the training data (x) of the multi-fidelity network during the training process high ,y high ) In x high And y high The time of day relationship between.
In the embodiment, control parameters and state parameters at a first moment of a target nuclear reactor are obtained; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment. The final simulation result can be output according to the coupling between different fidelity data, so that the simulation efficiency is improved while the simulation precision is ensured.
In one embodiment, a multi-fidelity network construction for nuclear reactor simulation testing, comprises:
acquiring a plurality of datasets of a sample nuclear reactor; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data.
A first fidelity network is obtained from first fidelity data of a sample nuclear reactor. Grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of second fidelity networks is the same as the number of fidelity levels.
And respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
Respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network; and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
The multi-fidelity network is used as a generator network, and a corresponding discriminator network is obtained according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain a trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to be used as a trained multi-fidelity network, wherein the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
The method comprises the following specific steps of carrying out simulation test on a target nuclear reactor through a trained multi-fidelity network: acquiring a control parameter and a first time state parameter of a target nuclear reactor; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
It should be understood that, although the steps in the flowcharts related to the embodiments as described above are sequentially displayed as indicated by arrows, the steps are not necessarily performed sequentially as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a part of the steps in the flowcharts related to the embodiments described above may include multiple steps or multiple stages, which are not necessarily performed at the same time, but may be performed at different times, and the execution order of the steps or stages is not necessarily sequential, but may be rotated or alternated with other steps or at least a part of the steps or stages in other steps.
Based on the same inventive concept, the embodiment of the application also provides a multi-fidelity network construction device for the nuclear reactor simulation test, which is used for realizing the multi-fidelity network construction method for the nuclear reactor simulation test. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, so specific limitations in one or more embodiments of the multi-fidelity network construction device for nuclear reactor simulation tests provided below can be referred to the limitations in the above multi-fidelity network construction method for nuclear reactor simulation tests, and are not described herein again.
In one embodiment, as shown in FIG. 5, there is provided a multi-fidelity network building apparatus 500 for nuclear reactor simulation testing, comprising: an acquisition module 501, a training module 502, and a combination module 503, wherein:
an obtaining module 501 is configured to obtain a first fidelity network from first fidelity data of a sample nuclear reactor, and obtain at least one second fidelity network from second fidelity data of the sample nuclear reactor.
A training module 502, configured to train the at least one second fidelity network with the second fidelity data to obtain at least one trained second fidelity network.
The combining module 503 is configured to combine the at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and train the multi-fidelity network by using the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In one embodiment, the acquisition module 501 is further configured to acquire a plurality of data sets of a sample nuclear reactor; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data.
In an embodiment, the obtaining module 501 is further configured to perform level division on the second fidelity data to obtain at least one fidelity level and sub data corresponding to each fidelity level; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of second fidelity networks is the same as the number of fidelity levels.
The training module 502 is further configured to train each second fidelity network by using the sub-data corresponding to each fidelity level, respectively, to obtain at least one trained second fidelity network.
In one embodiment, the combining module 503 is further configured to connect the output terminals of the trained second fidelity networks to one of the input terminals of the first fidelity network; and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
In one embodiment, the combining module 503 is further configured to use the multi-fidelity network as a generator network, and obtain a corresponding arbiter network according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
In one embodiment, the apparatus further comprises:
a test module 504 for obtaining control parameters and first time state parameters of a target nuclear reactor; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
The various modules in the multi-fidelity network construction apparatus for nuclear reactor simulation testing described above may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 6. The computer device comprises a processor, a memory, an Input/Output (I/O) interface and a communication interface. The processor, the memory and the input/output interface are connected through a system bus, and the communication interface is connected to the system bus through the input/output interface. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store neural network data. The input/output interface of the computer device is used for exchanging information between the processor and an external device. The communication interface of the computer device is used for connecting and communicating with an external terminal through a network. The computer program is executed by a processor to implement a method of multi-fidelity network construction for nuclear reactor simulation testing.
Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program: obtaining a first fidelity network from the first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from the second fidelity data of the sample nuclear reactor; training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network; combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a plurality of datasets of a sample nuclear reactor; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data.
In one embodiment, the processor, when executing the computer program, further performs the steps of: grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of second fidelity networks is the same as the number of fidelity levels.
In one embodiment, the processor, when executing the computer program, further performs the steps of: and respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
In one embodiment, the processor, when executing the computer program, further performs the steps of: respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network; and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
In one embodiment, the processor, when executing the computer program, further performs the steps of: the multi-fidelity network is used as a generator network, and a corresponding discriminator network is obtained according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
In one embodiment, the processor, when executing the computer program, further performs the steps of: acquiring a control parameter and a first time state parameter of a target nuclear reactor; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of: obtaining a first fidelity network from first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor; training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network; combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a plurality of datasets of a sample nuclear reactor; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data.
In one embodiment, the computer program when executed by the processor further performs the steps of: grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of second fidelity networks is the same as the number of fidelity levels.
In one embodiment, the computer program when executed by the processor further performs the steps of: and respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
In one embodiment, the computer program when executed by the processor further performs the steps of: respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network; and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
In one embodiment, the computer program when executed by the processor further performs the steps of: taking the multi-fidelity network as a generator network, and acquiring a corresponding discriminator network according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a control parameter and a first time state parameter of a target nuclear reactor; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
In one embodiment, a computer program product is provided, comprising a computer program which, when executed by a processor, performs the steps of:
obtaining a first fidelity network from first fidelity data of the sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor; training at least one second fidelity network by adopting second fidelity data to obtain at least one trained second fidelity network; combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a plurality of datasets of a sample nuclear reactor; determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as first fidelity data; and taking the data except the first fidelity data in the plurality of data sets as second fidelity data.
In one embodiment, the computer program when executed by the processor further performs the steps of: grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade; acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of second fidelity networks is the same as the number of fidelity levels.
In one embodiment, the computer program when executed by the processor further performs the steps of: and respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
In one embodiment, the computer program when executed by the processor further performs the steps of: respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network; and taking the input end of the first fidelity network and the input end of each trained second fidelity network as the input end of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
In one embodiment, the computer program when executed by the processor further performs the steps of: the multi-fidelity network is used as a generator network, and a corresponding discriminator network is obtained according to the generator network; constructing a generation countermeasure network according to the generator network and the discriminator network; training the generated countermeasure network by adopting the first fidelity data to obtain the trained generated countermeasure network; and acquiring the trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
In one embodiment, the computer program when executed by the processor further performs the steps of: acquiring a control parameter and a first time state parameter of a target nuclear reactor; inputting the control parameters and the first moment state parameters into the trained multi-fidelity network to obtain second moment state parameters of the target nuclear reactor; and obtaining a simulation test result of the target nuclear reactor based on the state parameters at the second moment.
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 for analysis, stored data, displayed data, etc.) referred to in the present application are information and data authorized by the user or fully authorized by each party, and the collection, use and processing of the related data need to comply with the relevant laws and regulations and standards of the relevant countries and regions.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, database, or other medium used in the embodiments provided herein may include at least one of non-volatile and volatile memory. The nonvolatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical Memory, high-density embedded nonvolatile Memory, resistive Random Access Memory (ReRAM), magnetic Random Access Memory (MRAM), ferroelectric Random Access Memory (FRAM), phase Change Memory (PCM), graphene Memory, and the like. Volatile Memory can include Random Access Memory (RAM), external cache Memory, and the like. 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), among others. The databases referred to in various embodiments provided herein may include at least one of relational and non-relational databases. The non-relational database may include, but is not limited to, a block chain based distributed database, and the like. The processors referred to in the various embodiments provided herein may be, without limitation, general purpose processors, central processing units, graphics processors, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, or the like.
All possible combinations of the technical features in the above embodiments may not be described for the sake of brevity, but should be considered as being within the scope of the present disclosure as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the present application. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present application shall be subject to the appended claims.
Claims (10)
1. A method of constructing a multi-fidelity network for nuclear reactor simulation testing, the method comprising:
obtaining a first fidelity network from first fidelity data of a sample nuclear reactor, and obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor;
training at least one second fidelity network by adopting the second fidelity data to obtain at least one trained second fidelity network;
combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
2. The method of claim 1, wherein prior to obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor, further comprising:
acquiring a plurality of datasets of the sample nuclear reactor;
determining the accuracy degree of each data set, and taking the data in the data set with the highest accuracy degree as the first fidelity data;
and taking data in the plurality of data sets except the first fidelity data as the second fidelity data.
3. The method of claim 1, wherein the obtaining at least one second fidelity network from second fidelity data of the sample nuclear reactor comprises:
grading the second fidelity data to obtain at least one fidelity grade and subdata corresponding to each fidelity grade;
acquiring a corresponding second fidelity network according to the subdata corresponding to each fidelity grade; the number of the second fidelity networks is the same as the number of the fidelity levels;
the training of the at least one second fidelity network by using the second fidelity data to obtain the at least one trained second fidelity network comprises:
and respectively adopting the subdata corresponding to each fidelity grade to train each second fidelity network to obtain at least one trained second fidelity network.
4. The method of claim 1, wherein combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network comprises:
respectively connecting the output end of each trained second fidelity network to one input end of the first fidelity network;
and taking the input end of the first fidelity network and the input ends of the trained second fidelity networks as the input ends of the multi-fidelity network together, and taking the output end of the first fidelity network as the output end of the multi-fidelity network to obtain the multi-fidelity network.
5. The method of claim 1, wherein the training the multi-fidelity network with the first fidelity data to obtain a trained multi-fidelity network comprises:
taking the multi-fidelity network as a generator network, and acquiring a corresponding arbiter network according to the generator network;
constructing a generation countermeasure network according to the generator network and the discriminator network;
training the generated countermeasure network by adopting the first fidelity data to obtain a trained generated countermeasure network;
and acquiring a trained generator network from the trained generation countermeasure network to serve as the trained multi-fidelity network.
6. The method of claim 1, further comprising:
acquiring a control parameter and a first time state parameter of the target nuclear reactor;
inputting the control parameters and the first-moment state parameters into the trained multi-fidelity network to obtain second-moment state parameters of the target nuclear reactor;
and obtaining a simulation test result of the target nuclear reactor based on the second moment state parameter.
7. A multi-fidelity network construction apparatus for nuclear reactor simulation testing, the apparatus comprising:
an acquisition module to acquire a first fidelity network from first fidelity data of a sample nuclear reactor and to acquire at least one second fidelity network from second fidelity data of the sample nuclear reactor;
the training module is used for training at least one second fidelity network by adopting the second fidelity data to obtain at least one trained second fidelity network;
the combination module is used for combining at least one trained second fidelity network with the first fidelity network to obtain a multi-fidelity network, and training the multi-fidelity network by adopting the first fidelity data to obtain the trained multi-fidelity network; the trained multi-fidelity network is used for carrying out simulation test on the target nuclear reactor.
8. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor, when executing the computer program, implements the steps of the method of any of claims 1 to 6.
9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program realizes the steps of the method of any one of claims 1 to 6 when executed by a processor.
Priority Applications (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211447961.8A CN115859783A (en) | 2022-11-18 | 2022-11-18 | Multi-fidelity network construction method and device for nuclear reactor simulation test |
| PCT/CN2023/132550 WO2024104485A1 (en) | 2022-11-18 | 2023-11-20 | Multi-fidelity network construction method and apparatus for nuclear reactor simulation test |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202211447961.8A CN115859783A (en) | 2022-11-18 | 2022-11-18 | Multi-fidelity network construction method and device for nuclear reactor simulation test |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN115859783A true CN115859783A (en) | 2023-03-28 |
Family
ID=85664126
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202211447961.8A Pending CN115859783A (en) | 2022-11-18 | 2022-11-18 | Multi-fidelity network construction method and device for nuclear reactor simulation test |
Country Status (2)
| Country | Link |
|---|---|
| CN (1) | CN115859783A (en) |
| WO (1) | WO2024104485A1 (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024104485A1 (en) * | 2022-11-18 | 2024-05-23 | 中广核研究院有限公司 | Multi-fidelity network construction method and apparatus for nuclear reactor simulation test |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015037296A1 (en) * | 2013-09-11 | 2015-03-19 | 株式会社日立製作所 | System analysis device |
| CN111178535B (en) * | 2018-11-12 | 2024-05-07 | 第四范式（北京）技术有限公司 | Method and apparatus for implementing automatic machine learning |
| WO2020263358A1 (en) * | 2019-06-24 | 2020-12-30 | Nanyang Technological University | Machine learning techniques for estimating mechanical properties of materials |
| US11431300B2 (en) * | 2020-06-12 | 2022-08-30 | Nokia Technologies Oy | Machine learning based digital pre-distortion for power amplifiers |
| CN112182938A (en) * | 2020-10-13 | 2021-01-05 | 上海交通大学 | Mesoscopic structural part mechanical property prediction method based on transfer learning-multi-fidelity modeling |
| CN113886992B (en) * | 2021-10-21 | 2024-05-07 | 大连理工大学 | Digital twin modeling method based on multi-fidelity data |
| CN114676522B (en) * | 2022-03-28 | 2024-03-29 | 西安交通大学 | Pneumatic shape optimization design method, system and equipment integrating GAN and migration learning |
| CN115859783A (en) * | 2022-11-18 | 2023-03-28 | 中广核研究院有限公司 | Multi-fidelity network construction method and device for nuclear reactor simulation test |
-
2022
- 2022-11-18 CN CN202211447961.8A patent/CN115859783A/en active Pending
-
2023
- 2023-11-20 WO PCT/CN2023/132550 patent/WO2024104485A1/en unknown
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2024104485A1 (en) * | 2022-11-18 | 2024-05-23 | 中广核研究院有限公司 | Multi-fidelity network construction method and apparatus for nuclear reactor simulation test |
Also Published As
| Publication number | Publication date |
|---|---|
| WO2024104485A1 (en) | 2024-05-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP7562576B2 (en) | Rapid Digital Reactor Design Using Machine Learning | |
| Dini et al. | A new method for simultaneous calibration of demand pattern and Hazen-Williams coefficients in water distribution systems | |
| Xiao et al. | Structural reliability analysis using combined space partition technique and unscented transformation | |
| Weller et al. | An investigation of the pineapple express phenomenon via bivariate extreme value theory | |
| Fontana et al. | Model risk in credit risk | |
| CN116933192A (en) | Concrete dam operation key part partition monitoring method and model | |
| Shi et al. | A novel seepage safety monitoring model of CFRD with slab cracks using monitoring data | |
| CN111428419A (en) | Suspended sediment concentration prediction method and device, computer equipment and storage medium | |
| CN115859783A (en) | Multi-fidelity network construction method and device for nuclear reactor simulation test | |
| CN110852415B (en) | Vegetation index prediction method, system and equipment based on neural network algorithm | |
| Li et al. | 3-Dimensional general ADT modeling and analysis: Considering epistemic uncertainties in unit, time and stress dimension | |
| Hoseyni et al. | A Bayesian ensemble of sensitivity measures for severe accident modeling | |
| Li et al. | A lstm-based method for comprehension and evaluation of network security situation | |
| Borgonovo | Sensitivity analysis of model output with input constraints: A generalized rationale for local methods | |
| Bani-Mustafa et al. | A practical approach for evaluating the strength of knowledge supporting risk assessment models | |
| El-Morshedy et al. | A New Probability Heavy‐Tail Model for Stochastic Modeling under Engineering Data | |
| Inzunza-Aragón et al. | Use of artificial neural networks and response surface methodology for evaluating the reliability index of steel wind towers | |
| De Rocquigny | Structural reliability under monotony: Properties of FORM, simulation or response surface methods and a new class of Monotonous Reliability Methods (MRM) | |
| Goode et al. | Characterizing climate pathways using feature importance on echo state networks | |
| Šimková | L‐moment homogeneity test in trivariate regional frequency analysis of extreme precipitation events | |
| Brodeur et al. | Synthetic forecast ensembles for evaluating forecast informed reservoir operations | |
| Shipley | Inferential permutation tests for maximum entropy models in ecology | |
| CN114548602A (en) | Line loss prediction method, device, equipment and medium based on LSTM network | |
| Wang et al. | Verifying empirical predictive modeling of societal vulnerability to hazardous events: A Monte Carlo experimental approach | |
| Lin et al. | On the multi‐parameters identification of concrete dams: A novel stochastic inverse approach |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |