Method and apparatus for making nuclear database

Abstract

The present application provides a method and apparatus for creating a target core database. The method comprises: obtaining a data set for constructing a target core database and determining target evaluation core data to be updated in the data set; extracting multiple data features corresponding to the target evaluation core data from the data set; using a trained value prediction model based on the multiple data features to predict target predicted values corresponding to the target evaluation core data; updating the target evaluation core data in the data set based on the target predicted values; and constructing a target core database based on the updated data set. This allows for automatic creation of a core database and improves the accuracy and effectiveness of the data in the core database.

Description

Technical Field

[0001] The present application relates to the field of data processing, and in particular to a method and device for producing a nuclear database. Background Art

[0002] For nuclear reactor physics, nuclide evaluation data are a crucial foundation for neutron transport calculations in nuclear reactors. These data provide a comprehensive description of basic nuclide information, including nuclear mass, nucleon density, and half-life, as well as data on interactions between nuclides, such as reaction cross sections, angular distributions, energy distributions, and reaction yields for various reaction types. However, experimental measurements of these data often contain errors or omissions due to limitations in experimental setup, measurement errors, and limited technological advancements. However, nuclear databases used for reactor physics calculations are typically derived from nuclide evaluation data processed using nuclear data processing software. The accuracy of these databases directly impacts the results of reactor physics calculations. Summary of the Invention

[0003] This application provides a method and apparatus for creating a nuclear database to at least partially address one of the technical issues in the related art. The technical solution of this application is as follows:

[0004] According to one aspect of the present application, a method for preparing a nuclear database is provided, comprising:

[0005] Acquiring a data set for constructing a target nuclear database, and determining target evaluation nuclear data to be updated in the data set;

[0006] extracting a plurality of data features corresponding to the target evaluation core data from the data set;

[0007] According to the multiple data features, using a trained value prediction model, predicting a target prediction value corresponding to the target evaluation core data;

[0008] Based on the target prediction value, updating the target evaluation core data in the data set;

[0009] The target nuclear database is constructed according to the updated data set.

[0010] According to another aspect of the present application, a nuclear database preparation device is provided, the device comprising:

[0011] a processing module, configured to obtain a data set for constructing a target nuclear database and determine target evaluation nuclear data to be updated in the data set;

[0012] A first extraction module is used to extract a plurality of data features corresponding to the target evaluation core data from the data set;

[0013] A prediction module, configured to predict a target prediction value corresponding to the target evaluation core data using a trained value prediction model based on the multiple data features;

[0014] An updating module, configured to update the target evaluation core data in the data set based on the target prediction value;

[0015] A construction module is used to construct the target core database according to the updated data set.

[0016] According to another aspect of the present application, a non-transitory computer-readable storage medium of computer instructions is provided, wherein the computer instructions are used to enable the computer to execute the nuclear database production method proposed in the above aspect of the present application.

[0017] According to another aspect of the present application, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the nuclear database production method proposed in the above aspect of the present application.

[0018] The technical solutions provided by the embodiments of this application bring at least the following beneficial effects:

[0019] A data set for constructing a target core database is obtained, and target evaluation core data to be updated in the data set is determined; multiple data features corresponding to the target evaluation core data are extracted from the data set; based on the multiple data features, a trained value prediction model is used to predict target prediction values corresponding to the target evaluation core data; based on the target prediction values, the target evaluation core data in the data set is updated; and based on the updated data set, a target core database is constructed. Thus, the value prediction model is used to predict the values of the evaluation core data to be updated, and based on the prediction results, the evaluation core data to be updated is updated or optimized. Furthermore, a core database is created based on the optimized evaluation core data. This not only allows the desired core database to be automatically created, but also improves the data accuracy of the created core database.

[0020] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0022] Figure 1 A flowchart of the method for creating a nuclear database provided in Example 1 of the present application;

[0023] Figure 2A flowchart of the method for creating a nuclear database provided in Example 2 of the present application;

[0024] Figure 3 A schematic diagram of the process of preparing a nuclear database provided in this application;

[0025] Figure 4 This is a structural diagram of the nuclear database production device provided in Example 3 of this application. DETAILED DESCRIPTION

[0026] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0027] The following describes the nuclear database production method and device according to the embodiment of the present application with reference to the accompanying drawings.

[0028] Figure 1 This is a flow chart of the method for producing a nuclear database provided in Example 1 of the present application.

[0029] The embodiment of the present application uses the example of the core database production method being configured in a core database production device. The core database production device can be applied to any electronic device so that the electronic device can perform the core database production function.

[0030] Among them, the electronic device can be any device with computing capabilities, such as a personal computer, mobile terminal, server, etc. The mobile terminal can be, for example, a mobile phone, tablet computer, personal digital assistant, wearable device, etc., which are hardware devices with various operating systems, touch screens and / or display screens.

[0031] like Figure 1 As shown, the nuclear database preparation method may include the following steps:

[0032] Step 101: Acquire a data set for constructing a target nuclear database, and determine target evaluation nuclear data to be updated in the data set.

[0033] Among them, the data set may include multiple evaluation nuclear data, and the evaluation nuclear data may be but not limited to nuclear mass, nucleon density, half-life, nuclide atomic number, incident energy, cross-section value, incident angle, etc. This application does not impose any restrictions on this.

[0034] It should also be noted that the evaluation nuclear data can be, for example, ENDF / B (Evaluated Nuclear Data File-B), JEFF (The Joint Evaluated Fission and Fusion File), CENDL (China Evaluation Nuclear Data Library) and other evaluation nuclear databases, and this application does not impose any restrictions on this.

[0035] In an embodiment of the present application, a data set for constructing a target core database may be acquired, and target evaluation core data to be updated in the data set may be determined.

[0036] It is understandable that the evaluation core data in the dataset may have missing values or insufficient accuracy. Therefore, as a possible implementation method, the target evaluation core data to be updated in the dataset can be determined based on user operations, such as click operations, input operations, sliding operations, etc.

[0037] As another possible implementation, a labeling model can be used to label the target evaluation core data in the dataset that needs to be updated. For example, the labeling model can be used to identify missing values and / or low-precision values in the dataset and identify the labeled missing values and / or low-precision values as the target evaluation core data to be updated. This can automatically determine the target evaluation and data to be updated.

[0038] It should be noted that the target evaluation core data can be one or more, and this application does not impose any restrictions on this.

[0039] It should also be noted that, when there are multiple target evaluation nuclear data, the target evaluation nuclear data can be of the same type or of different types. For example, the target evaluation nuclear data are all total cross sections, or for another example, the target evaluation nuclear data include total cross sections and incident energy.

[0040] Step 102: extract multiple data features corresponding to target evaluation core data from the data set.

[0041] It is understandable that the target evaluation core data may have corresponding data features.

[0042] For example, assuming that the target evaluation nuclear data is the total cross section, the corresponding data features include nuclide atomic number, atomic mass, interpolation method, incident energy, cross section value, etc.

[0043] In an embodiment of the present application, data features corresponding to the target evaluation core data may exist in the data set, so that multiple data features corresponding to the target evaluation core data can be extracted from the data set.

[0044] Step 103 : Based on multiple data features, a trained value prediction model is used to predict the target prediction value corresponding to the target evaluation kernel data.

[0045] Among them, the value prediction model can be used to predict the value of the target evaluation nuclear data.

[0046] Optionally, the value prediction model is a random forest regression model.

[0047] In an embodiment of the present application, a trained value prediction model can be used to predict the value of the target evaluation core data based on multiple data features to obtain a target predicted value corresponding to the target evaluation core data.

[0048] As an example, multiple data features are input into a trained value prediction model, and the value output by the value prediction model is determined as the target prediction value corresponding to the target evaluation kernel data.

[0049] In the case where the value prediction model is a random forest regression model, in order to clearly illustrate how the trained value prediction model is used based on multiple data features to predict the target prediction value corresponding to the target evaluation core data, as a possible implementation method, first, for any decision tree in the random forest regression model, at least one target data feature corresponding to the decision tree can be determined from multiple data features. For example, a data feature can be randomly selected from multiple data features as the target data feature corresponding to the decision number; secondly, the decision tree is used to predict the value of the target evaluation core data based on at least one target data feature corresponding to the decision tree to obtain the first prediction value corresponding to the decision tree; finally, according to the first prediction value corresponding to each decision tree, the target prediction value corresponding to the target evaluation core data is determined. For example, the weighted sum of each first prediction value can be performed to obtain the target prediction value.

[0050] Optionally, before predicting the target prediction value corresponding to the target evaluation core data using a trained value prediction model based on data characteristics, multiple sample evaluation core data can be extracted from multiple first evaluation core databases; according to the evaluation core database to which each sample evaluation core data belongs, a test set and a training set are determined from the multiple sample evaluation core data; and the training set and the test set are used to train and test the random forest regression model, respectively.

[0051] For example, multiple sample evaluation core data are extracted from evaluation core databases such as ENDF / B, JEFF, and CENDL; the sample evaluation core data belonging to the ENDF / B evaluation core database among the multiple sample evaluation core data are determined as the test set, and the sample evaluation core data except the test set among the multiple sample evaluation core data are determined as the training set; thus, the training set and test set are used to train and test the random forest regression model respectively.

[0052] In this way, the trained value prediction model can be obtained.

[0053] Step 104: Based on the target prediction value, the target evaluation kernel data in the data set is updated.

[0054] As an example, when the target evaluation data is a missing value, the target prediction value is added to the corresponding field of the target evaluation core data in the dataset; when the target evaluation core data is a low-precision value, the value under the corresponding field of the target evaluation core data in the dataset is updated to the target prediction value.

[0055] For example, suppose the target evaluation kernel data and its corresponding data features are shown in Table 1:

[0056] Table 1 Target evaluation core data and their corresponding data characteristics

[0057]

[0058] Among them, the target evaluation core data is a missing value, and the corresponding data feature 1 value is xxxx, the data feature 2 value is xxx, and the data feature 3 value is xxxxx. According to the values of the above data features 1, 2, and 3, the trained value prediction model is used to predict the target prediction value corresponding to the target evaluation core data; the target prediction value is added to the corresponding field of the target evaluation core data in the dataset.

[0059] Step 105: construct a target core database based on the updated data set.

[0060] Optionally, the target program can be executed to read the target instruction file, and the data in the updated data set can be processed based on the instructions of the target instruction file to obtain a processed target data set, and then, a target core database can be generated based on the target data set.

[0061] The target program is a pre-configured program, such as an NJOY (Nuclear Data Processing System) program, an AMPX (Automated Multi-Purpose Cross-section Generation System) program, etc., and this application does not impose any restrictions on this.

[0062] The processing may include format conversion, data cleaning, etc., which is not limited in this application.

[0063] The nuclear database production method of the embodiment of the present application obtains a data set for constructing a target nuclear database and determines the target evaluation nuclear data to be updated in the data set; extracts multiple data features corresponding to the target evaluation nuclear data from the data set; uses a trained value prediction model to predict target prediction values corresponding to the target evaluation nuclear data based on the multiple data features; updates the target evaluation nuclear data in the data set based on the target prediction values; and constructs the target nuclear database based on the updated data set. Thus, the value prediction model is used to predict the values of the evaluation nuclear data to be updated, and the evaluation nuclear data to be updated is updated or optimized based on the prediction results. Furthermore, a nuclear database is produced based on the optimized evaluation nuclear data. On the one hand, the required nuclear database can be automatically produced, and on the other hand, the data accuracy of the produced nuclear database can be improved.

[0064] It is understandable that after the target core database is constructed, the target core database can be inspected or tested to determine whether the target core database meets the production requirements. Figure 2 , the above process is explained in detail.

[0065] Figure 2 This is a flow chart of the method for creating a nuclear database provided in Example 2 of the present application.

[0066] like Figure 2 As shown, based on the previous embodiment of the present application, the method for preparing a nuclear database may further include the following steps:

[0067] In step 201 , for any reference benchmark question among a plurality of reference benchmark questions, a target core database is used to calculate the reference benchmark question to obtain a calculation result of the reference benchmark question, and a relative error of the reference benchmark question is determined based on the calculation result of the reference benchmark question and the experimental results.

[0068] Among them, the reference benchmark question can be, for example, a critical benchmark question, a fuel consumption benchmark question, etc., and this application does not impose any restrictions on this.

[0069] It should be noted that the reference benchmark question may have a corresponding experimental result, and the experimental result may be a numerical value obtained based on a large number of experiments.

[0070] In an embodiment of the present application, for any reference benchmark question among multiple reference benchmark questions, the target core database, that is, the relevant evaluation core data in the target core database, can be used to calculate the reference benchmark question, and the calculation result of the reference benchmark question can be obtained. Furthermore, the relative error of the reference benchmark question can be determined based on the calculation result and experimental results of the reference benchmark question.

[0071] As an example, assuming that the calculation result of the reference benchmark question T is a and its corresponding experimental result is b, the relative error w of the reference benchmark question can be determined according to the following formula:

[0072]

[0073] Step 202: Classify the calculation results of each reference benchmark question according to the relative error of each reference benchmark question to obtain at least one category.

[0074] Among them, the relative errors corresponding to the calculation results of the reference benchmark questions included in each classified category belong to the corresponding set value space.

[0075] Among them, the setting value space can be, for example, [0, 0.1‰), [0.1‰, 1‰), [1‰, 1], etc., and this application does not impose any restrictions on this.

[0076] For example, assume that the categories include category A and category B, where the relative error corresponding to the calculation results of the reference benchmark questions included in category A belongs to the corresponding set value space [0, 0.0001), and the relative error corresponding to the calculation results of the reference benchmark questions included in category B belongs to the corresponding set value space [0.0001, 0.001).

[0077] In an embodiment of the present application, the calculation results of each reference benchmark question can be classified according to the relative error of each reference benchmark question, thereby obtaining at least one category.

[0078] It can be understood that in order to improve the data accuracy of the target core database, a large number of reference benchmark questions can be used to detect the target core database. At this time, in order to quickly and accurately classify the calculation results of each reference benchmark question, as a possible implementation method, a logistic regression algorithm can be used to classify the calculation results of each reference benchmark question based on the relative error of each reference benchmark question, thereby obtaining at least one category.

[0079] Therefore, the use of the logistic regression algorithm can quickly and accurately classify the calculation results of each reference benchmark question. On the one hand, it can avoid the waste of human resources in the manual classification process, and on the other hand, it can avoid the situation where human factors lead to misclassification.

[0080] Step 203: Determine whether the target core database meets the production requirements based on at least one category.

[0081] In an embodiment of the present application, it is possible to determine whether the target core database meets the production requirements based on at least one obtained category, wherein the production requirements are, for example, the computational accuracy requirements of the database.

[0082] As a possible implementation manner, when only the first target category exists in at least one category, it is determined that the target core database meets the production requirement.

[0083] In an embodiment of the present application, the first target category can be pre-set. For example, the first target category is a category in which the relative error corresponding to the calculation result belongs to [0, 0.0001), or the first target category is a category in which the relative error corresponding to the calculation result belongs to [0, 0.0002). The present application does not impose any restrictions on the set value space corresponding to the first target category.

[0084] In a case where the target core database is obtained by executing the target program to process the data in the updated data set based on the read target instruction file, as another possible implementation manner, when the second target category exists in at least one category and the third target category does not exist, the target library parameters in the target instruction file are adjusted for no more than a set number of rounds;

[0085] Any round of adjustment process may include the following steps:

[0086] 1. For any round of adjustment process, adjust the target library parameters in the target instruction file;

[0087] 2. By executing the target program, based on the target instruction file using the target library parameters adjusted in this round, the data in the updated data set is processed to obtain the target core database of this round;

[0088] 3. For any reference benchmark question, use the current round target core database to calculate the reference benchmark question and obtain the current round calculation result of the reference benchmark question;

[0089] 4. Determine the relative error between the current round of calculation results and the corresponding experimental results of the reference benchmark problem;

[0090] 5. Classify the calculation results of each reference benchmark problem based on the relative error of the current round to obtain at least one category of the current round adjustment process;

[0091] 6. In response to only the first target category existing in at least one category in the current round of adjustment process, stopping iteration;

[0092] 7. In response to the second target category existing in at least one category of the current round of adjustment process and the third target category not existing, executing the next round of adjustment process.

[0093] In an embodiment of the present application, the second target category can be pre-set. For example, the second target category is a category in which the relative error corresponding to the calculation result belongs to [0.0001, 0.001). Alternatively, the second target category is a category in which the relative error corresponding to the calculation result belongs to [0.0002, 0.001). The present application does not impose any restrictions on the set value space corresponding to the second target category.

[0094] In the embodiment of the present application, the present application does not limit the setting value space corresponding to the third target category.

[0095] It should be noted that there is no intersection between the setting value spaces corresponding to the first target category, the second target category, and the third target category.

[0096] Among them, the target library parameters may include parameters lambda, sig0, etc., and this application does not impose any restrictions on this.

[0097] The set number of times may be pre-set, such as 10, 15, etc., and this application does not impose any restrictions on this.

[0098] It should be noted that, after a set number of rounds of adjustment of the target library parameters in the target instruction file, the calculation results of the reference benchmark question may be classified, and there may still be a second target category and / or a third target category in at least one category obtained, that is, there is a second target category and / or a third target category in at least one category of the last round of adjustment process. At this time, in a possible implementation method of an embodiment of the present application, when there is a second target category and / or a third target category in at least one category of the last round of adjustment process, the value prediction model can be retrained.

[0099] As another possible implementation, when a third target category exists in at least one category, the value prediction model may be directly retrained.

[0100] In this way, it is possible to effectively determine whether the target nuclear database meets the production requirements based on the obtained categories.

[0101] The core database preparation method of the embodiment of the present application uses a target core database to calculate any reference benchmark question from a plurality of reference benchmark questions, obtains a calculation result for the reference benchmark question, and determines the relative error of the reference benchmark question based on the calculation result and experimental results of the reference benchmark question. The calculation results of each reference benchmark question are classified according to the relative error of each reference benchmark question to obtain at least one category. The relative errors corresponding to the calculation results of the reference benchmark questions included in each classified category belong to a corresponding set value space. Based on the at least one category, it is determined whether the target core database meets the preparation requirements. In this way, the target core database can be tested, that is, whether the target core database meets the preparation requirements can be determined.

[0102] In order to clearly illustrate the nuclear database preparation method of the present application, a detailed description is given below with reference to examples.

[0103] As an example, the database preparation method of this application is as follows Figure 3 The following steps are shown:

[0104] Step 1: Obtain sample nuclide evaluation and nuclear data (referred to as sample evaluation and data in this application);

[0105] The sample nuclide evaluation nuclear data include but are not limited to the nuclide evaluation nuclear data in evaluation nuclear databases such as ENDF / B, JENDL, JEFF, CENDL and BROND.

[0106] Step 2: Divide the obtained sample nuclide evaluation nuclear data into a test set and a training set;

[0107] For example, the sample nuclide evaluation nuclear data belonging to the ENDF / B evaluation nuclear database can be used as a test set, and the remaining sample nuclide evaluation nuclear data can be used as a training set.

[0108] Step 3: Use the training set to train the random forest regression model;

[0109] In this model, the basic unit is the decision tree, which is composed of nodes. At each node of each tree, a data feature can be randomly selected; during the training process, the node's mean squared error can be used to evaluate the importance of the candidate feature. Specifically, for any node, the mean squared error before the node split is first calculated, and then the mean squared error after the node split is calculated. Then, the importance of each node (i.e., feature) is determined based on the difference between the mean squared error before the node split and the mean squared error after the node split.

[0110] It should be noted that the importance of a feature can be used to measure the contribution of the feature to the model prediction results.

[0111] It's also important to note that during training, each decision tree can be trained independently, learning different samples and features. In the random forest regression algorithm, each decision tree predicts a value, and each tree makes its own prediction. The average of the predictions from each tree is taken as the final prediction. The performance of the random forest regression model is then evaluated by calculating the sample mean squared error (MSE). When the sample mean squared error (MSE) is less than a set threshold, the random forest regression model is considered trained.

[0112] Step 4: Obtain a data set for preparing a target nuclear database and determine the target evaluation nuclear data to be updated in the data set;

[0113] The data set includes multiple nuclide evaluation nuclear data and data features corresponding to the nuclide evaluation nuclear data;

[0114] Step 5: Extract multiple data features corresponding to the target evaluation core data from the data set;

[0115] Step 6: Input multiple data features into the trained value prediction model to obtain the target prediction value corresponding to the target evaluation kernel data;

[0116] Step 7: Based on the target prediction value, update the target evaluation kernel data in the data set;

[0117] Among them, when the target evaluation core data is a missing value in the data set, the target prediction value is added to the corresponding field of the target evaluation core data; when the target evaluation core data is a low-precision value in the data set, the value under the corresponding field of the target evaluation core data in the data set is updated to the target prediction value;

[0118] Step 8: Create a target nuclear database based on the updated dataset;

[0119] As an example, a setting program (referred to as a target program in this application) is automatically executed. The function of this program is to automatically write input files (referred to as target instruction files in this application) for nuclear data processing software including but not limited to NJOY, AMPX, etc., and to produce target nuclear databases in formats including but not limited to ACE (A Compact ENDF), WIMS (WinfrithImproved Multigroup Scheme), etc. based on the updated data set.

[0120] Step 9: Using the target core database, calculate any international benchmark question from among the multiple international benchmark questions to obtain a corresponding calculation result;

[0121] Step 10: Determine the relative error of the calculation results of any international benchmark problem based on the calculation results and experimental results of the international benchmark problem;

[0122] Step 11: Based on the relative errors of the calculation results of each international benchmark question, a logistic regression model is used to classify the calculation results of each international benchmark question to obtain at least one category;

[0123] The relative errors corresponding to the calculation results of the international benchmark questions included in each classified category belong to the corresponding set value space;

[0124] Among them, a sufficient number of sample international benchmark questions can be obtained, and the sample international benchmark questions can be divided according to a set ratio to obtain a training set and a test set; the training set and the test set are used to train and test the logistic regression model respectively.

[0125] Step 12: When only the first target category exists in at least one category, determining whether the target kernel database meets the calculation accuracy requirement;

[0126] When a second target category exists in at least one category and a third target category does not exist, the target library parameters in the input file are adjusted for no more than a set number of rounds; wherein, in any round of adjustment, the target library parameters are modified, the target core database is remade, and at least one category of the current adjustment process is obtained; when only the first target category exists in at least one category of the current adjustment process, the iteration is stopped; when a second target category exists in at least one category of the current adjustment process and a third target category does not exist, the next round of adjustment is executed; when a second target category and / or a third target category exists in at least one category of the last round of adjustment, the value prediction model is retrained until the remade core database meets the computational accuracy requirements;

[0127] When a third target category exists in at least one category, the random forest regression model is retrained until the remade kernel database meets the computational accuracy requirements.

[0128] Among them, the value space of the relative error corresponding to the first target category is [0, 0.0001); the value space of the relative error corresponding to the second target category is [0.0001, 0.001); the value space of the relative error corresponding to the third target category is [0.001, 1].

[0129] Since the nuclear evaluation database stores a wide variety of nuclear data, and the research directions of different countries on nuclear data are not exactly the same, there are differences in the nuclide evaluation nuclear data published by different countries. By adopting the database preparation method of the present application, it is possible to extract the cross-sectional features of multiple nuclide evaluation nuclear data and use artificial intelligence machine learning to quickly and effectively predict the missing data of the nuclide evaluation nuclear data in experimental measurements, and to effectively correct the low-precision data in the nuclide evaluation nuclear data.

[0130] The nuclear database preparation method of the present application has at least the following advantages:

[0131] 1. The random forest regression model can effectively supplement the missing data in the nuclear data of nuclide evaluation and correct the data errors of the nuclear data of nuclide evaluation. The nuclear database data produced by the improved nuclear data of nuclide evaluation is more accurate.

[0132] 2. The logistic regression model is used to test the produced nuclear database, which reduces labor costs, avoids errors caused by human factors, and makes the nuclear database testing and verification work more efficient and accurate.

[0133] 3. Closely combining the random forest regression model with the logistic regression model can make the process of nuclear database production, testing and verification more efficient and the calculation results more accurate.

[0134] 4. Closely integrating artificial intelligence machine learning with nuclear database production and verification has achieved interdisciplinary integration of computer science and nuclear physics.

[0135] In summary, the nuclear database production method of the present application utilizes the random forest regression model after iterative training to accurately predict missing data and data with large errors in nuclear data for nuclide evaluation, and utilizes the trained logistic regression model to realize the testing and verification process of the nuclear database, thereby producing the nuclear database in a more efficient and accurate manner. Finally, the nuclear database production method of the present application improves the accuracy of the nuclear data in the nuclear database, and can promote nuclear science research and nuclear engineering applications in a data-driven manner.

[0136] With the above Figures 1 to 2 Corresponding to the method for making a nuclear database provided in the embodiment, the present application also provides a nuclear database making device. Figures 1 to 2 The embodiment provides a method for producing a nuclear database, and therefore the implementation method of the method for producing a nuclear database is also applicable to the nuclear database producing device provided in the embodiment of the present application, and will not be described in detail in the embodiment of the present application.

[0137] Figure 4 This is a structural diagram of the nuclear database production device provided in Example 3 of this application.

[0138] like Figure 4 As shown, the nuclear database preparation device 400 may include: an acquisition module 401 , a first extraction module 402 , a prediction module 403 , an update module 404 and a construction module 405 .

[0139] The acquisition module 401 is used to acquire a data set for constructing a target core database and determine target evaluation core data to be updated in the data set.

[0140] The first extraction module 402 is used to extract multiple data features corresponding to the target evaluation core data from the data set.

[0141] The prediction module 403 is used to predict the target prediction value corresponding to the target evaluation kernel data based on multiple data features using a trained value prediction model.

[0142] The updating module 404 is used to update the target evaluation kernel data in the data set based on the target prediction value.

[0143] The construction module 405 is used to construct a target core database according to the updated data set.

[0144] In a possible implementation of the embodiment of the present application, the nuclear database preparation device 400 may further include:

[0145] The calculation module is used to calculate any reference benchmark question among multiple reference benchmark questions using a target core database to obtain a calculation result of the reference benchmark question.

[0146] A first determining module is used to determine the relative error of the reference benchmark problem based on the calculation result and experimental result of the reference benchmark problem;

[0147] The classification module is used to classify the calculation results of each reference benchmark question according to the relative error of each reference benchmark question to obtain at least one category; wherein the relative error corresponding to the calculation results of the reference benchmark questions included in each classified category belongs to the corresponding set value space.

[0148] The second determination module is configured to determine whether the target nuclear database meets the production requirements based on at least one category.

[0149] In a possible implementation of the embodiment of the present application, the second determination module is configured to: in response to only the first target category existing in at least one category, determine that the target core database meets the production requirements.

[0150] In a possible implementation of the embodiment of the present application, the target core database is obtained by executing the target program to process the data in the updated data set based on the read target instruction file, and the second determination module is further used to: in response to the existence of the second target category and the absence of the third target category in at least one category, perform an adjustment process on the target library parameters in the target instruction file for no more than a set number of rounds; wherein, for any round of adjustment process, the target library parameters in the target instruction file are adjusted; by executing the target program to process the data in the updated data set based on the target instruction file using the target library parameters after this round of adjustment Processing to obtain the target core database of this round; for any reference benchmark question, use the target core database of this round to calculate the reference benchmark question to obtain the calculation result of this round of the reference benchmark question; determine the relative error of this round between the calculation result of this round of the reference benchmark question and the corresponding experimental result; classify the calculation result of this round of each reference benchmark question according to the relative error of this round of each reference benchmark question to obtain at least one category of this round of adjustment process; in response to the existence of only the first target category in at least one category of this round of adjustment process, stop iteration; in response to the existence of the second target category and the absence of the third target category in at least one category of this round of adjustment process, execute the next round of adjustment process.

[0151] In a possible implementation of an embodiment of the present application, the second determination module is further used to: retrain the value prediction model in response to the presence of a second target category and / or a third target category in at least one category of the last round of adjustment process.

[0152] In a possible implementation of the embodiment of the present application, the second determination module is further configured to: in response to the presence of a third target category in at least one category, retrain the value prediction model.

[0153] In a possible implementation of an embodiment of the present application, the value prediction model is a random forest regression model; the prediction module 403 is used to: for any decision tree in the random forest regression model, determine at least one target data feature corresponding to the decision tree from multiple data features; use the decision tree to predict the value of the target evaluation core data based on the at least one target data feature corresponding to the decision tree, and obtain a first prediction value corresponding to the decision tree; determine the target prediction value corresponding to the target evaluation core data based on the first prediction value corresponding to each decision tree.

[0154] In a possible implementation of an embodiment of the present application, the update module 404 is used to: in response to the target evaluation core data being a missing value, add the target prediction value to the corresponding field of the target evaluation core data; in response to the target evaluation core data being a low-precision value, update the value under the corresponding field of the target evaluation core data in the data set to the target prediction value.

[0155] In a possible implementation of the embodiment of the present application, the nuclear database preparation device 400 may further include:

[0156] A second extraction module is used to extract a plurality of sample evaluation core data from a plurality of first evaluation core databases;

[0157] A third determining module is used to determine a test set and a training set from the plurality of sample evaluation core data according to the evaluation core database to which each sample evaluation core data belongs;

[0158] The processing module is used to train and test the random forest regression model using the training set and the training set respectively.

[0159] The nuclear database production device of the embodiment of the present application obtains a data set for constructing a target nuclear database and determines the target evaluation nuclear data to be updated in the data set; extracts multiple data features corresponding to the target evaluation nuclear data from the data set; uses a trained value prediction model to predict target prediction values corresponding to the target evaluation nuclear data based on the multiple data features; updates the target evaluation nuclear data in the data set based on the target prediction values; and constructs a target nuclear database based on the updated data set. Thus, the value prediction model is used to predict the values of the evaluation nuclear data to be updated, and the evaluation nuclear data to be updated is updated or optimized based on the prediction results. Furthermore, a nuclear database is produced based on the optimized evaluation nuclear data. On the one hand, the required nuclear database can be automatically produced, and on the other hand, the data accuracy of the produced nuclear database can be improved.

[0160] In order to implement the above embodiments, the present application also proposes an electronic device, wherein the electronic device can be the server or detection device in the aforementioned embodiments; it includes: a memory, a processor, and a computer program stored on the memory and runnable on the processor, and when the processor executes the program, it implements the nuclear database production method proposed in any of the aforementioned embodiments of the present application.

[0161] In order to implement the above embodiments, the present application also proposes a non-temporary computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, it implements the nuclear database production method proposed in any of the above embodiments of the present application.

[0162] In order to implement the above embodiments, the present application also proposes a computer program product. When the instructions in the computer program product are executed by a processor, the nuclear database production method proposed in any of the above embodiments of the present application is executed.

[0163] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0164] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of the technical features being referred to. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of such features. Throughout the description of this application, "plurality" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0165] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application belong.

[0166] The logic and / or steps represented in the flowcharts or otherwise described herein, for example, can be considered as a sequenced list of executable instructions for implementing the logical functions, and can be embodied in any computer-readable medium for use by, or in conjunction with, an instruction execution system, apparatus, or device (e.g., a computer-based system, a system including a processor, or other system that can fetch and execute instructions from an instruction execution system, apparatus, or device). For purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by, or in conjunction with, an instruction execution system, apparatus, or device. More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection with one or more wires (electronic devices), a portable computer disk cartridge (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and programmable read-only memory (EPROM or flash memory), fiber optic devices, and a portable compact disc read-only memory (CDROM). Furthermore, the computer-readable medium may even be paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium and then editing, interpreting or processing it in another suitable manner if necessary, and then storing it in a computer memory.

[0167] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0168] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0169] In addition, the functional units in the various embodiments of the present application may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into a module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0170] The storage medium mentioned above may be a read-only memory, a magnetic disk, or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present application. Persons skilled in the art may make changes, modifications, substitutions, and variations to the above embodiments within the scope of the present application.

Claims

1. A method for preparing a nuclear database, characterized in that: The method comprises: Acquiring a data set for constructing a target nuclear database, and determining target evaluation nuclear data to be updated in the data set; extracting a plurality of data features corresponding to the target evaluation core data from the data set; According to the multiple data features, using a trained value prediction model, predicting a target prediction value corresponding to the target evaluation core data; Based on the target prediction value, updating the target evaluation core data in the data set; constructing the target nuclear database according to the updated data set; Also includes: For any one of the plurality of reference benchmark questions, using the target kernel database, performing calculations on the reference benchmark question to obtain a calculation result of the reference benchmark question, and determining a relative error of the reference benchmark question based on the calculation result of the reference benchmark question and an experimental result; Classifying the calculation results of each reference benchmark problem according to the relative error of each reference benchmark problem to obtain at least one category; wherein the relative errors corresponding to the calculation results of the reference benchmark problems included in each of the classified categories belong to a corresponding set value space; Based on the at least one category, it is determined whether the target core database meets the production requirements.

2. The method according to claim 1, characterized in that Determining whether the target core database meets the production requirements according to the at least one category includes: In response to only the first target category existing in the at least one category, it is determined that the target core database meets the production requirements.

3. The method according to claim 1, characterized in that The target core database is obtained by executing a target program to process data in the updated data set based on the read target instruction file. The method further includes: In response to the presence of a second target category and the absence of a third target category in the at least one category, performing an adjustment process on target library parameters in the target instruction file for no more than a set number of rounds; Among them, for any round of the adjustment process, the target library parameters in the target instruction file are adjusted; by executing the target program, the data in the updated data set are processed based on the target instruction file using the target library parameters adjusted in this round to obtain the target core database of this round; for any reference benchmark question, the target core database of this round is used to calculate the reference benchmark question to obtain the calculation result of this round of the reference benchmark question; the relative error of this round between the calculation result of this round of the reference benchmark question and the corresponding experimental result is determined; according to the relative error of this round of each reference benchmark question, the calculation result of this round of each reference benchmark question is classified to obtain at least one category of this round of adjustment process; in response to the existence of only the first target category in at least one category of this round of adjustment process, the iteration is stopped; in response to the existence of the second target category and the absence of the third target category in at least one category of this round of adjustment process, the next round of the adjustment process is executed.

4. The method according to claim 3, characterized in that The method further comprises: In response to the presence of a second target category and / or a third target category in at least one category of the last round of the adjustment process, the value prediction model is retrained.

5. The method according to claim 1, wherein The method further comprises: In response to the presence of a third target category in the at least one category, the value prediction model is retrained.

6. The method according to claim 1, characterized in that The value prediction model is a random forest regression model; The method of using a trained value prediction model based on the multiple data features to predict a target prediction value corresponding to the target evaluation core data includes: For any decision tree in the random forest regression model, determining at least one target data feature corresponding to the decision tree from the multiple data features; Using the decision tree to predict the value of the target evaluation core data based on at least one target data feature corresponding to the decision tree, to obtain a first predicted value corresponding to the decision tree; According to the first prediction value corresponding to each of the decision trees, the target prediction value corresponding to the target evaluation core data is determined.

7. The method according to claim 1, characterized in that The updating of the target evaluation core data in the data set based on the target prediction value corresponding to the target evaluation core data includes: In response to the target evaluation core data being a missing value, adding the target prediction value to a corresponding field of the target evaluation core data; In response to the target evaluation core data being a low-precision value, the value under the field corresponding to the target evaluation core data in the data set is updated to the target predicted value.

8. The method according to any one of claims 1 to 7, characterized in that: Before predicting the target prediction value corresponding to the target evaluation core data using the trained value prediction model according to the data characteristics, the method further includes: extracting a plurality of sample evaluation core data from a plurality of first evaluation core databases; determining a test set and a training set from the plurality of sample evaluation core data according to the evaluation core database to which each sample evaluation core data belongs; The random forest regression model is trained and tested using the training set and the training set, respectively.

9. A nuclear database production device, characterized in that: The device comprises: a processing module, configured to obtain a data set for constructing a target nuclear database and determine target evaluation nuclear data to be updated in the data set; A first extraction module is used to extract a plurality of data features corresponding to the target evaluation core data from the data set; A prediction module, configured to predict a target prediction value corresponding to the target evaluation core data using a trained value prediction model based on the multiple data features; An updating module, configured to update the target evaluation core data in the data set based on the target prediction value; A construction module, configured to construct the target nuclear database according to the updated data set; wherein, for any of the plurality of reference benchmark questions, the target kernel database is used to perform calculations on the reference benchmark question to obtain a calculation result of the reference benchmark question, and a relative error of the reference benchmark question is determined based on the calculation result of the reference benchmark question and the experimental result; Classifying the calculation results of each reference benchmark problem according to the relative error of each reference benchmark problem to obtain at least one category; wherein the relative errors corresponding to the calculation results of the reference benchmark problems included in each of the classified categories belong to a corresponding set value space; Based on the at least one category, it is determined whether the target core database meets the production requirements.

Images