Method and system for constructing and using nuclear structure material database and medium

Abstract

The invention discloses a nuclear structure material database construction and use method and system and a medium, and the method comprises the steps: obtaining original test data of a nuclear structure material, the original test data being in-pile and out-pile performance test data of the nuclear structure material; according to the original test data, first calculation information and second calculation information are obtained through calculation; the first calculation information is material calculation information for core structure material elements, and the second calculation information is material calculation information for describing core structure material components; performing data cleaning on the original test data, the first calculation information and the second calculation information, and filling the cleaned corresponding data into a database in a preset format one by one based on a Python program to form a nuclear structure material performance database; and extracting corresponding data from the database based on database use requirements of the set performance model. According to the method, construction and use of the kernel structure material database meeting the requirement of a high-accuracy kernel structure material machine learning model are realized.

Description

Technical Field

[0001] This invention relates to the field of materials database construction, and specifically to a method, system, and medium for constructing and using a nuclear structure materials database. Background Technology

[0002] Nuclear structural materials refer to a class of materials used in the service environment of nuclear facilities to play a structural role, such as nuclear fuel cladding materials and reactor internals materials. The most significant difference between nuclear structural materials and other materials lies in their specific requirements for irradiation characteristics. Traditional methods for developing high-performance nuclear structural materials mainly rely on traditional trial-and-error screening methods. This development model is severely constrained by economic costs and the scarcity of in-reactor testing resources, resulting in high development difficulty, limited testing resources, long development times, and high costs. Thanks to the rapid development of computer technology, a data-driven paradigm based on materials genome engineering and big data concepts has provided new methods and ideas for the development of novel nuclear structural materials, beyond traditional trial-and-error screening methods. Machine learning methods are one of the main approaches.

[0003] Building material performance models based on machine learning typically involves data collection and cleaning, feature recognition engineering, model selection and training, as well as analysis, verification, and re-optimization. This approach has already been applied in the design of various advanced materials. However, machine learning methods are less commonly used in nuclear structural materials due to several factors. First, irradiation performance data is limited due to scarce experimental resources and high costs. Many nuclear structural materials only have off-core data available, which cannot be directly used for design and selection. Therefore, the engineering application value of machine learning models based on off-core data is limited. Second, while mechanical properties, corrosion properties, and other on-core and off-core properties are correlated, their underlying mechanisms are often unclear. Directly applying machine learning to on-core and off-core properties makes it difficult to build highly accurate models. It requires integrating more types of physicochemical data and performing mathematical processing to supplement more dimensional parameter information. Therefore, conventional material databases are insufficient to meet the needs of machine learning for nuclear structural materials.

[0004] In view of the above, this application is hereby submitted. Summary of the Invention

[0005] The technical problem to be solved by this invention is that traditional material databases have narrow applicability, are difficult to meet the needs of subsequent machine learning models for nuclear structural materials, and are difficult to integrate different types of data. The purpose of this invention is to provide a method, system and medium for constructing and using a nuclear structural material database, which integrates multiple types of experimental data and computational information to achieve the construction and use of a nuclear structural material database that meets the needs of high-accuracy machine learning models for nuclear structural materials.

[0006] This invention is achieved through the following technical solution:

[0007] In a first aspect, the present invention provides a method for constructing and using a nuclear structure material database, the method comprising:

[0008] The original experimental data of nuclear structural materials is obtained. The original experimental data is the test data of the in-core and out-of-core performance of nuclear structural materials. The original experimental data includes basic experimental information, experimental characteristic information and data source.

[0009] Based on the original experimental data, first calculation information and second calculation information were obtained; the first calculation information is material calculation information for the elements of the nuclear structure material, and the second calculation information is material calculation information describing the composition of the components of the nuclear structure material.

[0010] The original experimental data, the first calculation information, and the second calculation information are cleaned, and the cleaned data are filled into the database with a preset format one by one based on the Python program to form a database of nuclear structure material properties.

[0011] Based on the database usage requirements of the defined performance model, relevant data are extracted from the nuclear structural material performance database.

[0012] Furthermore, the basic experimental information includes basic material characteristic information, material processing information, material property information, experimental characteristic markers, and experimental environment information. Specifically: Basic material characteristic information includes at least the material grade, constituent elements, and material application classification; Material processing information includes at least the processing methods (e.g., smelting, forging, hot rolling, cold rolling, welding, mechanical alloying, etc.) and their main process parameters; Material property information includes at least phase composition, grain size, grain size, roughness, and texture, and obtains density, thermal conductivity, resistivity, specific heat capacity, coefficient of linear expansion, phase transition temperature, melting point, and the corresponding physicochemical information obtained in step 2 (first calculation information and second calculation information) through experiments or calculations. Experimental characteristic markers indicate the performance category to which this data belongs, such as corrosion performance, creep performance, etc., and the test type, such as off-pile test, ion irradiation test, neutron irradiation test, etc. Experimental environment information includes at least the test temperature, pressure, irradiation dose, and irradiation damage. The above basic experimental information should be as complete as possible. Except for the basic material characteristics, the other characteristics are allowed to be omitted. A set of general default values ​​should be provided for specific types of materials in case they are needed.

[0013] The test characteristic information includes test feature information and performance feature information; taking the corrosion performance of zirconium alloy for nuclear fuel cladding as an example, the test feature information should at least include key test parameters such as corrosion medium, corrosion time, content of dissolved elements (such as oxygen), and pH value; the performance feature information should at least include: uniform corrosion rate, stress corrosion rate, nodular corrosion rate, oxide film thickness, corrosion weight gain, corrosion hydrogen absorption, hydride orientation, crack propagation rate, hydrogen concentration, and corrosion kinetic parameters.

[0014] Data sources include source category, document name, author, publication year, and link, to facilitate subsequent tracing.

[0015] Furthermore, based on the original experimental data and considering the fundamental physicochemical properties of the nuclear structure material itself, the first and second calculation information were obtained, including:

[0016] Based on the original experimental data, first-principles calculation tools and / or large-scale molecular dynamics software were used to calculate the first calculation information for common elements of nuclear fuel cladding materials. The first calculation information includes binding energy, lattice distortion, electronic localization function, electronic density of states, elastic constant, bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant.

[0017] Based on the first calculation information, and using the single-element characteristic parameters combined with the content of each element in the original experimental data, the second calculation information is obtained. The second calculation information includes mixing enthalpy, atomic radius difference, radius, Ω parameter, valence electron concentration, shear modulus difference, λ entropy, average binding energy, electronegativity mismatch, average shear modulus, shear modulus strengthening parameter, shear modulus mismatch, average atomic mass, radius mismatch, and electronegativity difference, etc.

[0018] Preferably, based on the original experimental data, the first calculation information and the second calculation information are calculated and obtained, including:

[0019] Step 21: Based on the original experimental data, construct a "cross-scale-multi-dimensional" element-level characteristic calculation system for common elements in nuclear fuel cladding materials, breaking through the limitations of traditional single-dimensional calculations. This system includes at least one or more types of information obtained through first-principles calculation tools, large-scale molecular dynamics software, or other material calculation methods, as well as literature review. These include element-level characteristics (mass number, electronegativity, etc.), structural characteristics (lattice parameters, interatomic spacing, atomic distribution, coordination number, degree of doped lattice distortion, diffusion coefficient, migration barrier, binding energy, etc.), and electronic characteristics (density of states, band gap, etc.). This first calculation information includes binding energy, lattice distortion, electronic localization function, electronic density of states, elastic constant, bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant.

[0020] Specifically, the parameter settings of the first-principles calculation tool are as follows: For 17 commonly used elements in nuclear structure materials (Fe, Zr, Sn, Nb, etc.), a 4×4×4 supercell model is used for VASP calculations, with the cutoff energy set to 520 eV and the K-point density set to 0.2. This parameter combination is the optimal solution after multiple trial calculations and verifications, which ensures both calculation accuracy and control of calculation costs, and is sufficient to support the needs of machine learning for nuclear materials.

[0021] Specifically, the computational parameters for the large-scale molecular dynamics software were set as follows: The universal machine learning potential function MACE (covering the entire periodic table, and compared with NEP89, mattersim, GPTFF, alignn, and MACE, MACE was found to be the best performing universal potential function) based on first-principles datasets was adopted, avoiding errors caused by using different potential functions for different alloying elements during calculations. Then, LAMMPS was used for large-scale molecular dynamics simulations, with a uniform energy convergence accuracy of 1e⁻⁸ and a force convergence accuracy of 1e⁻⁶.

[0022] Specifically, the nuclear materials were selectively screened for characteristic parameters: the 18 elemental-level features selected are all directly related to the key service performance of nuclear structure materials. For example, the lattice distortion parameter was specifically designed for atomic displacement effects under irradiation, the migration barrier directly reflects the material's ability to repair irradiation damage, and the elastic anisotropy constant is related to the material's stress corrosion cracking susceptibility.

[0023] Step 22: Based on the first calculation information, and using the single-element characteristic parameters combined with the content of each element in the original experimental data, calculate the second calculation information. The second calculation information includes mixing enthalpy, atomic radius difference, radius, Ω parameter, valence electron concentration, shear modulus difference, λ entropy, average binding energy, electronegativity mismatch, average shear modulus, shear modulus strengthening parameter, shear modulus mismatch, average atomic mass, radius mismatch, and electronegativity difference, etc.

[0024] Specifically, mathematical processing methods are used to obtain computational information on core structure materials oriented towards component composition. Based on the component alloy composition, basic physicochemical property data corresponding to each component entry are formed, realizing the quantitative transformation from single-element characteristics to the overall alloy performance. The basic property data should include one or more of the following: data trend (mean, median, maximum, minimum, difference, etc.) and dispersion (standard deviation, distribution, etc.), expanding the characteristic parameter dimensions of the component data points.

[0025] For example, a typical reflection of the average properties of component elements ,difference Dispersion characteristics The characteristic mathematical processing methods can be:

[0026]

[0027]

[0028]

[0029] In the formula: Indicates the proportion of elements. This represents the numerical value of the element under this feature, where i is the element's statistical index and n is the number of group members. For the average properties of the components, [ ] indicates the maximum value. [ ] indicates the minimum value. Due to differences in the properties of component elements, This refers to the dispersion characteristics of the component elements.

[0030] Descriptors can be constructed from literature or based on characteristics to obtain computational information on nuclear structural materials based on composition from elemental information. We focus on the needs of nuclear materials and have constructed several descriptors: a shear modulus strengthening parameter η, which quantifies the synergistic strengthening effect of alloy composition on shear modulus, taking into account deformation mechanisms such as dislocation slip and twinning. This is more accurate than traditional simple average descriptions and is suitable for the complex service environment within reactors. Another parameter is the radius γ, which better quantifies the structural mismatch between elements, correlating radius differences to characteristics such as internal stress and the density of zirconium alloy oxide films.

[0031] The above mathematical processing methods include fitting using Excel formulas and batch fitting using Python or other tools. Specifically, this invention develops an integrated script that includes feature calculation, outlier detection, and result verification, supporting multi-format data input (CSV, Excel, TXT) and batch processing. The script embeds a physical constraint verification module; for example, when the calculated enthalpy of mixing ΔH > 0, it is automatically marked as "thermodynamically unstable".

[0032] Examples are given below:

[0033] First, for the common constituent elements of nuclear fuel cladding materials Zr, Sn, Nb, Fe, Cr, Ni, Cu, Sb, Sc, Ge, Al, O, S, C, H, N, and Si, the first calculation information, including binding energy, lattice distortion, electronic localization function, electronic density of states, and elastic constant C, was calculated using the first-principles calculation tool VASP. ij The bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant were calculated. Binding energy, lattice distortion, electronic localization function, and electronic density of states were calculated using the first-principles calculation tool VASP and the corresponding PBE pseudopotentials for each element. The PBE pseudopotentials for each element were used as inputs to VASP, and the results were obtained by applying ±5% strain to the unit cell. Furthermore, the bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant were obtained based on the Voigt-Reuss-Hill approximation model. Secondly, the migration barriers and diffusion coefficients of vacancies and interstitial atoms were calculated using the large-scale molecular dynamics software LAMMPS, with the universal potential function being preferred.

[0034] Then, based on the single-element characteristic parameters and the content of each element in the dataset, mathematical processing methods are used to obtain the calculated information of the nuclear structure material oriented towards the component composition. By consulting literature, the radius, atomic mass, number of valence electrons, and electronegativity of each component element are obtained. Furthermore, the data trends (mean, median, maximum, minimum, difference, etc.) and dispersion (standard deviation, distribution, etc.) of the characteristics are calculated and analyzed. Based on the above calculated and queried information, the mathematical processing method of the nuclear structure material calculated information based on the component composition is used to perform batch fitting using Python. In this invention, the mixing enthalpy, atomic radius difference, radius γ, Ω parameter, valence electron concentration, shear modulus difference, λ entropy, average binding energy, electronegativity mismatch, average shear modulus, shear modulus strengthening parameter, shear modulus mismatch, average atomic mass, radius mismatch, electronegativity difference, etc., can be calculated using the following formula based on the first calculated information in S21 above to obtain the second calculated information.

[0035] Furthermore, data cleaning is performed on the original experimental data, the first calculation information, and the second calculation information, including:

[0036] The dataset formed by the original experimental data, the first calculation information, and the second calculation information is reviewed, and a dual deduplication strategy based on key fields and performance characteristics is adopted to remove all duplicate entries to obtain the first dataset.

[0037] Eliminate different expressions of the same meaning in the first dataset, and unify the different units of the same parameter into the same unit to obtain the second dataset;

[0038] The second dataset is filled with default values ​​to obtain the third dataset. The default value filling includes: for nuclear structure materials with grades, their composition and typical physical properties are filled with the typical parameters of the grade; for those tested according to test standards, their default test information is filled with the standard specification parameters and marked.

[0039] Based on a pre-defined data structure, the text data in the third dataset is digitally encoded using a One-Hot method to obtain the fourth dataset.

[0040] All entries in the fourth dataset are reviewed. If they meet the requirements for characterizing and constructing the performance characteristic model of the structural material, they are marked as valid entries; otherwise, they are marked as reference entries.

[0041] Furthermore, a dual deduplication strategy based on key fields and performance characteristics includes:

[0042] If the keyword fields in two data entries are the same and the performance characteristic difference meets the preset difference value, then the two entries are considered duplicates.

[0043] Furthermore, based on a pre-defined data structure, the text data in the third dataset is digitally encoded using a One-Hot method, including:

[0044] For heat treatment: Based on the preset data structure [heat treatment process name, temperature, time], each heat treatment process is treated as a sub-data item, and the heat treatment process name is numerically encoded using a label;

[0045] For phase composition: Based on the preset data structure [phase name, crystal structure, volume fraction, size], each phase is treated as a sub-data item, and the phase name is numerically encoded using a label.

[0046] Furthermore, the nuclear structure material performance database is stored in the form of an Excel file and a MySQL database, respectively.

[0047] Furthermore, after extracting relevant data from the nuclear structure material property database, the method also includes:

[0048] The extracted data is then normalized. Normalization includes:

[0049] For feature data with known numerical ranges and clear upper and lower limits, the maximum-minimum normalization method is used to normalize them.

[0050] For data with a good distribution and no need to change the distribution characteristics, the MAX data normalization method is used for normalization.

[0051] Secondly, the present invention provides a system for constructing and using a nuclear structure material database, the system comprising:

[0052] The acquisition unit is used to acquire raw experimental data of nuclear structural materials. The raw experimental data consists of test data on the in-core and out-of-core performance of nuclear structural materials. The raw experimental data includes basic experimental information, experimental characteristic information, and data sources.

[0053] The calculation unit is used to calculate and obtain first calculation information and second calculation information based on the original experimental data; the first calculation information is material calculation information for the elements of the nuclear structure material, and the second calculation information is material calculation information describing the composition of the components of the nuclear structure material.

[0054] The cleaning unit is used to clean the raw experimental data, the first calculation information, and the second calculation information.

[0055] The database forming unit is used to fill the cleaned data into a database with a preset format one by one based on a Python program, forming a database of nuclear structure material properties;

[0056] The data extraction unit is used to extract relevant data from the nuclear structural material performance database based on the database usage requirements of the set performance model.

[0057] Furthermore, the execution process of the cleaning unit is as follows:

[0058] The dataset formed by the original experimental data, the first calculation information, and the second calculation information is reviewed, and a dual deduplication strategy based on key fields and performance characteristics is adopted to remove all duplicate entries to obtain the first dataset.

[0059] Eliminate different expressions of the same meaning in the first dataset, and unify the different units of the same parameter into the same unit to obtain the second dataset;

[0060] The second dataset is filled with default values ​​to obtain the third dataset. The default value filling includes: for nuclear structure materials with grades, their composition and typical physical properties are filled with the typical parameters of the grade; for those tested according to test standards, their default test information is filled with the standard specification parameters and marked.

[0061] Based on a pre-defined data structure, the text data in the third dataset is digitally encoded using a One-Hot method to obtain the fourth dataset.

[0062] All entries in the fourth dataset are reviewed. If they meet the requirements for characterizing and constructing the performance characteristic model of the structural material, they are marked as valid entries; otherwise, they are marked as reference entries.

[0063] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the above-described method for constructing and using a nuclear structure material database.

[0064] Fourthly, the present invention also provides a computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the above-described method for constructing and using a nuclear structure material database.

[0065] Fifthly, the present invention also provides a computer program product, including a computer program / instruction that, when executed by a processor, implements the steps of the above-described method for constructing and using a nuclear structure material database.

[0066] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0067] This invention discloses a method, system, and medium for constructing and using a nuclear structural materials database. Addressing the limitations of traditional materials databases—such as narrow applicability, difficulty in meeting the modeling needs of machine learning models for nuclear structural materials, and difficulty in integrating different types of data—this method integrates various types of physicochemical data, including material experiments and calculations. It performs multi-type data fusion, mathematical processing, data cleaning, classification, and normalization to achieve the construction and use of a nuclear structural materials database that meets the requirements of high-accuracy machine learning models. This invention integrates different types of data, providing effective support for the construction of standardized databases for nuclear structural materials and subsequent model building, thus accelerating the research and development of nuclear structural materials. Attached Figure Description

[0068] The accompanying drawings, which are included to provide a further understanding of embodiments of the invention and form part of this application, do not constitute a limitation thereof. In the drawings:

[0069] Figure 1 This is a flowchart of a method for constructing and using a nuclear structure material database according to the present invention;

[0070] Figure 2 This is a flowchart of the data cleaning process of the present invention;

[0071] Figure 3 This is a schematic diagram of the binding energy, lattice distortion, and elastic modulus calculated in an embodiment of the present invention;

[0072] Figure 4 This is an example of how the present invention uses Python to process erroneous data in a dataset;

[0073] Figure 5 This is an example of how Python is used to fill in missing data in a dataset according to the present invention;

[0074] Figure 6 This is an example of how Python is used to process text features in this invention;

[0075] Figure 7 This is a block diagram of a system for constructing and using a nuclear structure material database according to the present invention. Detailed Implementation

[0076] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0077] This invention constructs a complete and comprehensive set of methods for building and using a machine learning database that integrates multiple types of information for nuclear structural materials, in order to achieve the construction and use of a nuclear structural material database that meets the needs of high-accuracy machine learning models for nuclear structural materials.

[0078] Example 1

[0079] like Figure 1 As shown, this invention provides a method for constructing and using a nuclear structural material database. This invention is applicable to nuclear structural material systems such as nuclear fuel cladding materials and in-core structural materials.

[0080] In this embodiment, the nuclear fuel cladding material is a zirconium alloy, and the key properties collected and focused on are corrosion resistance and creep resistance. The specific steps of the method include:

[0081] Step 1: Collect raw test data of nuclear structural materials from publicly available literature, test reports, and other sources. The raw test data includes test data on the in-pile and out-of-pile corrosion performance and creep performance of nuclear structural materials (e.g., nuclear fuel cladding system materials). The raw test data includes basic test information, test characteristic information, and data sources.

[0082] In this embodiment, the basic material characteristics information includes the material grade or name, main alloying elements of the matrix, coating (composition and thickness), phase composition, grain size, average grain size, and degree of recrystallization. The material processing information includes surface quality performance indicators (roughness, surface defects), microstructure control indicators (key impurity content, grain size, texture, microstructure), preparation process performance indicators (melting process, forging process, hot rolling process, cold rolling process), and reprocessing performance indicators (welding process, machining process, remanufacturing process). The material physical property information includes density, thermal conductivity, resistivity, specific heat capacity, coefficient of linear expansion, phase transition temperature, and melting temperature. The experimental environment information includes sample preparation method, thermal neutron cross-section, irradiation dose, ion irradiation intensity, experimental temperature, pH value, vacuum degree, sample processing procedures, chemical polishing (acid pickling), heat treatment process, N content, O content, H content, pressure, sample shape, sample size, sample thickness, and surface area. The experimental characteristic markings include corrosion performance and creep performance.

[0083] The test characteristic information includes test feature information and performance feature information. Corrosion test feature information includes corrosive medium, corrosion time, dissolved element content (oxygen, etc.), and pH value. Performance feature information includes uniform corrosion rate, stress corrosion rate, knot corrosion rate, oxide film thickness, corrosion weight gain, corrosion hydrogen absorption, hydride orientation, crack propagation rate, hydrogen concentration, and corrosion kinetic parameters. Creep test feature information includes creep test method, strain rate, biaxial circumferential stress, uniaxial stress, nanoindentation pressure, and creep time. Performance feature information includes yield strength, tensile strength, circumferential creep strain, axial creep strain, nanoindentation strain, creep rate, minimum creep rate, fracture time, fracture strain, creep activation energy, and stress index.

[0084] The data source information includes the source category, document name, author, publication year, and link.

[0085] Step 2: Based on the original experimental data and the material physicochemical properties of the zirconium alloy itself, calculate the relevant material calculation information using the composition of the zirconium alloy, including the first calculation information and the second calculation information;

[0086] In this embodiment, step 2 specifically includes:

[0087] Step 21: Calculate the first calculation information

[0088] For the elemental composition of Zr, Sn, Nb, Fe, Cr, Ni, Cu, Sb, Sc, Ge, Al, O, S, C, H, N, and Si in the dataset (i.e., the original experimental data), firstly, the binding energy, lattice distortion, electronic localization function, electronic density of states, and elastic constant C were calculated using the first-principles calculation tool VASP. ij The bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant were calculated. Binding energy, lattice distortion, electronic localization function, and electronic density of states were all calculated using VASP in a 4×4×4 supercell. A PEB pseudopotential was used, with a cutoff energy of 520 eV and a K-point density of 0.2. The elastic constant was obtained by applying ±5% strain to the unit cell, and the bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant were further obtained based on the Voigt-Reuss-Hill approximation model. Secondly, the migration barriers and diffusion coefficients of vacancies and interstitial atoms were calculated using the large-scale molecular dynamics software LAMMPS. A universal potential function was preferentially used. The migration barrier calculation included a configuration with 250 atoms, and the diffusion coefficient calculation included a configuration with 10,000 atoms. The elastic modulus was calculated using the elastic constant as follows:

[0089]

[0090]

[0091]

[0092]

[0093] In the formula, C11, C12, and C44 are elastic constants, G is the shear modulus, B is the volume modulus, and E is Young's modulus. It is Poisson's ratio.

[0094] In this embodiment, the alloy element binding energy, lattice distortion, bulk modulus, shear modulus, Young's modulus, and Poisson's ratio data obtained by calculating using the above parameter settings are as follows: Figure 3 As shown, Figure 3 The horizontal axis represents the element type, and the vertical axis represents the binding energy, RMS displacement, and modulus (including shear modulus, bulk modulus, and Young's modulus). The binding energies of alloying elements such as Sn and Nb are relatively low (≤0 eV), indicating a greater tendency to form solid solutions and a lower precipitation tendency (which reduces the performance of zirconium alloys), consistent with experimental measurements. The lattice distortion is also significantly lower than other alloying elements (0-0.02), indicating lower internal stress after alloying, which is consistent with Sn and Nb being the main additive elements in Zr alloys in practice. This provides reliable element-level data support for subsequent compositional characteristic calculations.

[0095] S22: Calculate the second calculation information

[0096] Based on the first calculation information (single-element feature parameters) obtained in S21, and combined with the content of each element in the dataset, mathematical processing methods are used to obtain calculation information of nuclear structure materials oriented towards component composition.

[0097] By consulting literature, the radius, atomic mass, number of valence electrons, and electronegativity of each constituent element were obtained. Furthermore, the data trends (mean, median, maximum, minimum, variance, etc.) and dispersion (standard deviation, distribution, etc.) of the characteristic data were calculated and analyzed.

[0098] Taking ZrSnNbFe as an example, based on the composition and elemental characteristics information in Table 1 (constructed as a built-in database), the average binding energy of -0.0013 eV / atom, the atomic radius difference of 1.598 Å, and the shear modulus strengthening parameter of 33.1 can be quickly calculated.

[0099] Table 1. Composition and Elemental Characteristics

[0100]

[0101] S23: A mathematical processing method for calculating information on nuclear structure materials based on composition, utilizing Python for batch fitting. Based solely on composition and some elemental data calculated in S21, various physicochemical characteristics of the alloy can be quickly obtained. In this invention, the following parameters are calculated: enthalpy of mixing, atomic radius difference, radius γ, Ω parameter, valence electron concentration, shear modulus difference, λ entropy, average binding energy, electronegativity mismatch, average shear modulus, shear modulus strengthening parameter, shear modulus mismatch, average atomic mass, radius mismatch, and electronegativity difference. These can be calculated using the following formulas based on the cross-scale calculation results of the material in S21, and the specific solution method is shown in Table 2:

[0102] Table 2. List of Second Calculation Information Formulas for Nuclear Structure Materials Oriented to Component Composition

[0103]

[0104] In Table 2, C i C j Indicates the number of group members. For the enthalpy of mixture of components, r i r is the component radius, and r is the average radius. max r is the maximum radius of the component. min T is the minimum radius of the component. m Melting point For mixed entropy, VEC i G represents the number of valence electrons in a component. i G is the component shear modulus, and E is the mean shear modulus. ci For component binding energy, For the component to be electronegative, M i It represents atomic mass.

[0105] Step 3: Clean the original experimental data, the first calculation information and the second calculation information, and fill the cleaned data into the database with a preset format one by one based on the Python program to form a database of nuclear structure material properties;

[0106] In this embodiment, as Figure 2 As shown, step 3 involves data cleaning of the original experimental data, the first calculation information, and the second calculation information, including:

[0107] Step 31: Review the dataset formed by the original experimental data, the first calculation information, and the second calculation information, and use a dual deduplication strategy based on key fields and performance characteristics to remove all duplicate entries to obtain the first dataset;

[0108] In this embodiment, a dual deduplication strategy based on key fields and performance characteristics is employed, including: if the keyword fields in two data entries are identical and the difference in performance characteristics meets a preset difference, then the two entries are determined to be duplicates. The key fields include four core elements: material grade, main components, process parameters, and test environment. Performance characteristics include key indicators such as corrosion rate and creep strain.

[0109] For example, for two corrosion performance data points for Zr-4 alloy, if the key fields are completely identical and the difference in corrosion rate is ≤5%, they are considered duplicate data and removed. This strategy can effectively avoid the misjudgment problem of "different batches of the same material with similar performance" in nuclear material data.

[0110] Step 32: Eliminate different expressions of the same meaning in the first dataset, and unify the different unit expressions of the same parameter into the same unit to obtain the second dataset;

[0111] For example, for the data in the "Material Grade or Name" column of the first dataset, the grade name is unified for the same grade. For example, M5Framatome, M5Framatomealloy, and M5® are all represented as M5 Zirconium Alloy, Zircaloy-2 and Zr-2 are represented as Zircaloy-2, and Zircaloy-4 and Zr-4 are represented as Zircaloy-4.

[0112] For parameters expressed in different units, use the same unit, except for units specific to nuclear materials (e.g., irradiation dose n / m). 2 s、n / m 2 Corrosion rate (unit: mg / dm²·d, mm / a) Develop conversion rules, for example, unify the irradiation dose to n / m 2 s, unifying the units for various corrosion rates to mg / dm 2 For example, ppm and wt.% are unified to wt.%, day / h / year is unified to day, and ℃ / K is unified to ℃. Incorrect data is corrected, such as RT being 25℃, data like 573-593 being represented as averages, and data types like <10 being changed to maximum values. Figure 4 This is an example of using Python to process erroneous data in a dataset.

[0113] Step 33: Fill the second dataset with default values ​​to obtain the third dataset;

[0114] For example, the default value filling includes: for nuclear structure materials with grades, their composition and typical physical properties are filled with the typical parameters of the grade; for tests conducted using reference test standards, their default test information is filled with standard specification parameters and specially marked; and the content of unmentioned impurity elements is marked as 0, the missing experimental temperature is set to 360℃, and the missing corrosion pressure is set to 10.8MPa, etc. Figure 5 This is an example of using Python to fill in missing data in a dataset.

[0115] Step 34: Based on the preset data structure, the text data in the third dataset is digitally encoded using the One-Hot method to obtain the fourth dataset;

[0116] In this embodiment, step 34 includes:

[0117] For heat treatment: convert it to a consistent data format: [[Heat Treatment Process 1 Name, Temperature, Time], [Heat Treatment Process 2 Name, Temperature, Time]]. Each heat treatment process is treated as a sub-data entry, including its heat treatment temperature and time. If the process is rolling, the heat treatment time is modified to the corresponding sample thickness after rolling. The heat treatment processes involved in the dataset include annealing, hot rolling, and cold rolling, represented by numbers 1-3 respectively.

[0118] For phase composition, the data format is: [[phase name 1, crystal structure, volume fraction, size], [phase name 2, crystal structure, volume fraction, size]]. Each phase is treated as a sub-data entry, including phase name, crystal structure, volume fraction, and phase size. Finally, the resulting two-dimensional array is flattened into a one-dimensional NumPy array. Figure 6 This is an example of using Python to process text features.

[0119] Step 35: Review all entries in the fourth dataset. If they meet the requirements for characterizing and constructing the performance characteristic model of the structural material, they are marked as valid entries; otherwise, they are marked as reference entries.

[0120] As a further implementation, step 3 also includes:

[0121] The entire data cleaning process for S35 is automated using Python scripts. For example, fuzzy matching rules are used for nuclear material grades (supporting intelligent recognition of Chinese and English abbreviations and the correspondence between old and new grades, such as using string similarity algorithms to uniformly label "Zr-2 alloy", "Zircaloy-2", and "zirconium-2" as the standard grade "Zircaloy-2"); logical verification rules are used for special test data such as irradiation and corrosion, such as automatically determining "irradiation dose <10"... 18 n / m²s and corrosion rate >100 mg / dm 2 "This is abnormal data."

[0122] In this embodiment, an example of the database style for nuclear structure material properties is shown in Table 3.

[0123] Table 3. Nuclear Structure Material Performance Database

[0124]

[0125] It should be noted that Table 2 is for illustrative purposes only. The various data parameters can be expanded according to actual needs to form a dataset suitable for specific nuclear materials. The database can be stored using MySQL, Excel, or an appropriate format.

[0126] Step 4: Based on the database usage requirements of the specific performance model, extract the corresponding data from the nuclear structure material performance database.

[0127] In this embodiment, the nuclear structure material performance database includes not only corrosion performance, but also performance data such as circumferential tensile performance and creep performance. Here, data tagged with "corrosion performance" is extracted.

[0128] As a further implementation, after extracting relevant data from the nuclear structure material performance database, to eliminate the excessive influence of dimensional differences between features on the model training process, necessary normalization processing is performed on some feature data. Therefore, the method also includes:

[0129] The extracted data is then normalized. Normalization includes:

[0130] (1) For feature data with known numerical ranges and relatively clear upper and lower limits, the maximum-minimum normalization method is used for normalization; the formula for the maximum-minimum normalization method is:

[0131]

[0132] in, and These are the parameter values ​​before and after normalization, respectively. and These represent the maximum and minimum parameter values, respectively.

[0133] (2) For irradiance, temperature, and stress characteristics whose numerical distributions are already relatively good and do not require alteration, the MAX data normalization method should be used for normalization. The formula for the MAX data normalization method is:

[0134]

[0135] in, and These are the parameter values ​​before and after normalization, respectively. This represents the maximum parameter value.

[0136] Example 2

[0137] like Figure 7 As shown, the difference between this embodiment and Embodiment 1 is that this embodiment provides a system for constructing and using a nuclear structure material database, which corresponds one-to-one with the method for constructing and using a nuclear structure material database in Embodiment 1; the system includes:

[0138] The acquisition unit is used to acquire raw experimental data of nuclear structural materials. The raw experimental data consists of test data on the in-core and out-of-core performance of nuclear structural materials. The raw experimental data includes basic experimental information, experimental characteristic information, and data sources.

[0139] The calculation unit is used to calculate and obtain first calculation information and second calculation information based on the original experimental data; the first calculation information is material calculation information for the elements of the nuclear structure material, and the second calculation information is material calculation information describing the composition of the components of the nuclear structure material.

[0140] The cleaning unit is used to clean the raw experimental data, the first calculation information, and the second calculation information.

[0141] The database forming unit is used to fill the cleaned data into a database with a preset format one by one based on a Python program, forming a database of nuclear structure material properties;

[0142] The data extraction unit is used to extract relevant data from the nuclear structural material performance database based on the database usage requirements of the set performance model.

[0143] As a further implementation, the cleaning unit's execution process is as follows:

[0144] The dataset formed by the original experimental data, the first calculation information, and the second calculation information is reviewed, and a dual deduplication strategy based on key fields and performance characteristics is adopted to remove all duplicate entries to obtain the first dataset.

[0145] Eliminate different expressions of the same meaning in the first dataset, and unify the different units of the same parameter into the same unit to obtain the second dataset;

[0146] The second dataset is filled with default values ​​to obtain the third dataset. The default value filling includes: for nuclear structure materials with grades, their composition and typical physical properties are filled with the typical parameters of the grade; for those tested according to test standards, their default test information is filled with the standard specification parameters and marked.

[0147] Based on a pre-defined data structure, the text data in the third dataset is digitally encoded using a One-Hot method to obtain the fourth dataset.

[0148] All entries in the fourth dataset are reviewed. If they meet the requirements for characterizing and constructing the performance characteristic model of the structural material, they are marked as valid entries; otherwise, they are marked as reference entries.

[0149] The execution process of each unit can be carried out according to the steps of the method for constructing and using a nuclear structure material database in Example 1, and will not be described in detail in this example.

[0150] Meanwhile, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that the processor executes the computer program to implement the above-mentioned method for constructing and using a nuclear structure material database.

[0151] Meanwhile, the present invention also provides a computer-readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the above-described method for constructing and using a nuclear structure material database.

[0152] Meanwhile, the present invention also provides a computer program product, including a computer program / instruction, which, when executed by a processor, implements the steps of the above-described method for constructing and using a nuclear structure material database.

[0153] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0154] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0155] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0156] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0157] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for constructing and using a nuclear structure material database, characterized in that, The method includes: Acquire raw experimental data of nuclear structural materials, wherein the raw experimental data are in-core and out-of-core performance test data of nuclear structural materials; the raw experimental data includes basic experimental information, experimental characteristic information, and data source. Based on the original experimental data, first calculation information and second calculation information are calculated and obtained; the first calculation information is material calculation information for the elements of the nuclear structure material, and the second calculation information is material calculation information describing the composition of the components of the nuclear structure material. The original experimental data, the first calculation information, and the second calculation information are cleaned, and the cleaned data are filled into a database with a preset format one by one based on a Python program to form a nuclear structure material performance database. Based on the database usage requirements of the defined performance model, relevant data are extracted from the nuclear structure material performance database.

2. The method for constructing and using a nuclear structure material database according to claim 1, characterized in that, The basic experimental information includes basic material characteristics, material processing information, material physical properties, experimental characteristic markings, and experimental environment information; The test characteristic information includes test feature information and performance feature information; The data sources include source category, document name, author, and year of publication.

3. The method for constructing and using a nuclear structure material database according to claim 1, characterized in that, Based on the original experimental data, first calculation information and second calculation information are calculated and obtained, including: Based on the original experimental data, first calculation information was obtained for common elements of nuclear fuel cladding materials using first-principles calculation tools and / or large-scale molecular dynamics software; the first calculation information includes binding energy, lattice distortion, electronic localization function, electronic density of states, elastic constant, bulk modulus, shear modulus, Young's modulus, Poisson's ratio, and elastic anisotropy constant. Based on the first calculation information, and using the single-element characteristic parameters and the content of each element in the original experimental data, the second calculation information is obtained. The second calculation information includes mixing enthalpy, atomic radius difference, radius, Ω parameter, valence electron concentration, shear modulus difference, λ entropy, average binding energy, electronegativity mismatch, average shear modulus, shear modulus strengthening parameter, shear modulus mismatch, average atomic mass, radius mismatch, and electronegativity difference.

4. The method for constructing and using a nuclear structure material database according to claim 1, characterized in that, Data cleaning is performed on the original experimental data, the first calculation information, and the second calculation information, including: The dataset formed by the original experimental data, the first calculation information, and the second calculation information is reviewed, and a dual deduplication strategy based on key fields and performance characteristics is adopted to remove all duplicate entries to obtain the first dataset. Eliminate different expressions of the same meaning in the first dataset, and unify the different unit expressions of the same parameter into the same unit to obtain the second dataset; The second dataset is filled with default values ​​to obtain the third dataset; the default value filling includes: for nuclear structure materials with grades, their composition and typical physical properties are filled with the typical parameters of the grade; for those tested according to test standards, their default test information is filled with the standard specification parameters and marked. Based on a preset data structure, the text data in the third dataset is digitally encoded using a One-Hot method to obtain the fourth dataset; All entries in the fourth dataset are reviewed. If they meet the requirements for characterizing and constructing the performance characteristic model of the structural material, they are marked as valid entries; otherwise, they are marked as reference entries.

5. The method for constructing and using a nuclear structure material database according to claim 4, characterized in that, The dual deduplication strategy based on key fields and performance characteristics includes: If the keyword fields in two data entries are the same and the performance characteristic difference meets the preset difference value, then the two entries are considered duplicates.

6. The method for constructing and using a nuclear structure material database according to claim 4, characterized in that, Based on a preset data structure, the text data in the third dataset is digitally encoded using a One-Hot method, including: For heat treatment: Based on the preset data structure [heat treatment process name, temperature, time], each heat treatment process is treated as a sub-data item, and the heat treatment process name is numerically encoded using a label; For phase composition: Based on the preset data structure [phase name, crystal structure, volume fraction, size], each phase is treated as a sub-data item, and the phase name is numerically encoded using a label.

7. The method for constructing and using a nuclear structure material database according to claim 1, characterized in that, The nuclear structure material performance database is stored in both Excel files and MySQL databases.

8. The method for constructing and using a nuclear structure material database according to claim 1, characterized in that, After extracting the relevant data from the nuclear structure material performance database, the method further includes: The extracted data is then normalized, the normalization process including: For feature data with known numerical ranges and clear upper and lower limits, the maximum-minimum normalization method is used to normalize them. For data with a good distribution and no need to change the distribution characteristics, the MAX data normalization method is used for normalization.

9. A system for constructing and using a nuclear structure material database, characterized in that, The system includes: The acquisition unit is used to acquire raw experimental data of nuclear structural materials, which are in-core and out-of-core performance test data of nuclear structural materials; the raw experimental data includes basic experimental information, experimental characteristic information, and data source. The calculation unit is used to calculate and obtain first calculation information and second calculation information based on the original experimental data; the first calculation information is material calculation information for the elements of the nuclear structure material, and the second calculation information is material calculation information describing the composition of the components of the nuclear structure material. The cleaning unit is used to clean the original experimental data, the first calculation information, and the second calculation information. The database forming unit is used to fill the cleaned data into a database with a preset format one by one based on a Python program, forming a database of nuclear structure material properties; The data extraction unit is used to extract relevant data from the nuclear structure material performance database based on the database usage requirements of the set performance model.

10. A system for constructing and using a nuclear structure material database according to claim 9, characterized in that, The cleaning unit performs the following process: The dataset formed by the original experimental data, the first calculation information, and the second calculation information is reviewed, and a dual deduplication strategy based on key fields and performance characteristics is adopted to remove all duplicate entries to obtain the first dataset. Eliminate different expressions of the same meaning in the first dataset, and unify the different unit expressions of the same parameter into the same unit to obtain the second dataset; The second dataset is filled with default values ​​to obtain the third dataset; the default value filling includes: for nuclear structure materials with grades, their composition and typical physical properties are filled with the typical parameters of the grade; for those tested according to test standards, their default test information is filled with the standard specification parameters and marked. Based on a preset data structure, the text data in the third dataset is digitally encoded using a One-Hot method to obtain the fourth dataset; All entries in the fourth dataset are reviewed. If they meet the requirements for characterizing and constructing the performance characteristic model of the structural material, they are marked as valid entries; otherwise, they are marked as reference entries.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements a method for constructing and using a nuclear structure material database as described in any one of claims 1 to 8.

12. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, it implements a method for constructing and using a nuclear structure material database as described in any one of claims 1 to 8.

13. A computer program product comprising a computer program / instructions, characterized in that, When executed by a processor, the computer program / instructions implement the steps of a method for constructing and using a nuclear structure material database as described in any one of claims 1 to 8.

Images