Method for Extracting Key Quality Characteristics of Nuclear Power Equipment from the Perspective of Materials

Abstract

The method for extracting key quality characteristics of nuclear power equipment from the perspective of materials provided by the present invention. First, collect the point cloud and energy data of the atomic clusters of key material elements of nuclear power equipment, and analyze and identify the element composition of the energy characteristics of the atomic clusters related to the structural stability of nuclear power equipment; input the quality characteristic elements of the atomic clusters into the XGboost and LightGBM prediction models for training respectively, and then judge whether the atomic cluster energy characteristic values output by the XGboost and LightGBM prediction models are optimal. If so, input the quality characteristic elements of the atomic cluster point cloud data to be measured into the optimal prediction model to find the optimal spatial structure and use the second-order energy difference function and dissociation energy to determine the relative stability of the atomic cluster. According to the feedback results, by analyzing the change of the atomic cluster energy, map the atomic cluster energy change to the quality characteristics of nuclear power equipment, so as to determine the requirements for each quality characteristic of nuclear power equipment according to the change law of atomic cluster quality, and support the quality control of nuclear power equipment.

Description

Technical Field

[0001] The present invention relates to a calculation method for quality characteristics, and particularly to a method for extracting key quality characteristics of nuclear power equipment from the perspective of materials. Background Art

[0002] Nuclear power equipment operates under special high-temperature and high-pressure water environment conditions. The quality of its key materials greatly affects the service life of the equipment. Studying the microstructure of key materials to improve the equipment stability is of great significance for strengthening the overall quality control of nuclear power equipment. Among them, atomic clusters are multi-nuclear aggregates between atoms / molecules and macroscopic substances, with a definite atomic composition and chemical structure, representing the nascent state of condensed matter, and being an ideal model for correlating macroscopic properties and the microscopic structure of substances. The quality characteristics, the structure and evolution of their elements play a crucial role in the state evolution process of the equipment stability. Therefore, revealing the quality characteristic law of clusters at the atomic level and understanding the correlation between cluster structure and function are of great significance for researching and preparing functional cluster-based materials and devices for nuclear power equipment and the quality control of the whole life cycle of nuclear power equipment.

[0003] Energy is a key factor affecting the quality characteristics of nuclear power equipment stability. However, different from other substances, energy cannot be obtained by simple linear or interpolation methods. Since the potential energy surface of clusters is relatively complex, the calculation of its potential energy and the search for the global optimized structure are rather troublesome. Traditional theoretical calculation methods require numerical iteration to solve the Schrödinger equation, and with the increase in the number of atoms, the high-precision theoretical calculation time shows exponential growth, which is very time-consuming, and its accuracy may not be guaranteed. Therefore, on the premise of controlling the nuclear power equipment stability within a certain range, improving the energy prediction efficiency can respond more quickly to the energy changes in the life cycle of nuclear power equipment to support quality control. Summary of the Invention

[0004] In view of this, the purpose of the present invention is to provide a method for extracting key quality characteristics of nuclear power equipment from the perspective of materials, that is, to establish a quality model of the energy characteristics of atomic clusters of nuclear power equipment through an efficient calculation method, and decompose it into multiple quality characteristic elements. This is of great significance for establishing a structural stability model and an energy state visualization evolution model of nuclear power equipment, revealing the composition and evolution law of quality characteristic elements of nuclear power equipment energy characteristics in the nucleation and growth evolution processes at the atomic level, realizing the visual traceability of energy characteristics in the co-evolution process, promoting the effective identification of nuclear power equipment stability quality characteristics, the systematic control of quality status, the accurate traceability of quality defects, and improving the quality of nuclear power equipment.

[0005] The method for extracting key quality characteristics of nuclear power equipment from the perspective of materials provided by the present invention includes:

[0006] First, collect the point cloud data and energy data of the atomic clusters of the key material elements of nuclear power equipment, and analyze and identify the constituent elements of the energy characteristics of the atomic clusters related to the structural stability of nuclear power equipment; among them, the mass characteristic elements related to the stability of nuclear power equipment include the point cloud spatial normal vector based on the neighborhood spatial distribution, the point cloud atomic cluster density based on the three-dimensional space, the point cloud atomic cluster spatial principal curvature, the Lennard-Jones potential energy of the atomic cluster, the Sutton-Chen potential energy of the atomic cluster, and the spatial mass symmetry of the point cloud atomic cluster.

[0007] Input the atomic cluster mass characteristic elements into the XGboost prediction model and the LightGBM prediction model for training respectively, and then judge whether the atomic cluster energy values output by the XGboost prediction model and the LightGBM prediction model are optimal. If so, input the mass characteristic elements of the atomic cluster point cloud data to be measured into the optimal prediction model to find the optimal spatial structure and determine the relative stability of the atomic cluster by using the second-order energy difference function and the dissociation energy.

[0008] Preferably: Judging whether the atomic cluster energy values output by the XGboost prediction model and the LightGBM prediction model are optimal includes:

[0009] Judge whether the energy values of the XGboost prediction model and the LightGBM prediction model are greater than the set value. If so, if not, re-train the XGboost prediction model and the LightGBM prediction model. If so, judge whether the error values of the XGboost prediction model and the LightGBM prediction model are less than the set threshold. If so, determine that the current XGboost prediction model or LightGBM prediction model is optimal. If not, re-train the XGboost prediction model and the LightGBM prediction model.

[0010] Preferably: The determination method for the point cloud spatial normal vector of the mass characteristic element based on the neighborhood spatial distribution is as follows:

[0011] Calculate by fitting the surface of the local area where the point is located. The plane is represented by a point and a normal vector. For each point P i , the corresponding covariance matrix C is calculated as follows:

[0012] where: k is the number of neighboring points of point p i , represents the three-dimensional centroid of the nearest neighbor elements, and λ j is the j-th eigenvalue of the covariance matrix, is the j-th eigenvector;

[0013] Process using PCA analysis method and covariance matrix C to obtain the normal vector of the point cloud space of quality characteristic elements based on the neighborhood space distribution.

[0014] Preferably, the spatial principal curvature of the point cloud features of quality characteristic elements is calculated by the following method: Set two mutually perpendicular planes U(0,1,0,0) and plane V(1,0,0,0), and respectively use the bilateral weighted-based method shown in the formula to obtain the change in the normal vector Δn of the current point p in the azimuth θ. θ ;

[0015]

[0016]

[0017]

[0018] where n is the number of neighborhood points; p i represents the i-th point within the neighborhood range; p i The coordinates of the point are (x i , y i , z i ); n p and n i are the normal vectors of the current point and its neighborhood points respectively; ||p i -p|| represents the distance between point p i and point p; δ is 3 times the local point cloud density; w 1i is weighted according to the Euclidean distance between p i and the current point p; w 2i is weighted according to the distance from the neighborhood point to the segmentation plane; the segmentation plane passes through the current point p and its projection in the local tangent plane is a straight line with the direction vector θ, which is determined by the parameters a, b, c, d.

[0019] Preferably, the Sutton-Chen potential energy of the atomic cluster of quality characteristic elements is calculated by the following formula:

[0020] where a represents the lattice constant of the atomic cluster, r ij represents the Euclidean distance between the i-th atom and the j-th atom, c is a constant, and the calculation formula of the intermediate variable p i is as follows:

[0021]

[0022] Preferably, the Lennard-Jones potential energy of the atomic cluster of quality characteristic elements is calculated by the following formula:

[0023] Lennard-Jones potential energy of single-atom clusters:

[0024]

[0025] Lennard-Jones potential energy of di-atom clusters:

[0026]

[0027] Where: r ij represents the Euclidean distance between the i-th atom and the j-th atom, αβ represents the atom type, N represents the number of atoms, ε represents the potential well depth, and σ represents the atomic distance when the potential energy is zero.

[0028] Preferably, the quality characteristic element is determined based on the point cloud atom cluster density in three-dimensional space as follows:

[0029] Define the density function D i :

[0030]

[0031] Where: r ax , r ay and r az are respectively the clustering shape parameters;

[0032] Use the clustering algorithm to update the density function D i :

[0033] Where, D i in this formula represents the density of certain detected clustering centers and other three-dimensional points that are definitely not clustering centers; x i , y i , z i represent the target points, and D cl is the density of the previous clustering center.

[0034] Preferably, the spatial quality symmetry of the quality characteristic element point cloud atom cluster is calculated as follows:

[0035] For each atom cluster, the calculation formula for spatial symmetry is as follows:

[0036]

[0037] H = (X 2 + Y 2 + Z 2 ) 0.5

[0038] Xi, Yi, and Zi are the spatial coordinates of the atomic cluster, X, Y, and Z are the symmetry eigenvalues of the atomic cluster in each direction obtained, and H is the eigenvalue of the symmetry of the finally obtained atomic cluster.

[0039] Preferably, the errors of the XGboost prediction model and the LightGBM prediction model include the mean absolute error, the mean squared error, the median absolute error, and the R2 score.

[0040] Preferably, the method for determining the relative stability of the atomic cluster includes calculating the second-order energy difference and the dissociation energy of the atomic cluster.

[0041] Preferably, analyze the change of the atomic cluster energy according to the feedback result, map the atomic cluster energy change to the quality characteristics of the nuclear power equipment, so as to determine the requirements for each quality characteristic of the nuclear power equipment according to the atomic cluster mass change law, and support the quality control of the nuclear power equipment.

[0042] The beneficial effects of the present invention: Through the present invention, it is possible to process atomic cluster structure data with a large amount of data and high dimensions. By analyzing the change of the atomic cluster energy, map the atomic cluster energy change to the quality characteristics of the nuclear power equipment, so as to determine the requirements for each quality characteristic of the nuclear power equipment according to the atomic cluster mass change law, and support the quality control of the nuclear power equipment. In addition, on the premise of ensuring the accuracy of the calculation results, this method can effectively save the algorithm time, improve the prediction efficiency, and shorten the response speed to the energy change. Description of the Drawings

[0043] Figure 1 It is a flowchart of the present invention.

[0044] Figure 2 Taking the Zr n (n = 2 - 16) atomic cluster as an example, it is a comparison chart of the prediction errors between the present invention and the traditional algorithm. Among them, (a) is the error chart of the traditional algorithm, and (b) is the error chart of the present invention. Detailed Embodiments

[0045] It should be noted that the following only describes the present invention in detail with preferred embodiments. Any modification and equivalent replacement of the technical solutions of the present invention by those skilled in the art are included in the scope of protection required by the technical solutions of this application.

[0046] The method for extracting key quality characteristics of nuclear power equipment based on the material perspective provided by the present invention includes:

[0047] Including:

[0048] First, collect the key material Zr of the nuclear power equipment nPoint cloud data and energy data of atomic clusters (n = 2 - 16), analyze and identify the component elements of the energy characteristics of atomic clusters related to the structural stability of nuclear power equipment; among them, the quality characteristic elements related to the stability of nuclear power equipment include the point cloud spatial normal vector based on the neighborhood spatial distribution, the point cloud atomic cluster density based on three-dimensional space, the spatial principal curvature of the point cloud atomic cluster, the Lennard-Jones potential energy of the atomic cluster, the Sutton-Chen potential energy of the atomic cluster, and the spatial mass symmetry of the point cloud atomic cluster;

[0049] Input the atomic cluster quality characteristic elements into the XGboost prediction model and the LightGBM prediction model for training respectively, and then judge whether the atomic cluster energy values output by the XGboost prediction model and the LightGBM prediction model are optimal. If so, input the quality characteristic elements of the atomic cluster point cloud data to be measured into the optimal prediction model to find the optimal spatial structure and use the second-order energy difference and dissociation energy function to determine the relative stability of the atomic cluster. Under the above method,

[0050] It can process atomic cluster structure data with a large amount of data and high dimensions. By analyzing the change of atomic cluster energy, map the change of atomic cluster energy to the quality characteristics of nuclear power equipment, so as to determine the requirements for each quality characteristic of nuclear power equipment according to the change law of atomic cluster quality, and support the quality control of nuclear power equipment. In addition, under the premise of ensuring the accuracy of the calculation results, this method can effectively save algorithm time, improve prediction efficiency, and shorten the response speed to energy changes.

[0051] Among them: The method for determining the point cloud spatial normal vector based on the neighborhood spatial distribution of the quality characteristic elements is as follows:

[0052] Calculate the surface of the local fit of the area where the point is located. The plane is represented by a point and a normal vector. For each point P i , the calculation of the corresponding covariance matrix C is as follows:

[0053] Among them: k is the number of neighboring points of point p i , represents the three-dimensional centroid of the nearest neighbor elements, and λ j is the j-th eigenvalue of the covariance matrix, is the j-th eigenvector;

[0054] Use PCA analysis method and covariance matrix C for processing to obtain the point cloud spatial normal vector based on the neighborhood spatial distribution. Under this quality characteristic element, it has 7-dimensional eigenvalues, which thus ensures the accuracy of the entire algorithm result.

[0055] In a preferred embodiment: determining whether the atomic cluster energy values output by the XGboost prediction model and the LightGBM prediction model are optimal includes:

[0056] Determining whether the energy values of the XGboost prediction model and the LightGBM prediction model are greater than a set value. If so, if not, then retrain the XGboost prediction model and the LightGBM prediction model. If so, then determine whether the error values of the XGboost prediction model and the LightGBM prediction model are less than a set threshold. If so, then determine that the current XGboost prediction model or LightGBM prediction model is optimal. If not, then retrain the XGboost prediction model and the LightGBM prediction model.

[0057] Among them, the errors of the XGboost prediction model and the LightGBM prediction model include the mean absolute error, the mean squared error, the median absolute error, and the R2 score. Through the above method, the accuracy of the final prediction result can be ensured.

[0058] In a preferred embodiment: the spatial principal curvature of the mass characteristic element point cloud features is calculated by the following method: Set two mutually perpendicular planes U(0,1,0,0) and plane V(1,0,0,0), and respectively use the method based on bilateral weighting shown in the formula to obtain the normal vector change Δn of the current point p in the azimuth θ θ ;

[0059]

[0060]

[0061]

[0062] Among them, n is the number of neighborhood points; p i represents the i-th point within the neighborhood range; p i The coordinates of the point are (x i , y i , z i ); n p and n i are the normal vectors of the current point and its neighborhood points respectively; ||p i -p|| represents the distance between point p i and point p; δ is 3 times the local point cloud density; w 1i is weighted according to the Euclidean distance between p i and the current point p; w 2iThe weighting is performed according to the distance from the neighborhood points to the segmentation plane; the segmentation plane passes through the current point p and its projection in the local tangent plane is a straight line with the direction vector θ, which is determined by the parameters a, b, c, and d.

[0063] In a preferred embodiment: the Sutton-Chen potential energy of the atomic cluster of the quality characteristic element is calculated by the following formula:

[0064] where a represents the lattice constant of the atomic cluster, r ij represents the Euclidean distance between the i-th atom and the j-th atom, c is a constant, and the calculation formula of the intermediate variable p i is as follows:

[0065]

[0066] In a preferred embodiment: the Lennard-Jones potential energy of the atomic cluster of the quality characteristic element is calculated by the following formula:

[0067] The Lennard-Jones potential energy of a single atomic cluster:

[0068]

[0069] The Lennard-Jones potential energy of a double atomic cluster:

[0070]

[0071] where: r ij represents the Euclidean distance between the i-th atom and the j-th atom, αβ represents the atomic type, N represents the number of atoms, ε represents the depth of the potential well, and σ represents the atomic distance when the potential energy is zero.

[0072] In a preferred embodiment: the quality characteristic element is determined based on the atomic cluster density of the point cloud in three-dimensional space as follows:

[0073] Define the density function D i :

[0074]

[0075] where: r ax , r ay and r az are respectively the clustering shape parameters;

[0076] Use the clustering algorithm to update the density function D i as follows:

[0077] where the D in this formula iRepresents the density of certain detected clustering centers and other three-dimensional points that are definitely not clustering centers; x i , y i , z i Represents the target point, D cl Is the density of the previous clustering center.

[0078] In a preferred embodiment: The spatial quality symmetry of the mass characteristic element point cloud atomic clusters is calculated as follows:

[0079] For each atomic cluster, the calculation formula for spatial symmetry is as follows:

[0080]

[0081] H = (X 2 + Y 2 + Z 2 ) 0.5

[0082] Xi, Yi, Zi are the spatial coordinates of the atomic cluster, X, Y, Z are the atomic cluster symmetry eigenvalue obtained in each direction, and H is the eigenvalue of the atomic cluster symmetry finally obtained.

[0083] In a preferred embodiment: The relative stability of Zr n (n = 2 - 16) clusters is calculated as follows:

[0084] By examining the second-order difference (Δ2E) and dissociation energy (ΔE) of the Zr n (n = 2 - 16) cluster energy to judge

[0085] Its relative stability. The second-order energy difference and dissociation energy are defined as follows:

[0086] Δ2E(n) = E t (Zr n+1 ) + E t (Zr n-1 ) - 2E t (Zr n )

[0087] ΔE(n) = E t (Zr n-1 ) + E t (Zr) - E t (Zr n )

[0088] Where E t (Zr n+1 ), E t (Zr n-1 ) and E t (Zr n ) respectively represent Zrn+1 , Zr n-1 and Zr n The total energy of the ground-state cluster, E t (Zr) represents the energy of a free Zr atom.

[0089] The second-order energy difference is a quantity that can sensitively reflect the relative stability of the cluster. The larger the second-order energy difference value, the higher the relative stability of the cluster compared to its neighbors. The dissociation energy indicates the energy required for the cluster to dissociate an atom. The larger its value, the more stable the cluster. Thus, it is possible to determine whether the atomic cluster has stability.

[0090] In a preferred embodiment: By analyzing the energy change of Zr n (n = 2 - 16) atomic clusters, map the energy change of the atomic clusters to the quality characteristics of nuclear power equipment, so as to determine the requirements for each quality characteristic of nuclear power equipment according to the law of atomic cluster mass change, and support the quality control of nuclear power equipment.

Claims

1. A method for extracting key quality characteristics of nuclear power equipment from the perspective of materials, characterized in that: Including: First, collect the point cloud data and energy data of the atomic clusters of nuclear power equipment materials, and analyze and identify the element composition of the energy characteristics of atomic clusters related to the structural stability of nuclear power equipment; among them, the quality characteristic elements related to the stability of nuclear power equipment include the point cloud space normal vector based on the neighborhood space distribution, the point cloud atomic cluster density based on the three-dimensional space, the main curvature of the point cloud atomic cluster space, the Lennard-Jones potential energy of the atomic cluster, the Sutton-Chen potential energy of the atomic cluster, and the spatial mass symmetry of the point cloud atomic cluster. Input the atomic cluster quality characteristic elements into the XGboost prediction model and the LightGBM prediction model for training respectively, and then judge whether the atomic cluster energy values output by the XGboost prediction model and the LightGBM prediction model are optimal. If so, input the quality characteristic elements of the atomic cluster point cloud data to be measured into the optimal prediction model to find the optimal spatial structure, and use the second-order energy difference function and dissociation energy to determine the relative stability of the atomic cluster. Analyze the change of atomic cluster energy according to the feedback result, map the atomic cluster energy change to the quality characteristics of nuclear power equipment, so as to determine the requirements for each quality characteristic of nuclear power equipment according to the change law of atomic cluster quality, and support the quality control of nuclear power equipment.

2. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1, characterized in that: Judging whether the atomic cluster energy values output by the XGboost prediction model and the LightGBM prediction model are optimal includes: Judge whether the energy values of the XGboost prediction model and the LightGBM prediction model are greater than the set value. If not, retrain the XGboost prediction model and the LightGBM prediction model. If so, judge whether the error values of the XGboost prediction model and the LightGBM prediction model are less than the set threshold. If so, determine that the current XGboost prediction model or LightGBM prediction model is optimal. If not, retrain the XGboost prediction model and the LightGBM prediction model.

3. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1 is characterized in that: for The determination method of the point cloud space normal vector of the quality characteristic element based on the neighborhood space distribution is as follows: By calculating the surface of the local fitting of the area where the point is located, the plane is represented by a point and a normal vector. For each point P i , the covariance matrix C is calculated as follows: where: k is the number of points p i adjacent to the point, denotes the three-dimensional centroid of the nearest neighbor elements, λ j is the j-th eigenvalue of the covariance matrix, and is the j-th eigenvector; Use the PCA analysis method and the covariance matrix C for processing to obtain the point cloud space normal vector of the quality characteristic element based on the neighborhood space distribution.

4. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1, wherein: The spatial principal curvature of the quality characteristic element point cloud feature is calculated by the following method: Set two mutually perpendicular planes U(0,1,0,0) and plane V(1,0,0,0), and use the bilateral weighted-based method shown in the formula to obtain the change in the normal vector Δn of the current point p in the azimuth θ θ ; Among them, n is the number of neighborhood points; p i represents the i-th point within the neighborhood range; p i The coordinates of the point are (x i , y i , z i ); n p and n i are the normal vectors of the current point and its neighborhood points respectively; ||p i - p|| represents the distance between point p i and point p; δ is three times the local point cloud density; w 1i is weighted according to the Euclidean distance between p i and the current point p; w 2i is weighted according to the distance from the neighborhood point to the segmentation plane; The segmentation plane passes through the current point p and its projection within the local tangent plane is a straight line with the direction vector θ, which is determined by the parameters a, b, c, and d.

5. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1, characterized in that: The Sutton-Chen potential energy of the atomic cluster of the quality characteristic element is calculated by the following formula: where a represents the lattice constant of the atomic cluster, r ij represents the Euclidean distance between the i-th atom and the j-th atom, c is a constant, and the intermediate variable p i is calculated as follows:

6. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1, characterized in that: The Lennard-Jones potential energy of the atomic cluster of the quality characteristic element is calculated by the following formula: The Lennard-Jones potential energy of a single atomic cluster The Lennard-Jones potential energy of a double atomic cluster where: r ij represents the Euclidean distance between the i-th atom and the j-th atom, αβ represents the atomic type, N represents the number of atoms, ε represents the depth of the potential well, and σ represents the atomic distance when the potential energy is zero.

7. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1, characterized in that: The determination of the point cloud atomic cluster density of the quality characteristic element based on the three-dimensional space is as follows, including the initial stage and the clustering stage. Among them, in the initial stage, the density function D is defined i : where: r ax , r ay and r az are clustering shape parameters, respectively; In the clustering stage, the density function D i is updated, and the update formula is as follows: wherein, D in this update formula i represents the density of the detected clustering center and other three-dimensional points that are not clustering centers; x i , y i , z i represent the target points, and D cl is the density of the previous clustering center.

8. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 1, wherein: The calculation of the spatial mass symmetry of the point cloud atomic cluster of the quality characteristic element is as follows: The calculation formula for the spatial symmetry of each atomic cluster is as follows H = (X 2 + Y 2 + Z 2 ) 0.5 X i , Y i , Z i are the spatial coordinates of the atomic cluster, X, Y, and Z are the symmetry eigenvalues of the atomic cluster in each obtained direction, and H is the eigenvalue of the symmetry of the finally obtained atomic cluster.

9. The method for extracting key quality characteristics of nuclear power equipment based on the material perspective according to claim 2, characterized in that: The errors of the XGboost prediction model and the LightGBM prediction model include the mean absolute error, the mean squared error, the median absolute error, and the R2 score.

Images