Re-crystallization nucleation site prediction method and system based on multi-factor scoring model

Abstract

The present application relates to a kind of recrystallization nucleation site prediction method and system based on multi-factor scoring model, the method includes: the electron backscattering diffraction scanning of cold deformation metal sample is carried out, and the electron backscattering diffraction data containing crystal orientation information is obtained;Based on the electron backscattering diffraction data, local orientation difference field is calculated;From local orientation difference field, the spatial distribution of the grain boundary information, including large-angle grain boundary and coincident site lattice grain boundary is extracted;Based on the morphological characteristics and spatial distribution of local orientation difference field, shear band region is automatically identified;Multi-factor weighted scoring model is established, and the nucleation tendency score of each pixel position is calculated, wherein multi-factor includes local energy storage factor, large-angle grain boundary distance factor, shear band factor and coincident site lattice grain boundary factor;According to the score threshold value, determine the preferential nucleation site, output nucleation site spatial distribution diagram and statistical result.Compared with prior art, the present application realizes the automatic and accurate prediction of nucleation site.

Description

Technical Field

[0001] This invention relates to the field of characterization and analysis of metallic materials, and in particular to a method for predicting recrystallization nucleation sites based on a multi-factor scoring model. Background Technology

[0002] Recrystallization is a significant microstructural evolution phenomenon that occurs in metallic materials during heat treatment, and it has a decisive impact on the final properties of the material. During recrystallization, nucleation is a key step that determines the final texture and grain structure. Accurately predicting the spatial distribution of recrystallization nucleation sites is crucial for understanding the recrystallization mechanism and optimizing heat treatment processes.

[0003] Taking grain-oriented silicon steel as an example, its excellent magnetic properties are highly dependent on the sharp Goss texture ({110}). <001> The nucleation sites of Goss grains are mainly distributed in the special microstructures formed during cold rolling deformation, including high-energy storage regions, near large-angle grain boundaries, and inside shear bands. However, the spatial distribution of these nucleation sites is complex, and traditional methods rely heavily on human experience and judgment, which is highly subjective. Different researchers may have significantly different judgments, making accurate quantitative prediction difficult.

[0004] Electron backscatter diffraction (EBSD) technology can provide rich information about the microstructure of materials, including crystal orientation, grain boundary type, and local orientation difference (KAM). Based on EBSD data, deformation energy storage distribution, grain boundary type, and shear band structure can be characterized.

[0005] However, existing methods for integrating electron backscatter diffraction data into nucleation site prediction have the following problems: On the one hand, existing analyses often only consider a single factor, such as focusing only on KAM or grain boundary type, lacking a quantitative description of the combined effects of multiple factors; on the other hand, as shear bands are important nucleation sites, existing automatic identification algorithms are still immature, making it difficult to achieve batch and standardized analysis, which greatly affects the accuracy and efficiency of recrystallization nucleation site prediction.

[0006] Therefore, there is a need for a method to predict recrystallization nucleation sites that can comprehensively consider multiple factors and achieve automated quantitative prediction. Summary of the Invention

[0007] The purpose of this invention is to overcome the shortcomings of the existing technology by providing a method and system for predicting recrystallization nucleation sites based on a multi-factor scoring model, thereby achieving automated and quantitative prediction of nucleation sites.

[0008] The objective of this invention can be achieved through the following technical solutions: A method for predicting recrystallization nucleation sites based on a multi-factor scoring model, comprising: Electron backscatter diffraction scanning was performed on cold-deformed metal samples to obtain electron backscatter diffraction data containing crystal orientation information. Calculate the local orientation difference distribution field based on electron backscattering diffraction data; Extract grain boundary information to obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping locations; Based on the morphological characteristics and spatial distribution of the local orientation difference distribution field, shear bands are automatically identified; A multi-factor weighted scoring model is established to score the nucleation tendency of each pixel. The multi-factors include local energy storage factor, large-angle grain boundary distance factor, shear band factor, and coincident position lattice grain boundary factor. Priority nucleation sites are determined based on the scoring threshold, and the spatial distribution map and statistical results of the nucleation sites are output.

[0009] Preferably, the calculation of the local orientation difference distribution field based on electron backscattering diffraction data specifically includes: For each pixel in the electron backscatter diffraction data, calculate its orientation difference with all pixels in its surrounding neighborhood, and take the average value as the local orientation difference of the current point. The neighborhood size is a 5×5 pixel window. The maximum orientation difference threshold is set to 5°. Adjacent points exceeding the preset maximum orientation difference threshold are filtered out to obtain the local orientation difference distribution field.

[0010] Preferably, the local energy storage factor is calculated using the following expression: f_KAM(x,y)=clip{[exp(α·(KAM(x,y)-K_low) / (K_high-K_low))-1] / (exp(α)- 1),0,1} , in, f_KAM(x,y) Represents pixels in two-dimensional electron backscattering diffraction data (x,y) The corresponding local energy storage factor; KAM(x,y) Pixels in two-dimensional electron backscatter diffraction data (x,y) The corresponding local orientation difference; α is a nonlinear shape parameter used to control the degree of nonlinearity of the contribution of energy storage to the nucleation driving force; K_low Low energy storage threshold; K_ high The high energy storage threshold is defined by the `clip` function, which is a clipping function.

[0011] Preferably, the calculation process for the large-angle grain boundary distance factor is as follows: All grain boundaries with an orientation difference greater than 15° are extracted as large-angle grain boundaries; The distance from each pixel to the nearest large-angle grain boundary is calculated using Euclidean distance transformation. d_HAGB(x,y) ; The expression for calculating the large-angle grain boundary distance factor is: f_HAGB(x,y)=clip[1-d_HAGB(x,y) / d_crit,0,1] , Where d_crit is the critical distance threshold for large-angle grain boundaries.

[0012] Preferably, the process of identifying the shear band includes the following steps: Extracting local orientation differences less than the low energy storage threshold K_low The region is selected as the candidate shear band region; Connectivity labeling and morphological analysis are performed on candidate shear band regions to filter out band regions with an aspect ratio greater than a set value. Calculate the angle between the main axis direction of the strip region and the rolling direction, and retain the region where the angle is within a specific range. The specific range includes two angle intervals: 12°-22° and 25°-35°, which correspond to the type A shear strip with an angle of 17° to the rolling direction and the type B shear strip with an angle of 30° to the rolling direction in cold-rolled iron-based alloys. Regions that satisfy the condition that the mean local orientation difference within the strip region is lower than the low energy storage threshold K_low and the mean local orientation difference between the adjacent regions on both sides of the strip region is higher than the high energy storage threshold K_high are identified as shear zones. The shear band factor is a binary factor. For pixels located within the identified shear band region, the shear band factor is set to 1, and for pixels located outside the shear band region, the shear band factor is set to 0.

[0013] Preferably, the calculation process for the grain boundary factor at the overlapping location lattice includes: Identify all CSL grain boundaries that conform to Brandon's criterion, including Σ3, Σ5, Σ9 and Σ11 grain boundaries, where the Brandon criterion tolerance angle is 15° / √Σ; For oriented silicon steel, calculate the distance d_CSL9(x,y) from each pixel to the nearest Σ9 grain boundary; Calculate the lattice grain boundary factor at the coincident site: f_CSL(x,y)=clip[1-d_CSL9(x,y) / d_CSL_crit,0,1] , Where d_CSL_crit is the critical distance between the lattice grain boundaries at the overlapping positions.

[0014] Preferably, the multi-factor weighted scoring model scores the nucleation tendency of each pixel: Score(x,y)=w 1 ×f_KAM+w 2 ×f_HAGB+w 3 ×f_shear+w 4 ×f_CSL , in, f_KAM As a local energy storage factor, f_HAGB For large-angle grain boundary distance factors, f_shearFor shear band factor, f_CSL For the lattice grain boundary factor at the coincident position; w 1 ~w 4 These are the corresponding factor weights, satisfying... w 1 + w 2 +w 3 +w 4 =1, with initial values ​​set to 0.35, 0.30, 0.25, and 0.10 respectively; Based on different material systems and process conditions, adjustments and optimizations are made to the initial values.

[0015] Preferably, the initial values ​​are adjusted and optimized based on the cold rolling reduction rate, initial grain size, and material composition, specifically including: When the material type is pure iron or IF steel, reduce the weight of the shear band factor w3, correspondingly increase the weight of the local energy storage factor w1 and the weight of the large angle grain boundary distance factor w2, and reduce the weight of the grain boundary factor w4 at the overlapping position lattice. When the material type is non-oriented silicon steel, increase the weight of the large-angle grain boundary distance factor w2 and decrease the weight of the shear band factor w3. When the material type is low carbon steel, increase the weight of the large-angle grain boundary distance factor w2 and decrease the weight of the shear band factor w3; When the material type is austenitic stainless steel, reduce the weight of the local energy storage factor w1 and increase the weight of the grain boundary factor w4 at the overlapping position lattice. When the cold rolling reduction rate is greater than the set value, the weight of the shear band factor w3 is further increased based on the adjustment according to the material type. When the initial grain size is greater than the set value, the weight of the large-angle grain boundary distance factor w2 is further increased based on the adjustment according to the material type.

[0016] Preferably, the method further includes classifying and statistically analyzing nucleation sites according to grain orientation, specifically including: Based on the crystal orientation information of each pixel in the two-dimensional electron backscatter diffraction data, the predicted nucleation sites are classified according to the orientation of their parent grains. The number and density of nucleation sites within grains of different orientation types were statistically analyzed. Output the comparison results of nucleation tendencies of grains with different orientations to evaluate the evolution trend of recrystallization texture.

[0017] A recrystallization nucleation site prediction system based on a multi-factor scoring model, using the method described above, the system comprising: The data acquisition module is used to perform electron backscatter diffraction scanning on the scanned cold-deformed metal sample to acquire electron backscatter diffraction data containing crystal orientation information. The local orientation difference calculation module is used to calculate the local orientation difference distribution field based on crystal orientation information; The grain boundary analysis module is used to extract grain boundary information and obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping locations; The shear band detection module is used to automatically identify shear bands based on the morphological characteristics and spatial distribution of the local orientation difference distribution field. The nucleation scoring module is used to perform multi-factor weighted scoring calculations and determine preferred nucleation sites based on scoring thresholds; The visualization module is used to display the distribution map of nucleation sites, the local orientation difference distribution field map, and the scoring heatmap; The report generation module is used to output analysis reports including nucleation density, nucleation site ratio, and orientation classification statistics.

[0018] Compared with the prior art, the present invention has the following advantages: (1) This invention is the first to model four types of factors—local energy storage, large-angle grain boundaries, shear bands, and overlapping lattice grain boundaries—within a unified framework. It uses multi-factor weighted scoring to quantify the nucleation tendency. On the one hand, standardized quantitative identification ensures the consistency and repeatability of identification. On the other hand, compared with single-factor analysis, this invention further improves the comprehensiveness and accuracy of recrystallization nucleation site identification.

[0019] (2) The size of the neighborhood window used in the calculation of the local energy storage factor directly determines the balance between spatial resolution and statistical stability of the measurement results. For oriented silicon steel, the typical scanning step size is 0.1~0.3μm. If a 3×3 neighborhood window is used, the number of sampling points is only 8, which is too few and easily leads to large statistical fluctuations. When the window size is increased to 7×7 or above, the spatial resolution is significantly reduced, which easily causes information aliasing between adjacent grains. The present invention uses a 5×5 neighborhood window, which corresponds to a physical size of about 0.5~1.5μm, matching the typical size of subgrain boundaries and dislocation cells (about 0.5~2μm), and can achieve a more reasonable balance between spatial resolution and statistical stability.

[0020] (3) Considering that the general threshold for large-angle grain boundaries in EBSD is 15°, while 5° corresponds to the dislocation wall / subgrain boundary category inside small-angle grain boundaries. Adjacent point pairs with orientation differences exceeding 5° are very likely to cross obvious subgrain boundaries. Including them in the local energy storage factor calculation will artificially increase the local energy storage value of pixels near the grain boundary, resulting in false high energy storage peaks at the grain boundary position, which interferes with the judgment of nucleation sites.

[0021] (4) The automatic shear band identification method proposed in this invention is based on four criteria: KAM threshold screening, morphological band analysis, 17° / 30° angle verification and KAM gradient confirmation on both sides. It can realize batch and standardized shear band detection and break through the bottleneck of existing methods that rely on manual judgment.

[0022] (5) Considering that in the high energy storage region, the dislocation density is close to the threshold of the driving force required for critical nucleation, a small increase in energy storage will cause a significant nonlinear increase in the nucleation probability. The present invention uses an exponential normalization formula to calculate the local energy storage factor. Since the exponential form has a larger slope at the high energy storage end in the normalization interval, it is consistent with the physical distribution law of recrystallization nucleation driving force, and more accurately reflects the nonlinear contribution of the high energy storage region to the nucleation driving force, and has better physical rationality.

[0023] (6) Based on a comprehensive consideration of the Goss nucleation physical mechanism of grain-oriented silicon steel, this invention clearly reveals the Σ9 grain boundary (38.94° / ) in grain-oriented silicon steel. <110> The special relationship between the Goss kernel and the Goss kernel is quantified as a spatial distance factor and incorporated into the score, which has better physical rationality.

[0024] (7) This invention has good versatility. By adjusting the weighting coefficients, it can be applied to the recrystallization nucleation prediction of various deformable metal materials such as oriented silicon steel, non-oriented silicon steel, pure iron, IF steel, low carbon steel, and stainless steel.

[0025] (8) The prediction results of the present invention can be directly used to guide the optimization of heat treatment process. The recrystallization texture evolution trend can be predicted by analyzing the nucleation tendency of grains with different orientations. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of the method flow of the present invention; Figure 2 Flowchart for shear tape recognition; Figure 3 This is a schematic diagram of the structure of a multi-factor weighted scoring model; Figure 4 System architecture diagram; Figure 5 The image shown is an EBSD orientation image of cold-rolled oriented silicon steel in the embodiment, where the horizontal and vertical axes are the pixel coordinates of the EBSD scan points, corresponding to a scan step size of 0.2 μm, and the color scale is the orientation color of the IPF-Z inverse pole figure. Figure 6 The image shows the KAM field distribution of cold-rolled oriented silicon steel in the example. The horizontal and vertical axes are the pixel coordinates of the EBSD scan points, corresponding to a scan step size of 0.2 μm. The color scale represents the local orientation difference, and the unit is °. Figure 7This is a grain boundary type distribution diagram of cold-rolled oriented silicon steel in the embodiment, where the horizontal and vertical axes are the pixel coordinates of EBSD scan points, corresponding to a scan step size of 0.2 μm, and the color scale represents the grain boundary orientation difference, in °. Figure 8 The image shows the shear band identification results of cold-rolled oriented silicon steel in the embodiment. The horizontal and vertical axes are the pixel coordinates of the EBSD scan points, corresponding to a scan step size of 0.2μm. The color scale is the shear band factor f_shear (dimensionless, with a value of 0 or 1). Figure 9 The image shows the CSL grain boundary identification results of cold-rolled oriented silicon steel in the example, where the horizontal and vertical axes are the pixel coordinates of the EBSD scan points, corresponding to a scan step size of 0.2 μm; Figure 10 The image shows the nucleation score of cold-rolled oriented silicon steel in the example. The horizontal and vertical axes are the pixel coordinates of the EBSD scan points, corresponding to a scan step size of 0.2 μm. The color scale is the nucleation tendency score (dimensionless, with a value range of 0 to 1). Figure 11 The nucleation site determination results of cold-rolled oriented silicon steel in the embodiment are shown. The horizontal and vertical axes are both EBSD scan point pixel coordinates, corresponding to a scan step size of 0.2μm. The color mark is a binary marker of the nucleation site (dimensionless, 0 represents a non-nucleation site, and 1 represents a nucleation site). Figure 12 The above are statistical results of the nucleation tendency of different orientation grains in cold-rolled oriented silicon steel in the examples. The horizontal axis represents the type of each texture component (Goss, Cube, RotCube, Brass, Copper, Gamma1, the meaning of which can be found in the specific implementation section of the specification), and the vertical axis represents the proportion of the number of nucleation sites in each texture component to the total number of nucleation sites (in %). Figure 13 This is a nucleation score diagram of the hot-rolled industrial pure iron sample in the example. The horizontal and vertical axes are the pixel coordinates of the EBSD scan points, corresponding to a scan step size of 0.3 μm. The color scale is the nucleation tendency score (dimensionless, with a value range of 0 to 1). Figure 14 The above are statistical results of the nucleation tendency of different orientation grains in the hot-rolled industrial pure iron samples in the examples. The horizontal axis represents the type of each texture component (Cube, RotCube, Brass, Gamma1, the meaning of which can be found in the specific implementation section of the specification), and the vertical axis represents the proportion of the number of nucleation sites in each texture component to the total number of nucleation sites (in %). Detailed Implementation

[0027] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0028] Example 1 like Figure 1 As shown, this embodiment provides a method for predicting recrystallization nucleation sites based on a multi-factor scoring model, including: S1. Perform electron backscatter diffraction scanning on the cold-deformed metal sample to obtain electron backscatter diffraction data containing crystal orientation information. S2. Calculate the local orientation difference distribution field based on electron backscattering diffraction data; S3. Extract grain boundary information to obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping positions; S4. Based on the morphological characteristics and spatial distribution of the local orientation difference distribution field, automatically identify shear bands; S5. Establish a multi-factor weighted scoring model to score the nucleation tendency of each pixel. The multi-factors include local energy storage factor, large-angle grain boundary distance factor, shear band factor and coincident position lattice grain boundary factor. S6. Determine the preferred nucleation sites based on the scoring threshold, and output the spatial distribution map and statistical results of the nucleation sites.

[0029] This embodiment also provides a recrystallization nucleation site prediction system based on a multi-factor scoring model, including: The data acquisition module is used to perform electron backscatter diffraction scanning on the scanned cold-deformed metal sample to acquire electron backscatter diffraction data containing crystal orientation information. The local orientation difference calculation module is used to calculate the local orientation difference distribution field based on crystal orientation information; The grain boundary analysis module is used to extract grain boundary information and obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping locations; The shear band detection module is used to automatically identify shear bands based on the morphological characteristics and spatial distribution of the local orientation difference distribution field. The nucleation scoring module is used to perform multi-factor weighted scoring calculations and determine preferred nucleation sites based on scoring thresholds; The visualization module is used to display the distribution map of nucleation sites, the local orientation difference distribution field map, and the scoring heatmap; The report generation module is used to output analysis reports including nucleation density, nucleation site ratio, and orientation classification statistics.

[0030] Example 2 like Figure 1 As shown, this embodiment provides a method for predicting recrystallization nucleation sites based on a multi-factor scoring model. The method includes the following steps: S1. Perform electron backscatter diffraction (EBSD) scans on the cold-deformed metal sample to obtain electron backscatter diffraction data containing crystal orientation information.

[0031] In this step, the cold-rolled and deformed metal sample is electrolytically polished and then scanned using a scanning electron microscope equipped with an electron backscatter diffraction (EBSD) probe. Scanning parameters include: scan step size of 0.1-0.5 μm, accelerating voltage of 20 kV, and working distance of 15-20 mm. The acquired EBSD data includes Euler angle information and confidence index (CI) for each measurement point, and the data is saved as a .ang or .ctf format file.

[0032] S2. Calculate the local orientation difference KAM distribution field based on electron backscatter diffraction data.

[0033] KAM (Kernel Average Misorientation) characterizes the local orientation gradient of a material's micro-regions and is related to local dislocation density and deformation energy storage. During calculation, for each pixel, the orientation difference between it and all surrounding pixels is calculated, and the average value is taken as the local orientation difference KAM value for that point. To avoid cross-grain boundary calculations, a maximum orientation difference threshold of 5° is set, and neighboring points exceeding the preset maximum orientation difference threshold are filtered out to obtain the local orientation difference distribution field.

[0034] The specific calculation process includes: For each pixel in the two-dimensional electron backscatter diffraction data, calculate its orientation difference with all pixels in its surrounding neighborhood, and take the average value as the local orientation difference of the current point.

[0035] The neighborhood size uses a 3×3 to 7×7 pixel window, preferably a 5×5 pixel window. For a typical EBSD scan with a scan step size of 0.1 to 0.3 μm, the 5×5 pixel window corresponds to a physical size of 0.5 to 1.5 μm, which matches the typical size of subgrains in cold-deformed metals, and can achieve a more reasonable balance between spatial resolution and statistical stability.

[0036] In this embodiment, the maximum orientation difference threshold is set in the range of 3° to 8°. Adjacent point pairs exceeding this threshold are excluded to avoid cross-grain boundary calculations. Further, the orientation difference threshold in this embodiment is preferably 5°. The advantage is that the general threshold for large-angle grain boundaries (HAGB) in EBSD is 15°, while 5° corresponds to the dislocation wall / subgrain boundary category within small-angle grain boundaries. Studies have shown that adjacent point pairs with orientation differences exceeding 5° are very likely to cross significant subgrain boundaries. Including them in KAM calculations would artificially inflate the KAM values ​​of pixels near the grain boundary, leading to false high-energy peaks at the grain boundary location and interfering with the determination of nucleation sites. When the threshold is set >5° (e.g., 8° or 10°), this grain boundary interference effect increases significantly, making it difficult to distinguish between the true high-energy region (within the shear band) and the region near the grain boundary.

[0037] S3. Extract grain boundary information to obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping positions.

[0038] Based on the orientation difference between adjacent pixels, the location and type of grain boundaries are identified: those with an orientation difference of less than 15° are small-angle grain boundaries (LAGB), and those with an orientation difference of 15° or greater are large-angle grain boundaries (HAGB).

[0039] Simultaneously, various CSL grain boundaries are identified according to the Brandon criterion (tolerance angle = 15° / √Σ), including Σ3 (twin boundary, 60° / <111> ), Σ5 (36.87° / <100> ), Σ9 (38.94° / <110> ), Σ11 (50.48° / <110> )wait.

[0040] S4. Based on the morphological characteristics and spatial distribution of the local orientation difference distribution field, automatically identify shear bands, such as... Figure 2 As shown, it specifically includes: S4-1, Extracting local orientation differences less than the low energy storage threshold K_low The region is selected as the candidate shear band region. In this embodiment, the low energy storage threshold is considered. K_low The angle was set to 0.5°. These regions exhibited lower local orientation difference (KAM) values ​​because they had partially recovered after deformation.

[0041] S4-2. Perform connected component labeling and morphological analysis on the candidate shear band regions to filter out band regions with an aspect ratio greater than a set value (set to 5 in this embodiment).

[0042] S4-3. Calculate the angle between the principal axis of the strip region and the rolling direction, retaining the region with the angle within a specific range. This specific range includes two angle intervals: 12°-22° and 25°-35°. These correspond to the Type A shear band (related to the Goss orientation grain activation slip system) with an angle of 17° to the rolling direction in cold-rolled iron-based alloys, and the Type B shear band (related to the {111} orientation grain activation slip system) with an angle of 30° to the rolling direction. By setting a ±5° tolerance based on the characteristic angle using these angle intervals, the angular deviation caused by grain orientation dispersion and rolling inhomogeneity in actual EBSD measurements can be effectively accommodated.

[0043] S4-4. Set a high energy storage threshold K_high=2.0°. Regions that meet the conditions that the average local energy storage within the strip region is lower than the low energy storage threshold K_low and the average local energy storage of adjacent regions on both sides of the strip region is higher than the high energy storage threshold K_high (there is a strain gradient) are identified as shear zones.

[0044] S5. Establish a multi-factor weighted scoring model to score the nucleation tendency of each pixel.

[0045] In this embodiment, a multi-factor weighted scoring model is constructed based on multiple factors, including local energy storage factor, large-angle grain boundary distance factor, shear band factor, and coincident position lattice grain boundary factor. Figure 3 As shown, the mathematical expression is: Score(x,y)=w 1 ×f_KAM+w 2 ×f_HAGB+w 3 ×f_shear+w 4 ×f_CSL , in, f_KAM As a local energy storage factor, f_HAGB For large-angle grain boundary distance factors, f_shear For shear band factor, f_CSL For the lattice grain boundary factor at the coincident position; w 1 ~w 4 These are the corresponding factor weights, satisfying... w 1 + w 2 +w 3 +w 4 =1.

[0046] The specific calculation process is as follows: 1) Local energy storage factor f_KAM The expression is: f_KAM(x,y)=clip{[exp(α·(KAM(x,y)-K_low) / (K_high-K_low))-1] / (exp(α)- 1),0,1} , in, f_KAM(x,y) Represents pixels in two-dimensional electron backscattering diffraction data (x,y) The corresponding local energy storage factor; KAM(x,y) Pixels in two-dimensional electron backscatter diffraction data (x,y) The corresponding local orientation difference; α is a nonlinear shape parameter, ranging from 2 to 5, used to control the degree of nonlinearity of the contribution of energy storage to the nucleation driving force. When α approaches 0, it degenerates into a linear clipping normalized form. K_low The low energy storage threshold is selected, with a value ranging from 0.3° to 0.8°, preferably 0.5°; K_ high The high energy storage threshold is set to a value ranging from 2.0° to 4.0°, preferably 3.0°. The clip function is used here to restrict the calculation results to the interval [0,1]. The higher the local orientation difference KAM value, the larger the local energy storage factor f_KAM, indicating higher energy storage and stronger nucleation driving force.

[0047] In the high energy storage region, the dislocation density is close to the threshold of the driving force required for critical nucleation. A small increase in energy storage will cause a significant nonlinear increase in the nucleation probability. The exponential form has a larger slope at the high energy storage end in the normalization interval, which is consistent with the physical distribution law of the recrystallization nucleation driving force.

[0048] 2) Coincident location lattice grain boundary factor f_CSL The calculation process includes: Identify all CSL grain boundaries that conform to Brandon's criterion, including Σ3, Σ5, Σ9 and Σ11 grain boundaries, where the Brandon criterion tolerance angle is 15° / √Σ; For oriented silicon steel, calculate the distance d_CSL9(x,y) from each pixel to the nearest Σ9 grain boundary; Calculate the lattice grain boundary factor at the coincident site: f_CSL(x,y)=clip[1-d_CSL9(x,y) / d_CSL_crit,0,1] , Wherein, d_CSL_crit is the critical distance of the lattice grain boundary at the overlapping position, and its value ranges from 2.0 to 4.0 μm. In this embodiment, d_CSL_crit = 3.0 μm.

[0049] 3) Shear band factor f_shear : Shear bands are important nucleation sites for Goss grains. In this embodiment, the shear band factor... f_shear f_shear is a binary factor. For pixels located within the identified shear band region, f_shear = 1; for pixels located outside the shear band region, f_shear = 0.

[0050] 4) Large-angle grain boundary distance factor f_HAGB The specific calculation process is as follows: All grain boundaries with an orientation difference greater than 15° are extracted as large-angle grain boundaries; The distance from each pixel to the nearest large-angle grain boundary is calculated using Euclidean distance transformation. d_HAGB(x,y) ; The expression for calculating the large-angle grain boundary distance factor is: f_HAGB(x,y)=clip[1-d_HAGB(x,y) / d_crit,0,1] , in, d_crit The critical distance threshold for large-angle grain boundaries is d_crit = 2.0 μm in this embodiment. The distance from each pixel to the nearest HAGB is calculated using distance transformation; the closer the distance, the larger f_HAGB, indicating proximity to the grain boundary nucleation site. For oriented silicon steel, there is a special relationship between Σ9 grain boundaries and Goss orientation; regions closer to Σ9 grain boundaries have a higher tendency for Goss nucleation.

[0051] 5) Factor weights w 1 ~w 4 : In this embodiment, factor weights w 1 ~w 4 The initial values ​​were set to 0.35, 0.30, 0.25, and 0.10, respectively. The weights were set based on: local orientation difference energy storage as the primary thermodynamic driving force, the kinetic advantages of large-angle grain boundaries, the contribution of shear bands as special nucleation sites for Goss grains, and the auxiliary selective effect of Σ9 grain boundaries on Goss nucleation. Furthermore, based on different material systems and process conditions, including cold rolling reduction rate, initial grain size, and material composition, adjustments and optimizations are made to the initial values, specifically including: When the material type is pure iron or IF steel, the shear band factor weight w3 is reduced to 0.10~0.20, the local energy storage factor weight w1 is increased to 0.35~0.45 and the large angle grain boundary distance factor weight w2 is increased to 0.35~0.45, and the overlapping position lattice grain boundary factor weight w4 is reduced to 0.03~0.07. When the material type is non-oriented silicon steel, the weight of the large-angle grain boundary distance factor w2 is increased to 0.33~0.38, and the weight of the shear band factor w3 is decreased to 0.18~0.22; When the material type is low carbon steel, the weight of the large-angle grain boundary distance factor w2 is increased to 0.33~0.38, and the weight of the shear band factor w3 is decreased to 0.18~0.22; When the material type is austenitic stainless steel, the local energy storage factor weight w1 is reduced to 0.28~0.32, and the lattice grain boundary factor weight w4 at the overlapping position is increased to 0.13~0.18. When the cold rolling reduction rate is greater than 80%, the weight of the shear band factor w3 is further increased based on the adjustment according to the material type. When the initial grain size is greater than 100 μm, the weight of the large-angle grain boundary distance factor w2 is further increased based on the material type adjustment.

[0052] S6. Determine the preferred nucleation sites based on the scoring threshold, and output the spatial distribution map and statistical results of the nucleation sites.

[0053] In this embodiment, a scoring threshold (default 0.6) is set, and pixels with a score greater than the scoring threshold are marked as priority nucleation sites.

[0054] The output includes: nucleation score heatmap, binary distribution map of nucleation sites, and nucleation density statistics (numbers / mm²). 2 ), the area ratio of nucleation sites, and statistics of nucleation sites classified by orientation.

[0055] The classification and statistics of nucleation sites according to grain orientation specifically include: Based on the crystal orientation information of each pixel in the two-dimensional electron backscatter diffraction data, the predicted nucleation sites are classified according to the orientation of their parent grains. The number and density of nucleation sites within grains of different orientation types were statistically analyzed. Output the comparison results of nucleation tendencies of grains with different orientations to evaluate the evolution trend of recrystallization texture.

[0056] This embodiment also provides a recrystallization nucleation site prediction system based on a multi-factor scoring model, applying the above-described method, such as... Figure 4 As shown, the system includes: The data acquisition module is used to perform electron backscatter diffraction scanning on cold-deformed metal samples to acquire electron backscatter diffraction data containing crystal orientation information, and supports EBSD data files in .ctf and .ang formats; The local orientation difference calculation module is used to calculate the local orientation difference distribution field based on crystal orientation information; The grain boundary analysis module is used to extract grain boundary information and obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping locations; The shear band detection module is used to automatically identify shear bands based on the morphological characteristics and spatial distribution of the local orientation difference distribution field. The nucleation scoring module is used to perform multi-factor weighted scoring calculations and determine preferred nucleation sites based on scoring thresholds; The visualization module is used to display the distribution map of nucleation sites, the local orientation difference distribution field map, and the scoring heatmap; The report generation module is used to output analysis reports including nucleation density, nucleation site ratio, and orientation classification statistics.

[0057] To verify the effectiveness of the method of the present invention, an application verification was carried out using a cold-rolled sample of ultra-thin oriented silicon steel (0.08 mm thick).

[0058] Sample preparation: The finished grain-oriented silicon steel sheet was cold-rolled to a thickness of 0.08 mm, and 5×5 mm samples were cut and electrolytically polished.

[0059] EBSD testing: A Zeiss field emission scanning electron microscope equipped with an Oxford Symmetry EBSD probe was used with a scan step size of 0.2 μm and an accelerating voltage of 20 kV. Figure 5 The image shows the EBSD orientation of the cold-rolled sample, revealing typical rolling deformation microstructure.

[0060] Data analysis: The analysis is performed according to the method described in this embodiment. Figure 6 The diagram shows the distribution of KAM fields. High KAM areas (red) represent high energy storage areas, while low KAM areas (blue) contain potential shear zones. Figure 7 The data represents the distribution of grain boundary types; red indicates HAGB and blue indicates CSL grain boundaries. Figure 8 The shear band identification results show multiple shear bands at angles of 17° and 30° to the rolling direction. Figure 9 The results show the CSL grain boundary distribution. Figure 10 For nucleation score chart, Figure 11 This is a map showing the predicted distribution of nucleation sites based on default parameters and a scoring threshold of 0.6.

[0061] Statistical results: The predicted nucleation site area ratio was 8.43%, and the nucleation density was approximately 8470 nuclei / mm². 2 .like Figure 12 The figure shows the statistics of nucleation tendencies of grains with different orientations. The terms on the horizontal axis are abbreviations for common texture components in cold-rolled / hot-rolled metals, specifically: Goss is {110}. <001> Orientation, Cube (cubic texture) is {001} <100> Orientation, RotCube (rotational cube texture) is {001} <110> Orientation, Brass (brass texture) is {110} <112> Orientation, Copper (copper texture) is {112} <111> Orientation, Gamma1 is {111} <110> Orientation (belonging to γ ​​fiber texture), the tolerance angle for each orientation is set to 15°. Figure 12 In the middle, Goss{110} <001> The nucleation site density is highest within the oriented grains (3662.1 × 10⁻⁶). 3pcs / mm 2 The proportion of Goss grains in oriented silicon steel is significantly higher than in other orientations, which is consistent with the preferential nucleation mechanism of Goss grains in oriented silicon steel; {100} <011> Orientation is secondary (2011.7×10) 3 pcs / mm 2 ), {110} <112> and {111} <110> The nucleation densities of the oriented structures are similar. This result indicates that the proposed method can effectively identify preferential nucleation sites that favor the formation of Goss texture.

[0062] To verify the versatility of the method of the present invention in non-woven materials, this embodiment uses hot-rolled samples of industrial pure iron as the object for application verification.

[0063] Sample preparation: 5×5mm samples were cut from hot-rolled industrial pure iron (Fe content ≥ 99.5%) and electrolytically polished. Compared with grain-oriented silicon steel, pure iron has a higher stacking fault energy, and its deformation mechanism is mainly dislocation slip, while shear band development is relatively weak.

[0064] EBSD testing: scan step size 0.3 μm, accelerating voltage 20 kV. The microstructure of cold-rolled pure iron exhibits typical fibrous deformed grains.

[0065] Parameter adjustment: The model parameters were adjusted to suit the characteristics of pure iron: (1) Since the shear band of pure iron is not as developed as that of silicon steel, w3(f_shear) was reduced from 0.25 to 0.15; (2) The weight of KAM energy storage factor w1 was increased to 0.40 accordingly; (3) The correlation between CSL grain boundaries and specific textures in pure iron is not as obvious as that in silicon steel, so w4 was reduced to 0.05; (4) The weight of large-angle grain boundary distance factor w2 was increased to 0.40 to reflect the important role of grain boundaries in the nucleation of pure iron. Figure 13 Score the nucleus.

[0066] The analysis results showed that the shear band identification result was 0, indicating that the number and development level of shear bands in pure iron were significantly lower compared with oriented silicon steel.

[0067] Statistical results: The predicted nucleation site area ratio was 11.20%, and the nucleation density was approximately 7.8 × 10⁻⁶. 4 pcs / mm 2 .like Figure 14The figure shows the predicted nucleation sites and nucleation density statistics for hot-rolled pure iron with different orientations. The terms on the horizontal axis are abbreviations for common texture components in cold-rolled / hot-rolled metals, specifically: Goss is {110}. <001> Orientation, Cube (cubic texture) is {001} <100> Orientation, RotCube (rotational cube texture) is {001} <110> Orientation, Brass (brass texture) is {110} <112> Orientation, Copper (copper texture) is {112} <111> Orientation, Gamma1 is {111} <110> Orientation (belonging to γ ​​fiber texture), the tolerance angle for each orientation is set to 15°. Figure 14 In the middle, pure iron {111} <110> The nucleation site density is highest in oriented grains (149.7 × 10⁻⁶). 3 pcs / mm 2 ), {100} <011> Orientation is secondary (122.8×10) 3 pcs / mm 2 This is consistent with the mechanism of preferential nucleation of γ-fiber and α-fiber textures during the recrystallization of pure iron. It is noteworthy that, unlike the preferential nucleation of Goss orientation in grain-oriented silicon steel, Goss{110} in pure iron… <001> The nucleation density of oriented nuclei was actually the lowest (44.6 × 10⁻⁶). 3 pcs / mm 2 This is because shear bands are underdeveloped in hot-rolled pure iron (shear band factor f_shear≈0), and nucleation mainly depends on the KAM energy storage factor and grain boundary distance factor. This embodiment verifies the universality and adaptability of the method of the present invention to different material systems by adjusting the weight configuration.

[0068] Table 1 Summary of parameters for the multi-factor scoring model

[0069] Table 2 Recommended weighting for different material types

[0070] In summary, the recrystallization nucleation site prediction method based on a multi-factor scoring model provided by this invention comprehensively considers multiple factors such as KAM energy storage, grain boundary type, shear band distribution, and CSL grain boundaries, achieving automated and quantitative prediction of nucleation sites. Experimental verification shows that the method of this invention is not only applicable to functional materials with high texture control requirements, such as oriented silicon steel, but can also be applied to various deformable metal materials such as pure iron, IF steel, low-carbon steel, and stainless steel by adjusting the weighting coefficients, demonstrating good versatility and practical value.

[0071] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any person skilled in the art can easily conceive of various equivalent modifications or substitutions within the technical scope disclosed in the present invention, and these modifications or substitutions should all be covered within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for predicting recrystallization nucleation sites based on a multi-factor scoring model, characterized in that, include: Electron backscatter diffraction scanning was performed on cold-deformed metal samples to obtain electron backscatter diffraction data containing crystal orientation information. Calculate the local orientation difference distribution field based on electron backscatter diffraction data; Extract grain boundary information to obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping locations; Based on the morphological characteristics and spatial distribution of the local orientation difference distribution field, shear bands are automatically identified; A multi-factor weighted scoring model is established to score the nucleation tendency of each pixel. The multi-factors include local energy storage factor, large-angle grain boundary distance factor, shear band factor, and coincident position lattice grain boundary factor. Based on the scoring threshold, priority nucleation sites are determined, and the spatial distribution map and statistical results of the nucleation sites are output. The local energy storage factor is calculated using the following expression: f_KAM(x,y)=clip{[exp(α·(KAM(x,y)-K_low) / (K_high-K_low))-1] / (exp(α)-1),0, 1} ， in, f_KAM(x,y) Represents pixels in two-dimensional electron backscattering diffraction data (x,y) The corresponding local energy storage factor; KAM(x,y) Pixels in two-dimensional electron backscatter diffraction data (x,y) The corresponding local orientation difference; α is a nonlinear shape parameter used to control the degree of nonlinearity of the contribution of energy storage to the nucleation driving force; K_low Low energy storage threshold; K_high The high energy storage threshold is defined by the `clip` function, which is a clipping function. The calculation process for the large-angle grain boundary distance factor is as follows: All grain boundaries with an orientation difference greater than 15° are extracted as large-angle grain boundaries; The distance from each pixel to the nearest large-angle grain boundary is calculated using Euclidean distance transformation. d_HAGB(x,y) ; The expression for calculating the large-angle grain boundary distance factor is: f_HAGB(x,y)=clip[1-d_HAGB(x,y) / d_crit,0,1] ， Where d_crit is the critical distance threshold for large-angle grain boundaries; The process of identifying the shear band includes the following steps: Extracting local orientation differences less than the low energy storage threshold K_low The region is selected as the candidate shear band region; Connectivity labeling and morphological analysis are performed on candidate shear band regions to filter out band regions with an aspect ratio greater than a set value. Calculate the angle between the main axis direction of the strip region and the rolling direction, and retain the region where the angle is within a specific range. The specific range includes two angle intervals: 12°-22° and 25°-35°, which correspond to the type A shear strip with an angle of 17° to the rolling direction and the type B shear strip with an angle of 30° to the rolling direction in cold-rolled iron-based alloys. Regions that satisfy the condition that the mean local orientation difference within the strip region is lower than the low energy storage threshold K_low and the mean local orientation difference between the adjacent regions on both sides of the strip region is higher than the high energy storage threshold K_high are identified as shear zones. The shear band factor is a binary factor. For pixels located within the identified shear band region, the shear band factor is set to 1, and for pixels located outside the shear band region, the shear band factor is set to 0. The calculation process for the grain boundary factor at the overlapping location lattice includes: Identify all CSL grain boundaries that conform to Brandon's criterion, including Σ3, Σ5, Σ9 and Σ11 grain boundaries, where the Brandon criterion tolerance angle is 15° / √Σ; For oriented silicon steel, calculate the distance d_CSL9(x,y) from each pixel to the nearest Σ9 grain boundary; Calculate the lattice grain boundary factor at the coincident site: f_CSL(x,y)=clip[1-d_CSL9(x,y) / d_CSL_crit,0,1] ， Where d_CSL_crit is the critical distance of the lattice grain boundary at the overlapping position; The multi-factor weighted scoring model scores the nucleation tendency of each pixel: Score(x,y)=w 1 ×f_KAM+w 2 ×f_HAGB+w 3 ×f_shear+w 4 ×f_CSL ， in, f_KAM As a local energy storage factor, f_HAGB For large-angle grain boundary distance factors, f_shear For shear band factor, f_CSL For the lattice grain boundary factor at the coincident position; w 1 ~w 4 These are the corresponding factor weights, satisfying... w 1 + w 2 +w 3 +w 4 =1; Based on different material systems and process conditions, adjustments and optimizations are made to the initial values.

2. The method for predicting recrystallization nucleation sites based on a multi-factor scoring model according to claim 1, characterized in that, The calculation of the local orientation difference distribution field based on electron backscattering diffraction data specifically includes: For each pixel in the electron backscatter diffraction data, calculate its orientation difference with all pixels in its surrounding neighborhood, and take the average value as the local orientation difference of the current point. The neighborhood size is a 5×5 pixel window. The maximum orientation difference threshold is set to 5°. Adjacent points exceeding the preset maximum orientation difference threshold are filtered out to obtain the local orientation difference distribution field.

3. The method for predicting recrystallization nucleation sites based on a multi-factor scoring model according to claim 1, characterized in that, Based on the cold rolling reduction rate, initial grain size, and material composition, adjustments and optimizations are made to the initial values, specifically including: When the material type is pure iron or IF steel, reduce the weight of the shear band factor w3, correspondingly increase the weight of the local energy storage factor w1 and the weight of the large angle grain boundary distance factor w2, and reduce the weight of the grain boundary factor w4 at the overlapping position lattice. When the material type is non-oriented silicon steel, increase the weight of the large-angle grain boundary distance factor w2 and decrease the weight of the shear band factor w3. When the material type is low carbon steel, increase the weight of the large-angle grain boundary distance factor w2 and decrease the weight of the shear band factor w3; When the material type is austenitic stainless steel, reduce the weight of the local energy storage factor w1 and increase the weight of the grain boundary factor w4 at the overlapping position lattice. When the cold rolling reduction rate is greater than the set value, the weight of the shear band factor w3 is further increased based on the adjustment according to the material type. When the initial grain size is greater than the set value, the weight of the large-angle grain boundary distance factor w2 is further increased based on the adjustment according to the material type.

4. The method for predicting recrystallization nucleation sites based on a multi-factor scoring model according to claim 1, characterized in that, It also includes the classification and statistics of nucleation sites according to grain orientation, specifically including: Based on the crystal orientation information of each pixel in the two-dimensional electron backscatter diffraction data, the predicted nucleation sites are classified according to the orientation of their parent grains. The number and density of nucleation sites within grains of different orientation types were statistically analyzed. Output the comparison results of nucleation tendencies of grains with different orientations to evaluate the evolution trend of recrystallization texture.

5. A recrystallization nucleation site prediction system based on a multi-factor scoring model, characterized in that, The system comprising the method of claim 1, wherein the method of claim 1 is: The data acquisition module is used to perform electron backscatter diffraction scanning on the scanned cold-deformed metal sample to acquire electron backscatter diffraction data containing crystal orientation information. The local orientation difference calculation module is used to calculate the local orientation difference distribution field based on crystal orientation information; The grain boundary analysis module is used to extract grain boundary information and obtain the spatial distribution of large-angle grain boundaries and lattice grain boundaries at overlapping locations; The shear band detection module is used to automatically identify shear bands based on the morphological characteristics and spatial distribution of the local orientation difference distribution field. The nucleation scoring module is used to perform multi-factor weighted scoring calculations and determine preferred nucleation sites based on scoring thresholds; The visualization module is used to display the distribution map of nucleation sites, the local orientation difference distribution field map, and the scoring heatmap; The report generation module is used to output analysis reports including nucleation density, nucleation site ratio, and orientation classification statistics.

Images