Component design method for titanium diboride reinforced aluminum-silicon-based composite material

Abstract

The invention provides a titanium diboride reinforced aluminum-silicon-based composite material component design method, which comprises the following steps: based on a high-flux first principle calculation method, establishing a view of regulating and controlling the dispersibility of titanium diboride particles in an aluminum-silicon alloy matrix by interface modification elements for evaluating the dispersing capacity of alloy elements on the titanium diboride particles; on the basis of a theoretical interface modification element distribution picture, alloying elements which are beneficial to dispersion of titanium diboride reinforced particles in an aluminum-silicon matrix and can regulate and control the strengthening and toughening capacity of the aluminum-silicon alloy are determined, and a self-adaptive learning optimization method based on machine learning is used for achieving quantitative design of aluminum-silicon matrix components in the composite material. And titanium diboride particles are added into the obtained optimized matrix alloy components, and the target composite material is obtained. The invention provides an efficient new thought for designing the particle reinforced metal-based composite material, and rapid design of the high-toughness titanium diboride reinforced aluminum-silicon-based composite material can be realized.

Description

Technical Field

[0001] The present invention relates to the technical field of composite material design, and in particular, to a method for designing the composition of a titanium diboride reinforced aluminum-silicon matrix composite material. Background Art

[0002] Titanium diboride (TiB2) ceramic particles are ceramic particles with high stiffness and high strength. By introducing them into aluminum alloy to prepare TiB2 particle-reinforced aluminum-silicon matrix composite materials, the strength performance can be greatly improved, while the plasticity is significantly reduced. The fundamental reason for this phenomenon is that the interfacial energy between TiB2 particles and the aluminum matrix is relatively high, which easily leads to particle agglomeration. In the aluminum matrix composite material containing particle agglomeration, stress concentration is likely to occur around the agglomerates during the loading process, resulting in the initiation of cracks and material failure. Numerous studies have found that modifying the interfacial alloy elements can reduce the interfacial energy between TiB2 particles and the aluminum matrix, thereby realizing the effective dispersion of the reinforcing particles and simultaneously improving the strength and plasticity of the composite material. At the same time, by selecting and adding alloying elements to affect the characteristics of the alloy matrix, the strength and plasticity of the aluminum matrix composite material can also be regulated. However, the traditional design method mainly relies on the empirical trial-and-error method and lacks a multi-objective material design criterion based on the synergistic regulation of the dispersion of reinforcing particles and the strengthening and toughening of the matrix, resulting in difficulty in achieving a breakthrough in the strength-plasticity matching of the material.

[0003] In recent years, the development of machine learning technology has had a huge impact on the research in the field of metal material design technology. Through machine learning methods, researchers can fully explore the internal relationship between "composition / process - properties" of materials, and conduct forward or reverse design of materials, greatly reducing the experimental cost. However, due to the relatively small amount of relevant research, if machine learning development is directly carried out for TiB2 particle-reinforced aluminum-silicon matrix composite materials, it will be difficult to perform model fitting due to the small amount of data. Considering that the TiB2 particle-reinforced aluminum-silicon matrix composite material is composed of TiB2 particles and an alloy matrix, and the amount of data of the alloy matrix is relatively sufficient. Therefore, based on the pre-designed evaluation of the ability of interfacial modification elements to regulate the dispersion of TiB2 particles in the aluminum-silicon alloy matrix, and then comprehensively considering the strengthening and toughening ability of the alloying elements themselves for the composition design of the alloy matrix, the development of TiB2 particle-reinforced aluminum-silicon matrix composite materials can be realized. Summary of the Invention

[0004] Aiming at the defects in the prior art, the purpose of the present invention is to provide a method for designing the composition of a titanium diboride reinforced aluminum-silicon matrix composite material, which solves the problem that machine learning methods cannot be used for the composition design of titanium diboride particle-reinforced aluminum-silicon matrix composite materials due to the small amount of data, and proposes a new idea for the efficient design of particle-reinforced metal matrix composite materials, which can realize the rapid design of high-strength and high-toughness titanium diboride reinforced aluminum-silicon matrix composite materials.

[0005] To solve the above problems, the technical solution of the present invention is as follows:

[0006] A method for designing the composition of a titanium diboride reinforced aluminum-silicon matrix composite, comprising the following steps:

[0007] Based on the high-throughput first-principles calculation method, establish a picture of the dispersion of titanium diboride particles in the aluminum-silicon alloy matrix regulated by interface modification elements, and use it to evaluate the dispersion ability of alloying elements on titanium diboride particles;

[0008] Based on the theoretical picture of the distribution of interface modification elements, determine the alloying elements that are beneficial to the dispersion of titanium diboride reinforcing particles in the aluminum-silicon matrix and can also regulate the strengthening and toughening ability of the aluminum-silicon alloy. Use the adaptive learning optimization method based on machine learning to realize the quantitative design of the aluminum-silicon matrix composition in the composite material;

[0009] Add titanium diboride particles to the obtained optimized matrix alloy composition to obtain the target composite material.

[0010] Preferably, the step of establishing a picture of the dispersion of titanium diboride particles in the aluminum-silicon alloy matrix regulated by interface modification elements based on the high-throughput first-principles calculation method and using it to evaluate the dispersion ability of alloying elements on titanium diboride particles specifically includes:

[0011] Construct an initial Si-Al(111) / TiB2(0001) interface model;

[0012] In the initial Si-Al(111) / TiB2(0001) interface model, select the modification element X Ⅱ , and perform the second modification doping of X Ⅱ -Si-Al(111) / TiB2(0001) interface for high-throughput calculation of relative formation energy;

[0013] For each interface modification element X Ⅱ , select the X Ⅱ -Si-Al(111) / TiB2(0001) interface model with the lowest relative formation energy to obtain the relative interface formation energy and relative adhesion work;

[0014] Taking the relative interface formation energy as the horizontal axis and the relative adhesion work as the vertical axis, establish a picture of the dispersion of TiB2 reinforcing particles in the aluminum-silicon matrix regulated by interface modification elements in the composite material, and evaluate the particle dispersion ability of the interface modification elements.

[0015] Preferably, the constructed initial Si-Al(111) / TiB2(0001) interface model is composed of two 6-layer Al(111) surface models sandwiching a 9-layer TiB2(0001) surface model. This interface model contains two interfaces and adopts a Ti-centered stacking mode. The calculation formula for the relative interface formation energy of the Si-doped Si-Al(111) / TiB2(0001) is as follows:

[0016]

[0017] Among them, ΔE f0 is the relative interface formation energy; N Total is the total number of atoms in the interface model; and are the energies of the Si-doped interface and the initial interface respectively; and are the average atomic energies of element Si and Al in the bulk structure respectively.

[0018] Preferably, the calculation formula for the relative interface formation energy of the second modified doped X Ⅱ -Si-Al(111) / TiB2(0001) interface is as follows:

[0019]

[0020] Among them, ΔE f is the relative interface formation energy; N Total is the total number of atoms in the interface model; and are the energies of the doped interface and the initial interface respectively; and are the average atomic energies of alloy element X Ⅱ and Al in the bulk structure respectively.

[0021] Preferably, the calculation formula for the relative adhesion work of the second modified doped X Ⅱ -Si-Al(111) / TiB2(0001) interface is as follows:

[0022]

[0023] Among them, is the adhesion work of the X Ⅱ -Si-Al(111) / TiB2(0001) interface model; and are obtained by separately calculating the single-point energies after separating the optimized X Ⅱ -Si-Al(111) / TiB2(0001) interface model into an Al surface and a TiB2 surface along the interface. is X Ⅱ - The total energy of the Si-Al(111) / TiB2(0001) interface model; is the adhesion work of the Si-Al(111) / TiB2(0001) interface model; and is obtained by performing single-point energy calculations on the optimized Si-Al(111) / TiB2(0001) interface model after separating it into an Al surface and a TiB2 surface along the interface; is the total energy of the Si-Al(111) / TiB2(0001) interface model; ΔW ad is the relative adhesion work.

[0024] Preferably, based on the theoretical interface modification element distribution pattern, determining alloying elements that are beneficial to the dispersion of titanium diboride reinforcing particles in the aluminum-silicon matrix and can also regulate the strengthening and toughening ability of the aluminum-silicon alloy, and using an adaptive learning optimization method based on machine learning to achieve the quantitative design of the aluminum-silicon matrix composition in the composite material, specifically including:

[0025] Establish a "composition / process - property" database based on the historical data information of aluminum-silicon alloys under the target process;

[0026] Use a machine learning method based on the characteristics of alloying element types to construct a machine learning relationship model of "composition / process - property", and combine the distributed SHAP analysis method to evaluate the strengthening and toughening ability of alloying elements themselves;

[0027] Use a machine learning method based on the physical and chemical characteristics of alloying elements, and through a three-step feature engineering method, construct a "composition / process - property" machine learning model with high prediction accuracy;

[0028] Use an adaptive learning optimization strategy to determine the multi-objective expected improvement values of yield strength and elongation in the composition search space, and through multiple calculations and combined with experiments for iterative optimization, obtain the optimized matrix alloy composition corresponding to the optimal performance.

[0029] Preferably, the specific steps of the distributed SHAP analysis method are: list the SHAP values corresponding to all feature values; when there are several samples with the same feature values, take the average of their SHAP values as the SHAP value under this feature value; use the feature value as the abscissa and the SHAP value as the ordinate to draw a bar chart.

[0030] Preferably, the machine learning method based on the physical and chemical characteristics of alloying elements is specifically: select 55 common physical and chemical characteristics of each alloying element, and convert them into 110 alloy factor characteristics according to the following formula:

[0031] fmi = ∑(f ij ×α j ) / ∑α j

[0032] f vi = ∑[(f ij - f mi ) 2 ×a j ) / ∑a j

[0033] where f mi and f vi are the average factor and variance factor of the i-th physicochemical characteristic respectively; α j represents the atomic ratio of the j-th alloying element.

[0034] Preferably, the three-step feature engineering method specifically includes: correlation screening, recursive elimination, and exhaustive screening.

[0035] Preferably, the specific calculation formula of the multi-objective expected improvement MOEI is:

[0036]

[0037] z = (μ - μ * ) / σ

[0038] MOEI = EI YS × EI EL

[0039] where EI YS / EL represents the expected improvement value of yield strength or elongation; (z) and φ(z) are the probability density function and cumulative distribution function respectively; μ * is the reference value, set to 140 MPa for YS and 9% for EL; σ and μ are the uncertainty (RMSE value) and average predicted value calculated by the bootstrap resampling method respectively.

[0040] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0041] 1. The present invention provides a high-throughput first-principles calculation method for the Al / TiB2 interface model. By pre-doping the solid solution element Si in the interface model, it can more realistically compare the ability of different interface modification elements (X II ) to modify the Al / TiB2 interface containing the solid solution Si element, and guide the screening of alloying elements beneficial to the dispersion of TiB2 particles.

[0042] 2. The present invention uses two feature methods for machine learning, namely, a machine learning method based on the types of alloying elements combined with the distributed SHAP analysis method to obtain the strengthening and toughening ability of the alloying elements themselves, and a machine learning method based on the physical and chemical characteristics of the alloying elements combined with the three-step feature engineering method to construct a machine learning model with high prediction accuracy.

[0043] 3. The present invention introduces a machine learning method into the design of TiB2-reinforced aluminum-silicon matrix composites. By mainly designing the composition of the alloy matrix and comprehensively considering the strengthening and toughening ability of the alloying elements themselves and their dispersing ability for TiB2 particles, the efficient development of TiB2-reinforced aluminum-silicon matrix composites is realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] By reading the following detailed description of non-limiting embodiments with reference to the accompanying drawings, other features, objects, and advantages of the present invention will become more apparent:

[0045] Figure 1 It is a flow chart of the composition design method for TiB2-reinforced aluminum-silicon matrix composites of the present invention;

[0046] Figure 2 It is a flow chart of the high-throughput first-principles calculation of the present invention;

[0047] Figure 3 It is a graph of the distributed SHAP analysis results of the yield strength for different features of the present invention;

[0048] Figure 4 It is a specific step diagram of the three-step feature engineering method of the present invention;

[0049] Figure 5 It is a comparison graph of the properties of the alloy matrix and composites prepared by the present invention with the remaining samples in the database. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0050] The present invention will be described in detail below with reference to specific embodiments. The following embodiments will help those skilled in the art to further understand the present invention, but do not limit the present invention in any form. It should be noted that those of ordinary skill in the art can make several changes and improvements without departing from the concept of the present invention. These all belong to the protection scope of the present invention.

[0051] Specifically, the present invention provides a method for designing the composition of TiB2-reinforced aluminum-silicon matrix composites, as Figure 1 shown, the method includes the following steps:

[0052] S1: Based on the high-throughput first-principles calculation method, establish a picture of the dispersion of interface modification elements regulating TiB2 particles in the aluminum-silicon alloy matrix to evaluate the dispersion ability of alloying elements for TiB2 particles;

[0053] Specifically, step S1 includes the following steps:

[0054] S11: Construct an initial Si-Al(111) / TiB2(0001) interface model;

[0055] Construct an Al(111) / TiB2(0001) interface model with the best interface coherency in the titanium diboride (TiB2) and aluminum (Al) system. For the target aluminum-silicon alloy matrix system, pre-dope silicon (Si) in the Al layer, select different doping sites, and calculate the relative interface formation energy after doping with the corresponding Si element. After comparison, the model with the lowest relative interface formation energy is selected as the initial Si-Al(111) / TiB2(0001) interface model;

[0056] In this step, the specifically constructed Si-Al(111) / TiB2(0001) interface model is as follows: The constructed Si-Al(111) / TiB2(0001) interface model is in a "sandwich" shape, composed of two 6-layer Al(111) surface models sandwiching a 9-layer TiB2(0001) surface model. This interface model contains two interfaces and adopts a Ti-centered stacking mode.

[0057] The calculation formula for the relative interface formation energy of the Si-Al(111) / TiB2(0001) doped with the Si element is as follows:

[0058]

[0059] where ΔE f0 is the relative interface formation energy; N Total is the total number of atoms in the interface model; and are the energies of the Si element doped interface and the initial interface respectively; and are the average atomic energies of the elements Si and Al in the bulk structure respectively.

[0060] S12: In the initial Si-Al(111) / TiB2(0001) interface model, select a modification element (X Ⅱ ), and perform high-throughput calculation of the relative formation energy of the second modified doped X Ⅱ -Si-Al(111) / TiB2(0001) interface;

[0061] In this step, Figure 2It is the high-throughput first-principles calculation flow chart of the present invention. The specific steps include: rewriting the POSCAR file according to the doping behavior and inputting it into the VASP software; extracting key information from the result file for convergence judgment; storing the target information in the material database; data processing and visualization of the results.

[0062] In this step, the second modification doping X Ⅱ - The calculation formula for the relative interface formation energy of the Si-Al(111) / TiB2(0001) interface is as follows:

[0063]

[0064] Among them, ΔE f is the relative interface formation energy; N Total is the total number of atoms in the interface model; and are the energies of the doped interface and the initial interface respectively; and are the average atomic energies of alloy element X Ⅱ and Al in the bulk structure respectively.

[0065] S13: For each interface modification element (X Ⅱ ), select the X Ⅱ

[0066] -Si-Al(111) / TiB2(0001) interface model with the lowest relative formation energy to obtain the relative interface formation energy and the relative adhesion work;

[0067] In this step, the second modification doping X Ⅱ - The calculation formula for the relative adhesion work of the Si-Al(111) / TiB2(0001) interface is as follows:

[0068]

[0069] Among them, is the adhesion work of the X Ⅱ -Si-Al(111) / TiB2(0001) interface model; and are obtained by separately calculating the single-point energies after separating the optimized X Ⅱ -Si-Al(111) / TiB2(0001) interface model into the Al surface and the TiB2 surface along the interface; is the total energy of the X Ⅱ -Si-Al(111) / TiB2(0001) interface model; is the adhesion work of the Si-Al(111) / TiB2(0001) interface model; and It is obtained by performing single-point energy calculations on the optimized Si-Al(111) / TiB2(0001) interface model after separating it into the Al surface and the TiB2 surface along the interface respectively; is the total energy of the Si-Al(111) / TiB2(0001) interface model; ΔW ad is the relative adhesion work.

[0070] S14: Taking the relative interface formation energy as the horizontal axis and the relative adhesion work as the vertical axis, establish a picture of the dispersion of TiB2 reinforcing particles in the aluminum-silicon matrix in the composite material regulated by interface modification elements, and evaluate the particle dispersion ability of the interface modification elements.

[0071] S2: Based on the theoretical picture of the distribution of interface modification elements, determine alloying elements that are both beneficial to the dispersion of titanium diboride reinforcing particles in the aluminum-silicon matrix and can regulate the strengthening and toughening ability of the aluminum-silicon alloy, and use an adaptive learning optimization method based on machine learning to realize the quantitative design of the composition of the aluminum-silicon matrix in the composite material;

[0072] Specifically, the step S2 includes the following steps:

[0073] S21: Establish a "composition / process - property" database according to the historical data information of aluminum-silicon alloy under the target process;

[0074] S22: Use a machine learning method based on the type characteristics of alloying elements to construct a machine learning relationship model of "composition / process - property", and combine the distributed SHAP analysis method to evaluate the strengthening and toughening ability of the alloying elements themselves;

[0075] Specifically, use a machine learning method based on the type characteristics of alloying elements to construct a relationship of "composition / process - property", and combine the distributed SHAP analysis method to evaluate the strengthening and toughening ability of the alloying elements themselves. This method directly uses the features in the database, and respectively uses seven machine learning models of linear regression (LIN), polynomial regression (POLY), random forest regression (RF), extreme gradient boosting regression (XGB), support vector machine regression with linear kernel (SVR.L), support vector machine regression with polynomial kernel (SVR.P), and support vector machine regression with Gaussian radial basis kernel (SVR.R) for fitting, and selects the model with the highest fitting accuracy through the cross-validation method. Based on this machine learning model, list the SHAP values corresponding to all feature values; when the feature values of several samples are the same, take the average of their SHAP values as the SHAP value under this feature value; take the feature value as the abscissa and the SHAP value as the ordinate, and draw a bar chart. According to this bar chart, evaluate the strengthening and toughening ability of the alloying elements.

[0076] Figure 3It is a distributed SHAP analysis result graph of different features on the yield strength, which can reflect the contribution of different features to the target performance at different feature values.

[0077] S23: Use a machine learning method based on the physicochemical characteristics of alloying elements, and through a three-step feature engineering method, construct a "composition / process - performance" machine learning model with high prediction accuracy;

[0078] In this step, a machine learning method based on the physicochemical characteristics of alloying elements is used. Specifically, 55 common physicochemical characteristics of each alloying element are selected and transformed into 110 alloy factor characteristics according to the following formula:

[0079] f mi =∑(f ij ×α j ) / ∑α j

[0080] f vi =∑[(f ij -f mi ) 2 ×a j / ∑a j

[0081] Among them, f mi and f vi are the average factor and variance factor of the i-th physicochemical characteristic (both belong to alloy factors); α j represents the atomic ratio of the j-th alloying element.

[0082] In this step, the three-step feature engineering methods used are correlation screening, recursive elimination, and exhaustive method screening.

[0083] Figure 4It is a specific step diagram of the three-step feature engineering method. In the correlation screening, first calculate the Pearson correlation coefficient and feature importance. When the Pearson correlation coefficient value between two features is greater than or equal to 0.95, delete the feature with lower feature importance, and update the Pearson correlation coefficient and feature importance. Repeat this step until there are no two highly correlated features in the feature pool. In the recursive elimination, assume that there are n features in the current feature pool. First, take out one feature from the feature pool, then perform machine learning fitting on the remaining n - 1 features, and calculate the fitting error using the cross-validation method. Then, put this feature back into the feature pool, select another different feature and calculate the fitting error of the remaining features in the same way until all features in the feature pool have been taken out. Therefore, delete the feature corresponding to the smallest fitting error. Keep performing this step until the fitting error reaches the minimum extreme value. In the exhaustive search screening, calculate the fitting error for all feature combinations using the cross-validation method, and select the feature combination with the smallest fitting error.

[0084] S24: Use the adaptive learning optimization strategy to determine the multi-objective expected improvement values of yield strength and elongation in the composition search space, and carry out iterative optimization through multiple calculations and combined with experiments to obtain the optimized matrix alloy composition corresponding to the optimal performance.

[0085] In this step, the multi-objective expected improvement (MOEI), the specific calculation formula is:

[0086]

[0087] z = (μ - μ * ) / σ

[0088] MOEI = EI YS × EI EL

[0089] where EI YS / EL represents the expected improvement value of yield strength or elongation; (z) and φ(z) are the probability density function and cumulative distribution function respectively; μ * is the benchmark value, set to 140 MPa for YS and 9% for EL; σ and μ are the uncertainty (RMSE value) and average predicted value calculated by the bootstrap resampling method respectively.

[0090] The matrix alloy composition corresponding to the optimal performance is obtained through multiple iterations of optimization, specifically: in each global optimization process, the first three component points with the highest MOEI values are selected for experimental verification, and their strength and elongation are measured. The three samples measured in the experiment are put back into the dataset for re-training of the machine learning model and calculation of the MOEI value, so as to generate a series of candidate components. When the material performance at the nth iteration meets the set requirements and the material performance obtained after the (n + 1)th iteration does not increase compared with the nth iteration, the material composition at the nth iteration is selected as the matrix alloy composition.

[0091] S3: Titanium diboride particles are added to the obtained optimized matrix alloy composition to obtain the target composite material.

[0092] Specifically, Figure 5 This is a comparison chart of the performance of the alloy matrix and composite material prepared in this step with the rest of the samples in the database. As Figure 5 shown, the comprehensive mechanical properties of the alloy matrix are in a relatively balanced position, and the comprehensive mechanical properties of the composite material have reached a relatively high level.

[0093] The specific embodiments of the present invention have been described above. It should be understood that the present invention is not limited to the above specific embodiments, and those skilled in the art can make various changes or modifications within the scope of the claims, which do not affect the essence of the present invention. Without conflict, the embodiments of the present application and the features in the embodiments can be combined arbitrarily with each other.

Claims

1. A method for designing the composition of a titanium diboride reinforced aluminum-silicon matrix composite, characterized in that The method includes the following steps: Based on the high-throughput first-principles calculation method, establish a picture of the dispersion of TiB2 particles in the Al-Si alloy matrix regulated by interface modification elements, which is used to evaluate the dispersion ability of alloy elements on TiB2 particles; Based on the theoretical picture of interface modification element distribution, determine alloying elements that are beneficial to the dispersion of TiB2 reinforcing particles in the Al-Si matrix and can also regulate the strengthening and toughening ability of the Al-Si alloy. Use the adaptive learning optimization method based on machine learning to achieve the quantitative design of the composition of the Al-Si matrix in the composite material; Add TiB2 particles to the obtained optimized matrix alloy composition to obtain the target composite material.

2. The composition design method of the titanium diboride reinforced aluminum-silicon matrix composite material according to claim 1, characterized in that The step of establishing a picture of the dispersion of TiB2 particles in the Al-Si alloy matrix regulated by interface modification elements based on the high-throughput first-principles calculation method, which is used to evaluate the dispersion ability of alloy elements on TiB2 particles, specifically includes: Construct an initial Si-Al(111) / TiB2(0001) interface model; In the initial Si-Al(111) / TiB2(0001) interface model, the modified element X Ⅱ , and then doped with X for the second time Ⅱ - High-throughput calculation of the relative formation energy of the Si-Al(111) / TiB2(0001) interface; For each interface modification element X Ⅱ , select the X with the lowest relative formation energy Ⅱ -Si-Al(111) / TiB2(0001) interface model to obtain the relative interface formation energy and relative adhesion work; Taking the relative interface formation energy as the horizontal axis and the relative adhesion work as the vertical axis, establish a picture of the dispersion of TiB2 reinforcing particles in the Al-Si matrix in the composite material regulated by interface modification elements, and evaluate the particle dispersion ability of the interface modification elements.

3. The composition design method of the titanium diboride reinforced aluminum-silicon matrix composite material according to claim 2, characterized in that, The constructed initial Si-Al(111) / TiB2(0001) interface model is composed of two 6-layer Al(111) surface models sandwiching a 9-layer TiB2(0001) surface model. This interface model contains two interfaces and adopts the Ti-centered stacking mode; The calculation formula for the relative interface formation energy of the Si-Al(111) / TiB2(0001) doped with Si element is: Among them, ΔE f0 is the relative interface formation energy; N Total is the total number of atoms in the interface model; and are the energies of the Si element-doped interface and the initial interface, respectively; and are the average atomic energies of elements Si and Al in the bulk structure, respectively.

4. The composition design method of titanium diboride reinforced aluminum-silicon matrix composite according to claim 2, characterized in that, The second modified doped X Ⅱ - The calculation formula for the interfacial formation energy of the Si-Al(111) / TiB2(0001) interface is as follows: Among them, ΔE f is the relative interfacial formation energy; N Total is the total number of atoms in the interface model; and are the energies of the doped interface and the initial interface, respectively; and are the average atomic energies of alloying element X Ⅱ and Al in the bulk structure, respectively.

5. The composition design method of titanium diboride reinforced aluminum-silicon matrix composite according to claim 2, characterized in that The second modified doped X Ⅱ - The calculation formula for the relative adhesion work of the Si-Al(111) / TiB2(0001) interface is as follows: Among them, is the work of adhesion of the X Ⅱ -Si-Al(111) / TiB2(0001) interface model; and are obtained by calculating the single-point energies of the Al surface and the TiB2 surface respectively after separating the optimized X Ⅱ -Si-Al(111) / TiB2(0001) interface model along the interface; is the total energy of the X Ⅱ -Si-Al(111) / TiB2(0001) interface model; is the work of adhesion of the Si-Al(111) / TiB2(0001) interface model; and are obtained by calculating the single-point energies of the Al surface and the TiB2 surface respectively after separating the optimized Si-Al(111) / TiB2(0001) interface model along the interface; is the total energy of the Si-Al(111) / TiB2(0001) interface model; ΔW ad is the relative work of adhesion.

6. The composition design method of titanium diboride reinforced aluminum-silicon matrix composite according to claim 1, characterized in that, The step of determining alloying elements that are beneficial to the dispersion of TiB2 reinforcing particles in the Al-Si matrix and can also regulate the strengthening and toughening ability of the Al-Si alloy based on the theoretical picture of interface modification element distribution, and using the adaptive learning optimization method based on machine learning to achieve the quantitative design of the composition of the Al-Si matrix in the composite material, specifically includes: Establish a "composition / process-property" database based on the historical data information of the Al-Si alloy under the target process; Use a machine learning method based on the type characteristics of alloying elements to construct a machine learning relationship model of "composition / process-property", and combine the distributed SHAP analysis method to evaluate the strengthening and toughening ability of the alloying elements themselves; Use a machine learning method based on the physical and chemical characteristics of alloying elements, and through a three-step feature engineering method, construct a machine learning model of "composition / process-property" with high prediction accuracy; Use the adaptive learning optimization strategy to determine the multi-objective expected improvement values of yield strength and elongation in the composition search space. Through multiple calculations and combined with experiments for iterative optimization, obtain the optimized matrix alloy composition corresponding to the optimal performance.

7. The composition design method of the titanium diboride reinforced aluminum-silicon matrix composite material according to claim 6, characterized in that The specific steps of the distributed SHAP analysis method are: list the SHAP values corresponding to all feature values; when the feature values of several samples are the same, take the average of their SHAP values as the SHAP value under this feature value; Taking the feature value as the abscissa and the SHAP value as the ordinate, draw a bar chart.

8. The method for designing the composition of the titanium diboride reinforced aluminum-silicon matrix composite according to claim 6, characterized in that The specific machine learning method using the physical and chemical characteristics of alloy elements is as follows: 55 common physical and chemical characteristics of each alloy element are selected and converted into 110 alloy factor characteristics according to the following formula: where, f mi and f vi are the average factor and variance factor of the i-th physicochemical characteristic, respectively; α j represents the atomic ratio of the j-th alloying element.

9. The composition design method of the titanium diboride reinforced aluminum-silicon matrix composite according to claim 6, characterized in that The three-step feature engineering method specifically includes: correlation screening, recursive elimination, and exhaustive method screening.

10. The composition design method of the titanium diboride reinforced aluminum-silicon matrix composite material according to claim 6, characterized in that, The specific calculation formula of the multi-objective expected improvement MOEI is: z = (μ - μ * ) / σ MOEI = EI YS × EI EL where EI YS / EL represents the desired improvement value of the yield strength or elongation; and φ(z) are the probability density function and the cumulative distribution function, respectively; μ * is the reference value, set to 140 MPa for YS and 9% for EL; σ and μ are the uncertainty (RMSE value) and the average predicted value calculated by the bootstrap resampling method, respectively.

Images