Experimental data driven multi-principal element alloy strength prediction and design method

Abstract

The invention relates to an experimental data-driven multi-principal-element alloy strength prediction and design method, which comprises the following steps: data acquisition: measuring or collecting room-temperature tensile curves of various large-grain alloy samples to obtain yield strength sigma y; carrying out characteristic engineering, defining an electronegative mismatch parameter delta x, measuring local structure distortion parameters Uiso and epsilon 1st, and calculating to obtain an alloy poisson ratio v; constructing a model, adopting a symbolic regression algorithm, and inputting characteristics including delta x, Uiso, epsilon 1st and Poisson's ratio v; the expression complexity and the fitting precision are balanced through a multi-objective optimization algorithm, an optimal solution set is generated, and a formula with the training set and the test set in front, the difference is minimum and the formula complexity is not larger than 6 is selected from the optimal solution set to serve as a yield strength prediction formula; component design is conducted, the yield strength of different alloys is solved through multiple analytical formulas, and alloy components with the large product of the electronegative difference and the shear modulus are screened out; and determining a single-phase FCC component window through phase diagram calculation. According to the method, the component design efficiency is improved; a quantitative design tool is provided for high-strength material development in the fields of aerospace, nuclear energy and the like, and the research and development period is shortened.

Description

Technical Field

[0001] The present invention relates to the technical field of material computational design, and in particular to an experimental data-driven method for predicting and designing the strength of multi-principal element alloys, and in particular to a material composition optimization design technology with the electronegativity mismatch (Δχ) as the core feature. Background Art

[0002] Existing material strength prediction models mostly rely on empirical formulas and first-principles calculation parameter inputs, and have the following defects:

[0003] 1. Parameter limitations: Traditional strengthening theories (such as the dislocation pinning model) cannot explain the complex solid solution strengthening mechanism of multi-principal element alloys;

[0004] 2. High computational cost: First-principles calculations are difficult to cover a wide composition space;

[0005] 3. Defects of black-box models: Deep learning models lack interpretability and are difficult to guide composition design. Summary of the Invention

[0006] The problem solved by the present invention is to provide an experimental data-driven method for predicting and designing the strength of multi-principal element alloys, which improves the efficiency of composition design; provides a quantitative design tool for the development of high-strength materials in fields such as aerospace and nuclear energy, and shortens the R & D cycle.

[0007] In order to achieve the above object, the present invention adopts the following technical solutions:

[0008] An experimental data-driven method for predicting and designing the strength of multi-principal element alloys, characterized by comprising the following steps:

[0009] Step 1. Data collection

[0010] Measure or collect the room-temperature tensile curves of a variety of large-grain alloy samples to obtain the yield strength σ y ;

[0011] Step 2. Feature engineering:

[0012] Define the electronegativity mismatch parameter Δχ, measure the local structure distortion parameters U iso and ε 1st , and calculate the Poisson's ratio ν of the alloy;

[0013] Step 3. Model construction:

[0014] Balance the expression complexity and fitting accuracy through a multi-objective optimization algorithm to generate an optimal solution set, and select a formula with a high ranking in both the training set and the test set, the smallest difference, and a formula complexity not greater than 6 as the yield strength prediction formula;

[0015] Step 4. Composition design:

[0016] Use multiple analytical formulas to solve the yield strength of different alloys, and screen out alloy components with a large product of electronegativity difference and shear modulus; determine the single-phase FCC composition window through phase diagram calculation.

[0017] The beneficial effects of the present invention are as follows:

[0018] 1. Strength prediction model: Break through the limitations of traditional theories, and reveal the linear strengthening law dominated by Δχ through symbolic regression.

[0019] 2. Compared with the traditional trial-and-error method, the component design efficiency is increased by at least more than 10 times.

[0020] 3. Commercial value: Provide a quantitative design tool for the development of high-strength materials in fields such as aerospace and nuclear energy, and shorten the R & D cycle.

[0021] As an improved technical solution, the calculation of Δχ adopts the Allen electronegativity scale, and the calculation formula of Δχ is:

[0022]

[0023] where c i represents the concentration of element i, and χ i represents the Allen electronegativity of element i.

[0024] represents the average electronegativity of the alloy.

[0025] As an improved technical solution, the local structure distortion parameters U iso and ε 1st are quantified: Use synchrotron radiation powder diffraction to obtain the isotropic atomic displacement parameter (U iso ); then, use the Fourier transform powder diffraction signal to obtain the pair distribution function of atoms, and measure the deviation of the first peak of the pair distribution function to obtain the local distortion strain ε 1st .

[0026] As an improved technical solution, use the mixing rule to calculate the G and ν of the alloy through the shear modulus G and Poisson's ratio ν of the elemental elements.

[0027] As an improved technical solution, the analytical formula adopts one of the following three formulas, and the formula is BRIEF DESCRIPTION OF THE DRAWINGS

[0028] Figure 1 . Stress-strain curve of Au85Ta15 alloy under room temperature compression. DETAILED DESCRIPTION OF THE INVENTION

[0029] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0030] An experimental data-driven multi-principal element alloy strength prediction and design method, characterized by comprising the following steps:

[0031] Step 1. Data acquisition

[0032] Measure or collect the room temperature tensile curves of a variety of large-grain alloy samples to obtain the yield strength σ y ;

[0033] Step 2. Feature engineering:

[0034] Define the electronegativity mismatch parameter Δχ; the calculation of Δχ uses the Allen electronegativity scale, and the calculation formula of Δχ is:

[0035]

[0036] where c i represents the concentration of element i, χ i represents the Allen electronegativity of element i, represents the average electronegativity of the alloy.

[0037] Quantify the local structure distortion parameters U iso and ε 1st : Use synchrotron radiation powder diffraction to obtain the isotropic atomic displacement parameter (U iso ); then, use the Fourier transform powder diffraction signal to obtain the pair distribution function of atoms, and measure the deviation of the first peak of the pair distribution function to obtain the local distortion strain ε 1st . Use the mixing rule to calculate the G and ν of the alloy through the shear modulus G and Poisson's ratio ν of the elemental elements.

[0038] Step 3. Model construction:

[0039] Adopt the symbolic regression algorithm, and the input features include Δχ, U iso , ε 1st and Poisson's ratio ν; through genetic programming optimization, obtain formulas with top performance in both the training set and the test set and a formula complexity not greater than 6, and select the one with the smallest performance difference between the training set and the test set as the yield strength prediction formula; the analysis formula adopts one of the following three formulas, and the formula is and See Table 1 for details

[0040] Formulas with high performance in both the training set and the test set and a formula complexity not greater than 6

[0041]

[0042]

[0043] Step 4. Composition design:

[0044] Solve the yield strength of different alloys using multiple analytical formulas, and screen out alloy components with a large product of electronegativity difference and shear modulus; determine the single-phase FCC composition window through phase diagram calculation.

[0045] The effects achieved by using the above method are as follows:

[0046] 1. Strength prediction model: Break through the limitations of traditional theories, and reveal the linear strengthening law dominated by Δχ through symbolic regression.

[0047] 2. Compared with the traditional trial-and-error method, the composition design efficiency is increased by at least 10 times or more.

[0048] 3. Commercial value: Provide a quantitative design tool for the development of high-strength materials in fields such as aerospace and nuclear energy, and shorten the R & D cycle.

[0049] The above data can be statistically incorporated into a thermodynamic calculation database that includes a Δχ optimization module and single-phase composition screening. Provide data support for further research and development.

[0050] Taking the alloy design of the Au-Ta system as an example, the specific steps are as follows:

[0051] Step 1. Data preparation: Collect Δχ, σ y and G data of no less than 10 multi-principal element alloys;

[0052] Step 2. Model training: Use the Python programming language and the gplearn library for symbolic regression, and select the optimal model with a formula complexity length ≤ 6 and a performance in the dataset and test set of not less than 80%.

[0053] Step 3. Composition optimization: Calculate the solid solution composition that can form a single phase in the Au-Ta system, calculate the electronegativity difference, screen out the maximum value Δχ = 0.113, calculate the shear modulus of the corresponding composition Au85Ta15 to be 35 GPa, and predict the yield strength range of the Au85Ta15 alloy to be 229 according to the first formula in Table 1;

[0054] Step 4. Experimental verification: Prepare the Au85Ta15 alloy by arc melting, and measure the actual σy = 234 MPa in the room temperature compression test, with a deviation from the prediction of < 5%. The stress-strain curve of the Au85Ta15 alloy at room temperature compression can be referred toFigure 1 as shown

[0055] As mentioned above, it is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention, according to the technical solution and inventive concept of the present invention, making equivalent substitutions or changes should be covered within the protection scope of the present invention.

Claims

1. An experimental data-driven method for predicting and designing the strength of multi-principal element alloys, characterized in that, Including the following steps: Step 1. Data collection Measure or collect the room temperature tensile curves of a variety of large-grained alloy samples to obtain the yield strength v y ; Step 2. Feature engineering: Define the electronegativity mismatch parameter Δχ and measure the local structure distortion parameters U iso and ε 1st , and calculate the Poisson's ratio ν of the alloy; Step 3. Model construction: By using a multi-objective optimization algorithm to balance the expression complexity and fitting accuracy, generate an optimal solution set, and select a formula with the best performance in both the training set and the test set, the smallest difference, and a formula complexity not greater than 6 as the yield strength prediction formula; Step 4. Composition design: Solve the yield strength of different alloys with various analytical formulas, and screen out alloy compositions with a large product of the electronegativity difference and shear modulus; determine the single-phase FCC composition window through phase diagram calculation.

2. The method for predicting and designing the strength of a multi-principal element alloy driven by experimental data according to claim 1, wherein The calculation of the said Δχ adopts the Allen electronegativity scale, and the calculation formula of the said Δχ is: where c i represents the concentration of element i, χ i represents the Allen electronegativity of element i, and represents the average electronegativity of the alloy.

3. A method for predicting and designing the strength of a multi-principal element alloy driven by experimental data according to claim 1, characterized in that Quantify the local structural distortion parameters U iso and ε 1st : Obtain the isotropic atomic displacement parameter (U iso ) using synchrotron radiation powder diffraction; then, obtain the pair distribution function of atoms by using the Fourier transform powder diffraction signal, and measure the deviation of the first peak of the pair distribution function to obtain the local distortion strain ε 1st .

4. A method for predicting and designing the strength of a multi-principal element alloy driven by experimental data according to claim 1, characterized in that Using the mixing rule, calculate the G and ν of the alloy through the shear modulus G and Poisson's ratio ν of the elemental elements.

5. A method for predicting and designing the strength of a multi-principal element alloy driven by experimental data according to claim 1, characterized in that The analysis formula adopts one of the following three formulas, and the formulas are and

Images