Machine learning-based mixed rare earth magnesium alloy heat treatment process and mechanical property optimization method
Through machine learning model and multi-objective optimization algorithm, the heat treatment process of mixed rare earth magnesium alloys is solved, and the problem of insufficient mechanical properties and heat resistance of traditional cast magnesium alloys is realized, low-cost and high-performance magnesium alloy preparation is achieved, reducing rare earth costs and improving performance.
Patent Information
- Application Number
- CN202510353558.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-03-25
- Publication Date
- 2025-07-08
AI Technical Summary
Traditional cast magnesium alloys are insufficient in terms of mechanical properties and heat resistance, and the cost of high-risk earth content alloys is too high, and the mixed rare earth ratio and heat treatment process are complicated, resulting in low efficiency.
The machine learning model is used to predict the mixed rare earth ratio and heat treatment process, and combined with multi-objective optimization algorithms, low-cost and high-performance preparation solutions are generated, including data preprocessing, machine learning model establishment, multi-objective optimization and data expansion.
It has achieved high strength and heat resistance of low-cost magnesium alloys, reduced rare earth costs by 30%-50%, improved mechanical properties, shortened R&D cycle, ensured that process parameters are accurate and controllable, and in line with the trend of green manufacturing.
Smart Images
Figure CN120280055A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of magnesium alloy material design, and particularly relates to an optimization method for the heat treatment process of a rare earth mixed magnesium alloy based on machine learning, which realizes high strength and heat resistance by replacing pure rare earth with low-cost mixed rare earth. Background Technique
[0002] Magnesium and its alloys, as an innovative structural material, have been widely used in the fields of aerospace, electronic communication, and military equipment due to their low density, high specific strength, large elastic modulus, and excellent heat dissipation performance, and their market demand is rising rapidly. Among the applications of magnesium alloys, cast magnesium alloy products dominate the market. However, compared with cast aluminum alloys, traditional cast magnesium alloys are slightly insufficient in mechanical properties and heat resistance. Due to the unique valence electron structure of rare earth elements and their significant strengthening effect in magnesium, rare earth (RE) alloying has become one of the key means to improve the strength and toughness of magnesium alloys and plays an important role in the research and development of new magnesium-based materials.
[0003] One of the main problems restricting the wide application of rare earth magnesium alloys is the high alloy cost, especially for magnesium alloys with a high rare earth content. One of the main ways to solve this problem at present is to use cheaper mixed rare earth to replace pure rare earth. For example, Ce / La mixed rare earth, as a by-product of extracting precious metals such as Nd and Pr, has a very low price, only half of the price of pure Ce or pure La. It has been found through research that when the Ce / La mixed rare earth accounts for more than 20wt.% of the total rare earth elements, the high-temperature mechanical properties of rare earth magnesium alloys can be significantly improved. However, the influence of its ratio and heat treatment process on performance is complex, and the efficiency is low when using the traditional trial-and-error method.
[0004] It should be noted that the information disclosed in this background technical part is only intended to deepen the understanding of the overall background technology of the present invention, and should not be regarded as an admission or any form of implication that this information constitutes the prior art known to those skilled in the art. Summary of the Invention
[0005] In view of this, the purpose of the present invention is to provide a method for predicting the influence of the mixed rare earth ratio and heat treatment process on mechanical properties through a machine learning model, and combining a multi-objective optimization algorithm to generate a preparation plan with low cost and high performance.
[0006] To achieve the above purpose, the present invention adopts the following technical solutions:
[0007] In the first aspect, the present invention provides an optimization method for the heat treatment process of a rare earth mixed magnesium alloy based on machine learning, including the following steps:
[0008] (a) Construct a database containing mixed rare earth components, heat treatment process parameters, and mechanical properties;
[0009] (b) Perform data preprocessing on the said database, including feature dimensionality reduction and outlier removal;
[0010] (c) Establish a machine learning model with the mixed rare earth ratio and heat treatment parameters as inputs and mechanical properties as outputs;
[0011] (d) Use a multi-objective optimization algorithm to iteratively optimize the said model to generate a combination of mixed rare earth ratio and heat treatment process that meets the target performance;
[0012] (e) Output the range of preparation parameters for the mixed rare earth magnesium alloy.
[0013] Further, the machine learning model in step (c) is a deep neural network, and the activation functions include ReLU, Tanh or Sigmoid.
[0014] Further, the heat treatment process parameters include solution temperature, solution time, aging temperature and aging time.
[0015] Further, the mixed rare earth includes Ce and La, the proportion of Ce / La is 10 - 70 wt.%, and the total rare earth content ≤ 5 wt.%.
[0016] Further, it further includes generating synthetic data through a generative adversarial network and expanding the database after performing a consistency test with the original data distribution.
[0017] Further, the optimization objectives of the multi-objective optimization algorithm include tensile strength, elongation and yield strength, and the cost of the mixed rare earth is constrained.
[0018] Further, the multi-objective optimization algorithm includes genetic algorithm, particle swarm algorithm, simulated annealing algorithm.
[0019] Further, in step (b), the data preprocessing methods used include Pearson correlation coefficient and recursive feature elimination.
[0020] Further, this method also includes optimizing the hyperparameters of the established neural network to obtain the neural network model with the best prediction effect.
[0021] In the second aspect, the present invention provides a mixed rare earth magnesium alloy product prepared by using the method described in the first aspect above, wherein the proportion of Ce / La mixed rare earth ≥ 20 wt.%, and the tensile strength ≥ 200 MPa.
[0022] The method for optimizing the heat treatment process of the mixed rare earth magnesium alloy based on machine learning provided by the present invention has the following remarkable advantages:
[0023] 1. Synergy of low cost and high performance: By using mixed rare earths to replace pure rare earth elements, while ensuring that the tensile strength of the magnesium alloy is ≥200 MPa and the elongation is ≥8%, the rare earth cost is reduced by 30%-50%, breaking through the bottleneck of relying on traditional high-purity rare earths.
[0024] 2. Multi-objective optimization to avoid performance imbalance: Combining optimization algorithms to synergistically optimize strength, elongation and cost, solving the problem of performance degradation caused by single-objective optimization. For example, when the optimized Ce / La ratio is 25% in the embodiment, the tensile strength is increased by 15% and the elongation remains stable.
[0025] 3. Data-driven efficient design: Using machine learning models to accurately predict mechanical properties (prediction error ≤5%), reducing the number of traditional trial-and-error experiments by more than 80% and shortening the R & D cycle; expanding data through generative adversarial networks to improve the generalization ability of the model and solve the limitation of small samples.
[0026] 4. Precise control of process parameters: Defining key process windows such as solution temperature (460±10°C) and aging time (6±2 h) to guide industrial production and ensure performance consistency.
[0027] 5. Environmental protection and resource utilization: Using Ce / La mixed rare earths (metallurgical by-products) to achieve resource recycling, reducing the environmental pressure of rare earth mining, and conforming to the trend of green manufacturing.
[0028] The present invention provides a low-cost and high-precision process design solution for high-performance magnesium alloy components in fields such as aerospace and new energy vehicles, with significant economic benefits and technical promotion value. Description of the Drawings
[0029] Figure 1 It is a flowchart of the method for predicting the mechanical properties and optimizing the process of rare earth magnesium alloy based on machine learning provided by the embodiment of the present invention, showing the complete process of data collection, modeling and optimization;
[0030] Figure 2 It is a comparison curve of the tensile strength and the actual value in the database of ultrasonic pulping semi-solid rheoforming rare earth magnesium alloy by the neural network model in the embodiment of the present invention;
[0031] Figure 3 It is the prediction result of the yield strength by the neural network model in the database of ultrasonic pulping semi-solid rheoforming rare earth magnesium alloy in the embodiment of the present invention;
[0032] Figure 4 It is the prediction result of the elongation by the neural network model in the database of ultrasonic pulping semi-solid rheoforming rare earth magnesium alloy in the embodiment of the present invention;
[0033] Figure 5In the embodiments of the present invention, the microstructure and property results of rare earth magnesium alloys with different Ce / La ratios designed according to the machine learning results. Specific embodiments
[0034] The technical features and advantages of the present invention are described in more detail below with reference to the accompanying drawings, so that the advantages and features of the present invention can be more easily understood by those skilled in the art, and thus a clearer and more definite definition of the protection scope of the present invention can be made.
[0035] The embodiments of the present invention provide an optimization method for rare earth composition and heat treatment process of ultrasonic pulping semi-solid rheoforming rare earth magnesium alloys using a machine learning model. The method first collects a database of mixed rare earth magnesium alloy composition - heat treatment process - mechanical properties, then constructs a machine learning model, and predicts the mechanical properties of magnesium alloy castings by inputting the proportioning composition of the mixed rare earth magnesium alloy and the heat treatment process parameters. At the same time, based on the model, the composition of the mixed rare earth magnesium alloy and the heat treatment process under the best mechanical properties are obtained through algorithm optimization to guide the design of a multi-component mixed rare earth magnesium alloy with low cost, high strength and heat resistance containing mixed rare earth. In addition, according to the mechanical properties required for different magnesium alloy products, different mixed rare earth or pure yttrium rare earth compositions and heat treatment process designs can be provided. It effectively reduces the production and design costs of rare earth magnesium alloys and realizes the design of the mixed rare earth composition and heat treatment process of mixed rare earth magnesium alloys. The inventive concept of the method in this embodiment is summarized as follows:
[0036] 1. Data collection: Design mixed rare earth magnesium alloys with different ratios, record the heat treatment process parameters, and test the mechanical property data of the corresponding castings to establish a database of rare earth composition - heat treatment process - mechanical properties of rare earth magnesium alloys.
[0037] 2. Model construction: Construct a machine learning model with rare earth composition and heat treatment process parameters as inputs and mechanical properties as outputs.
[0038] 3. Multi-objective optimization: Simultaneously use multiple mechanical properties as optimization objectives and adopt a multi-objective optimization algorithm to generate an optimal heat treatment process plan.
[0039] 4. Data expansion: Generate synthetic data through GAN, and update the model after verifying the distribution consistency.
[0040] As an example, the above method is described in detail below using a neural network method as the machine learning model, a genetic algorithm as the multi-objective optimization algorithm, and Y, Ce, and La as rare earth elements. The flow chart of the method refers to Figure 1 to show the complete processes of data collection, modeling, and optimization. The specific steps include:
[0041] Step 1: According to the strengthening effect of rare earth elements content and types on magnesium alloys, design a variety of rare earth magnesium alloys with different mixed rare earth elements content and carry out melting.
[0042] Step 2: Design and record different heat treatment processes of magnesium alloys to control the internal structure of magnesium alloy components.
[0043] Step 3: Take samples from the finally obtained magnesium alloy components for mechanical property determination, and combine the previous rare earth composition and heat treatment process parameters to establish a database of rare earth composition - heat treatment process - mechanical properties of rare earth magnesium alloys. For example, the following table shows the composition, process parameters and mechanical property data of some rare earth magnesium alloys in the database. In order to evaluate the performance of the mixed rare earth magnesium alloy, some pure Y rare earth magnesium alloys are additionally designed as a comparison in the database.
[0044] To improve the accuracy of model prediction, the established database can contain 100 - 200 pieces of data. Combining the applicant's experience and comprehensively considering cost and alloy performance, the content of Y element in the alloy is controlled at 0 - 4wt.%, the content of La element is controlled at 0 - 2wt.%, the Ce / La ratio is controlled at 0 - 50%, the solution temperature is designed as 400 - 500°C, and the solution time is designed as 0 - 10min.
[0045] Table 1 Partial content of the database of rare earth composition - heat treatment process - mechanical properties
[0046]
[0047]
[0048] Step 4: Perform data preprocessing on the established database of mixed rare earth composition - heat treatment process - mechanical properties of rare earth magnesium alloys, and screen the initial data and features to optimize the process of establishing the neural network model. In this process, the Pearson correlation coefficient is used to calculate the correlation between features. The results show that the correlation between Ce content and La content is relatively high. In the subsequent training and model establishment process, the Ce element content is removed. The hyperparameters used for the established neural network are: the number of hidden layers is 2, the number of neurons is 48, the activation function is relu, the α value is 7.26e - 5, the learning rate is 3.43e - 5. The comparison between the predicted values and the actual values of the neural network model for tensile strength, yield strength and elongation is respectively as Figures 2 - 4As shown. Among them, the RMSE of the tensile strength prediction result is 5.37, the MSE is 4.96, and the R2 is 0.94. The RMSE of the yield strength prediction result is 5.37, the MSE is 4.17, and the R2 is 0.96. The RMSE of the elongation prediction result is 1.04, the MSE is 0.82, and the R2 is 0.91. Table 2 shows the comparison between the measured mechanical properties of the rare earth magnesium alloy and the predicted values of the machine learning model, and the errors are all within 10%.
[0049] Step 5: Establish a neural network model and optimize the hyperparameters of the neural network to obtain the neural network model with the best prediction effect.
[0050] Table 2 Results of the composition and process parameters of the rare earth magnesium alloy designed based on the machine learning model
[0051]
[0052] It can be seen from the results in Table 2 that for the rare earth magnesium alloy designed by the neural network model constructed in the embodiments of the present invention, the differences between the actual values and the predicted values of the tensile strength and elongation are relatively small, and the prediction accuracy is relatively high.
[0053] Step 6: Based on the optimized neural network model, use the genetic algorithm, with the tensile strength, yield strength, and elongation of the rare earth magnesium alloy as the optimization objectives, to optimize the neural network model, and obtain the composition and process parameters of the rare earth magnesium alloy under the simultaneous optimization of multiple objectives, as shown in Table 3.
[0054] Table 3 Results of the composition and process parameters of the rare earth magnesium alloy optimized based on machine learning
[0055] UTS = 180 - 210; E = 5 - 8 UTS = 210 - 240; E = 8 - 10 UTS = 240+; E = 10+ Solution time 4.7 5.1 4.5 Solution temperature 470.3 460.6 455.1 Ce 0.0 0.0 0.0 La 0.0 0.0 0.0 Y 2.6 2.3 1.2
[0056] It can be seen from the results in Table 3 that compared with the mixed rare earth magnesium alloy, the pure yttrium rare earth magnesium alloy has better performance.
[0057] Step 7: Use the deep learning method to learn the database of the composition- heat treatment process- mechanical properties of the rare earth magnesium alloy to generate multiple groups of new data that conform to the distribution of the original database, and use the optimized neural network model to perform performance prediction to obtain a new database.
[0058] Step 8: Perform data analysis on the new database to obtain the range of rare earth components and heat treatment process parameters of the mixed rare earth magnesium alloy under different mechanical properties, as shown in Table 4.
[0059] Table 4 Results of the composition and process parameters of the mixed rare earth magnesium alloy designed based on machine learning
[0060]
[0061] According to the machine learning results, rare earth magnesium alloys with different Ce / La ratios are designed, and their microstructures are as follows Figure 5 shown Figure 5 Figures a-d respectively show the microstructure diagrams with the rare earth mixture ratios of 5-15%, 15-25%, 25-35%, and 35-45%, and their mechanical properties respectively reach the mechanical properties corresponding to the rare earth mixture contents in Table 4
[0062] It can be seen from the results of Table 3 and Table 4 that the higher the proportion of rare earth mixtures such as La and Ce, the poorer the performance of the magnesium alloy. However, the use of rare earth mixtures can achieve mechanical properties similar to those of pure yttrium rare earths, effectively reducing the cost of rare earth magnesium alloys
[0063] In summary, the design model of the present invention guided by mechanical properties provides a rapid and effective method for realizing the rare earth mixture composition and heat treatment process of rare earth magnesium alloys
[0064] In the description of the embodiments of the present invention, it should be understood that the orientation or positional relationships indicated by the terms "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "center", "top", "bottom", "top part", "bottom part", "inner", "outer", "inner side", "outer side", etc. are the orientation or positional relationships based on the orientation or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and thus cannot be construed as a limitation of the present invention. Among them, the "inner side" refers to the internal or enclosed area or space. The "periphery" refers to the area around a specific component or specific area
[0065] In the description of the embodiments of the present invention, the terms "first", "second", "third", "fourth" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly specifying the quantity of the indicated technical features. Thus, the features defined with "first", "second", "third", "fourth" may explicitly or implicitly include one or more of such features. In the description of the present invention, unless otherwise specified, the meaning of "a plurality" is two or more
[0066] In the description of the embodiments of the present invention, it should be noted that unless otherwise clearly specified and limited, the terms "installed", "connected", "joined", "assembled" should be understood in a broad sense. For example, it may be a fixed connection, a detachable connection, or an integral connection; it may be directly connected, or indirectly connected through an intermediate medium, and it may be the communication inside two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances
[0067] In the description of the embodiments of the present invention, specific features, structures, materials, or characteristics may be combined in a suitable manner in any one or more embodiments or examples.
[0068] In the description of the embodiments of the present invention, it should be understood that "-" and "~" represent the range between two numerical values, and this range includes the endpoints. For example, "A-B" represents a range greater than or equal to A and less than or equal to B. "A~B" represents a range greater than or equal to A and less than or equal to B.
[0069] In the description of the embodiments of the present invention, the term "and / or" herein is merely a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " herein generally represents an "or" relationship between the associated objects before and after.
[0070] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for optimizing the heat treatment process of a rare earth magnesium alloy based on machine learning, characterized in that, It includes the following steps: (a) Construct a database containing mixed rare earth components, heat treatment process parameters, and mechanical properties; (b) Perform data preprocessing on the database, including feature dimensionality reduction and outlier removal; (c) Establish a machine learning model with the mixed rare earth ratio and heat treatment parameters as inputs and the mechanical properties as outputs; (d) Use a multi-objective optimization algorithm to iteratively optimize the model to generate a combination of mixed rare earth ratios and heat treatment processes that meet the target performance; (e) Output the preparation parameter range of the mixed rare earth magnesium alloy.
2. The method according to claim 1, wherein The machine learning model in step (c) is a deep neural network, and the activation functions include ReLU, Tanh, or Sigmoid.
3. The method according to claim 1, wherein The heat treatment process parameters include solution temperature, solution time, aging temperature, and aging time.
4. The method according to claim 1, characterized in that, The mixed rare earth includes Ce and La, the proportion of Ce / La is 10 - 70 wt.%, and the total rare earth content ≤ 5 wt.%.
5. The method according to claim 1, wherein It further includes generating synthetic data through a generative adversarial network and expanding the database after consistency testing of the original data distribution.
6. The method according to claim 1, characterized in that The optimization objectives of the multi-objective optimization algorithm include tensile strength, elongation, and yield strength, and the cost of the mixed rare earth is constrained.
7. The method according to claim 1, wherein The multi-objective optimization algorithm includes genetic algorithm, particle swarm algorithm, and simulated annealing algorithm.
8. The method according to claim 1, wherein The data preprocessing methods used in step (b) include Pearson correlation coefficient and recursive feature elimination.
9. The method according to claim 2, wherein It further includes optimizing the hyperparameters of the established neural network to obtain the neural network model with the best prediction effect.
10. A mixed rare earth magnesium alloy product, characterized in that, Prepared by the method according to any one of claims 1 - 9, wherein the proportion of the Ce / La mixed rare earth is ≥ 20 wt.%, and the tensile strength is ≥ 200 MPa.