A Machine Learning-Based Method for Composition Optimization and Mechanical Property Prediction of High-Strength Deformable Rare Earth Magnesium Alloys; Computer-Readable Storage Medium

Abstract

This invention provides a method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys based on machine learning, as well as a computer-readable storage medium. The method includes the following steps: S1. Collecting magnesium alloy data and corresponding mechanical property data, constructing an alloy database, and cleaning and standardizing the data to construct a feature matrix containing compositional and process characteristics; S2. Based on the feature matrix, analyzing the influence of each compositional and process characteristic on preset mechanical property indicators, and identifying the key influencing factors affecting each mechanical property indicator; S3. Training a machine learning prediction model and performing collaborative multi-objective optimization to obtain an optimized solution set that meets the preset performance objectives; S4. Within a preset compositional range, designing an alloy composition that meets the target mechanical properties based on the above; S5. Based on the prediction model and optimization algorithm, optimizing key process parameters and outputting the design results. This invention can significantly improve the comprehensive performance of rare-earth magnesium alloys.

Description

Technical Field

[0001] This invention relates to the field of magnesium alloy material design technology, and in particular to a method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare earth magnesium alloys based on machine learning, as well as a computer-readable storage medium. Background Technology

[0002] Magnesium alloys, as the lightest metallic structural materials currently used in engineering applications, possess broad application prospects in aerospace, transportation, 3C electronics, and defense industries due to their low density, high specific strength, high specific stiffness, and excellent thermal conductivity, damping, and electromagnetic shielding properties. Achieving lightweight equipment is one of the key goals continuously pursued in these fields, making magnesium alloys a highly attractive material choice.

[0003] However, traditional magnesium alloys (such as the AZ and AM series) generally lack sufficient absolute strength, room-temperature plasticity, and high-temperature creep resistance, making it difficult to meet the performance requirements of key load-bearing structural components or high-temperature service environments. This severely restricts their wider application. To overcome these performance bottlenecks, rare earth element alloying is considered one of the most effective technical approaches. By adding rare earth elements such as gadolinium (Gd) and yttrium (Y), the strength and heat resistance of magnesium alloys can be significantly improved. The strengthening mechanism mainly stems from the solid solution strengthening of rare earth atoms in the magnesium matrix, as well as the precipitation strengthening brought about by the formation of rare earth-rich nanoscale precipitates (such as β′ phase and LPSO structural phase) during subsequent heat treatment.

[0004] In recent years, Mg-Gd-Y-Zn-Zr alloys have become a research hotspot for high-performance rare-earth magnesium alloys due to their excellent room-temperature strength, high-temperature creep resistance, and certain plasticity. In this system, the combined addition of Gd and Y elements not only produces a strong solid solution strengthening effect, but also promotes the precipitation of various strengthening phases through interaction with Zn elements, thereby significantly improving the overall mechanical properties of the alloy.

[0005] Nevertheless, existing high-performance rare-earth magnesium alloys based on the Mg-Gd-Y-Zn-Zr system still face a series of prominent challenges in their path to large-scale engineering applications, the core of which lies in: 1) The contradiction between strength and processing performance: High alloying composition (especially high rare earth content) designed to obtain ultra-high strength usually leads to a decrease in the casting fluidity of the alloy, a narrowing of the hot working plasticity window, and a sharp increase in deformation resistance. This makes the alloy prone to defects during casting and faces problems such as high risk of cracking and difficulty in forming during plastic processing such as hot extrusion and rolling, which seriously affects the yield and processing efficiency of the material.

[0006] 2) Challenges in controlling performance stability: The microstructure and final properties of this alloy system are extremely sensitive to compositional fluctuations and hot working parameters (such as homogenization temperature, extrusion temperature and speed, aging regime, etc.). Slight deviations in process parameters can lead to significant changes in the type, size, distribution, and volume fraction of the strengthening phase, resulting in batch-to-batch performance instability and poor reproducibility, making it difficult to meet the stringent requirements for material performance consistency in engineering applications.

[0007] 3) The challenge of balancing comprehensive performance: Existing research and technologies often focus on improving ultimate strength by increasing rare earth content or optimizing a single process. However, there is still a lack of systematic solutions on how to effectively improve the room temperature plasticity, fracture toughness, and deformation processing capability of alloys while ensuring high strength through precise composition design and synergistic control of multi-stage processes. The optimization potential of composition and process has not yet been fully explored.

[0008] Therefore, how to achieve a synergistic improvement in high strength, good plasticity, and excellent hot working performance within the Mg-Gd-Y-Zn-Zr alloy system framework through innovative composition optimization and process design, while breaking through the traditional dilemma that "high strength is inevitably accompanied by low plasticity and difficult processing," and ensuring the stability of microstructure and properties, has become a key technical bottleneck that urgently needs to be overcome in this field. Solving this problem has significant practical implications for promoting the application of high-performance rare-earth magnesium alloys from the laboratory to practical engineering applications.

[0009] In view of this, the present invention is proposed. Summary of the Invention

[0010] The purpose of this invention is to provide a method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys based on machine learning, as well as a computer-readable storage medium, thereby solving at least one of the problems mentioned in the background art. For example, the method of this invention addresses the shortcomings of existing rare-earth magnesium alloys in terms of mechanical properties and heat resistance, particularly the difficulty in simultaneously achieving high strength, toughness, and deformability.

[0011] In a first aspect, the present invention provides a method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys based on machine learning, comprising the following steps: Step S1. Data collection and preprocessing: Collect composition data, process parameter data and corresponding mechanical property data of Mg-Gd-Y-Zn series magnesium alloys, construct an alloy database (such as a high-strength forged rare earth magnesium alloy database), and clean and standardize the data to construct a feature matrix containing composition features and process features. Data cleaning may include removing outliers and severely missing data. The above feature matrix can be constructed by extracting key features. Step S2. Analysis of factors affecting mechanical properties: Based on the feature matrix, machine learning methods are used to analyze the influence of the characteristics of each component and process on the preset mechanical performance indicators, and to determine the key factors affecting each mechanical performance indicator; Step S3. Construct a performance prediction model and perform collaborative optimization: Use component parameters and process parameters as input features and mechanical performance indicators as output to train a machine learning prediction model; and use optimization algorithms to perform collaborative multi-objective optimization of component and process parameters to obtain an optimized solution set that meets the preset performance objectives. Step S4. Composition optimization design based on prediction results: Within the preset composition range, composition optimization is performed based on the optimization algorithm, and based on the optimization results, an alloy composition that meets the target mechanical properties is designed. Step S5. Process Parameter Optimization and Alloy Preparation: Based on the prediction model and optimization algorithm, optimize the key process parameters and output the designed alloy composition and its corresponding key process parameters.

[0012] In some embodiments, in step S1: the composition data includes at least the elemental contents of Gd, Y, Zn, and Zr; the process parameter data includes one or more of the following: solution temperature, solution time, extrusion ratio, and extrusion temperature; the mechanical property indicators include at least one of yield strength, tensile strength, and elongation after fracture. These data can be collected from existing literature, such as from at least one of the following literature databases: Elsevier ScienceDirect, Springer, EBSCO, and CNKI.

[0013] In some embodiments, in step S1, the range of the composition data includes: Gd: 0~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%; the range of the process parameters includes: solution temperature: 500~600℃, solution time: 10~30 hours, extrusion ratio: 10~20, extrusion temperature: 400~500℃; the range of the mechanical property indicators includes: yield strength: 250~550MPa, tensile strength: 350~650MPa, elongation after fracture: 5~15%.

[0014] Specifically, in step S2, determining the key influencing factors affecting each mechanical performance index may include: The effects of Y, Zn, Mn element content, extrusion temperature, homogenization temperature, solution temperature, and solution time on yield strength were analyzed. The effects of the content of Y, Zn, Mn, and Cu elements, as well as extrusion temperature, homogenization temperature, solution temperature, and solution time, on tensile strength were analyzed. The effects of Y, Zn, Gd element content, extrusion temperature, homogenization temperature, aging temperature, and time on elongation after fracture were analyzed.

[0015] Common high-strength wrought rare-earth magnesium alloys do not contain Cu, but some papers have investigated the effect of Cu on the properties of rare-earth magnesium alloys. Analysis of the collected data revealed that Cu also has a significant impact on the tensile strength of magnesium alloys. (See...) Figure 2 and Figure 3 Therefore, Cu was also included in the elements analyzed above.

[0016] In some embodiments, determining the key influencing factors affecting each mechanical performance index in step S2 includes: constructing a prediction model using a random forest algorithm or a gradient boosting decision tree algorithm, and ranking the features output by the model based on their importance, in order to determine the key factors affecting mechanical performance.

[0017] In some embodiments, in step S3: the machine learning prediction model is a model built based on a tree ensemble model, and hyperparameter optimization is performed using cross-validation and parameter search; the optimization algorithm is a swarm intelligence optimization algorithm used to perform collaborative multi-objective optimization of component parameters and process parameters; preferably, the tree ensemble model is a random forest model, and the swarm intelligence optimization algorithm is a particle swarm optimization algorithm.

[0018] In some embodiments, step S3 includes: establishing and optimizing a machine learning prediction model; and using an optimization algorithm to perform synergistic multi-objective optimization of composition and process parameters.

[0019] Specifically, step S3 may include: S31. Establish a preliminary prediction model using machine learning algorithms (e.g., random forest algorithm); S32. Optimize the hyperparameters of the model to improve its accuracy; S33. Particle swarm optimization is used for multi-objective optimization to achieve synergistic optimization of composition and process parameters.

[0020] Preferably, step S31 may further include: based on the preliminary prediction model, using cross-validation technology to select the optimal model parameters and train an accurate prediction model. Preferably, optimizing the hyperparameters of the model in step S32 may specifically include: scanning the hyperparameter space of the model using a grid search algorithm to seek the optimal combination of hyperparameters.

[0021] Preferably, step S33 may specifically include: initializing the population, finding the optimal solution through iterative calculation, until the convergence condition is met.

[0022] In some embodiments, in step S4, the component optimization is performed based on a genetic algorithm within a preset component range. The optimization objective is, for example, to maximize the yield strength, tensile strength, and elongation simultaneously. For example, new individuals are generated through crossover and mutation operations, and surviving individuals are selected through competitive selection. Preferably, the preset component range is: Gd: 2~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%, with the remainder being Mg and unavoidable impurities.

[0023] In some embodiments, in step S4, the target mechanical properties of the optimized target composition alloy are: yield strength ≥ 395 MPa, tensile strength ≥ 469 MPa, and elongation after fracture ≥ 13.5%. After obtaining a series of optimized compositions, a specific composition can be selected by referring to the composition performance data in relevant patent literature.

[0024] Specifically, in step S4, the optimization of components can be combined with micro-organism regulation mechanisms.

[0025] Preferably, the micro-organism regulation mechanism may include: Refine recrystallized grains through particle-induced nucleation mechanism; The precipitation, morphology, and distribution of the second phase at grain boundaries are regulated to pin grain boundaries and suppress slip.

[0026] In some embodiments, step S5 includes: using process parameter optimization methods (e.g., response surface methodology) to analyze the effects of extrusion ratio, extrusion temperature, homogenization temperature, and solution temperature on mechanical properties, and determining the optimal combination of process parameters; and adjusting the strength and toughness matching of the alloy by controlling the aging temperature and time, and by inhibiting or promoting recrystallization mechanisms.

[0027] Specifically, the microstructure of the material can be adjusted by controlling the temperature and time during the aging process; the balance between the strength and elongation of the alloy can be altered by inhibiting or promoting the recrystallization mechanism; and the elongation of the material can be improved by adding trace amounts of alloying elements or second-phase particles to promote dynamic recrystallization nucleation or inhibit recrystallization grain growth. Other processes known in the art can also be used to promote dynamic recrystallization nucleation or inhibit recrystallization grain growth.

[0028] A second aspect of the present invention provides a high-strength deformable rare-earth magnesium alloy, which is designed by the above-described method and has the following constituent elements and element mass percentages as follows: Gd: 6-9%, Y: 4-6%, Zn: 0.5-3%, Zr: 0.2-0.5%, and the balance Mg.

[0029] A third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the machine learning-based method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys as described above.

[0030] Compared with existing technologies, this invention provides a method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys based on machine learning, which has the following beneficial effects: 1. By collecting and analyzing a certain amount of Mg-Gd-Y-Zn system magnesium alloy data (e.g., 502 records), a database of high-strength forged rare earth magnesium alloys was constructed, providing systematic data support for the composition optimization and process improvement of rare earth magnesium alloys, and effectively solving the problems of incomplete and unsystematic data in the existing technology; 2. Through in-depth analysis of three indicators—yield strength, tensile strength, and elongation after fracture—key factors affecting mechanical properties were identified from aspects such as alloy composition and process parameters. This enabled refined control of the properties of rare earth magnesium alloys, overcoming the shortcomings of insufficient alloy performance in existing technologies. 3. By adopting a machine learning model based on composition and process prediction performance, combined with genetic algorithm for composition optimization, the alloy composition with target performance can be quickly optimized within a reasonable range of key components such as Gd, Y, Zn, and Zr, thereby improving the comprehensive performance of rare earth magnesium alloys. 4. By optimizing process parameters such as extrusion temperature and homogenization temperature, the deformation performance of magnesium alloys was improved, effectively solving the problem that it is difficult to simultaneously achieve high strength and good deformation performance in existing technologies. 5. Based on the composition optimization results of the predicted mechanical properties, the designed alloy has excellent mechanical properties, with the target values ​​of yield strength, tensile strength and elongation after fracture reaching 395 MPa, 469 MPa and 13.5% respectively, which significantly improves the comprehensive performance of rare earth magnesium alloy. Attached Figure Description

[0031] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0032] Figure 1 This is a diagram illustrating the component-process-mechanical property database in an embodiment of the present invention; Figure 2 This is a diagram illustrating the correlation analysis between input and output variables in an embodiment of the present invention; Figure 3 This is a diagram showing the order of importance of features affecting yield strength, tensile strength, and elongation after fracture in an embodiment of the present invention; Figure 4 This is a diagram showing the predicted yield strength model in an embodiment of the present invention. Figure 5 This is a diagram showing the predicted tensile strength model in an embodiment of the present invention. Figure 6 This is a diagram showing the prediction effect of the post-fracture elongation model in an embodiment of the present invention; Figure 7 This is a diagram illustrating the optimization of yield strength, tensile strength, and elongation after fracture based on a genetic algorithm in an embodiment of the present invention. Detailed Implementation

[0033] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0034] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form includes the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0035] It should be noted that if the text uses terms such as "first" or "second", these terms are only used to distinguish similar objects and should not be interpreted as indicating or implying their relative importance, order of precedence, or implicitly indicating the number of technical features indicated. It should be understood that the data in the descriptions of "first" and "second" can be interchanged where appropriate.

[0036] The technical solution of the present invention will be clearly and completely described below with reference to the embodiments. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0037] Example 1 This embodiment provides a machine learning-based method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys. The specific steps are as follows: Step 1: Data Collection and Preprocessing Step 101: Collect composition, processing, and mechanical property data of Mg-Gd-Y-Zn magnesium alloys from literature to construct a database of high-strength forged rare-earth magnesium alloys. For example... Figure 1 As shown, the database specifically includes the composition content of Gd: 0~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%, Ce: 0~0.26 wt%, La: 0~0.3 wt%, Nd: 0~3.1 wt%, Ni: 0~1 wt%, Cu: 0~1 wt%, Mn: 0~2.56 wt%, Ti: 0~0.62 wt%, solution temperature: 500~600℃, solution time: 10~30 hours, extrusion ratio: 10~20, extrusion temperature: 400~500℃, and other process parameters, as well as mechanical property indicators such as yield strength: 250~550 MPa, tensile strength: 350~650 MPa, and elongation after fracture: 5~15%.

[0038] Step 102: Clean and standardize the collected data, remove outliers and duplicate data, and normalize the data of each indicator to make it range from 0 to 1.

[0039] Step 103: Extract key features and construct a feature matrix of components and process parameters. The contents of Gd, Y, Zn, and Zr are selected as component features, while solution temperature, solution time, extrusion ratio, and extrusion temperature are selected as process features.

[0040] Step 2, Analysis of Factors Affecting Mechanical Properties: Step 201: Through Pearson correlation coefficient analysis, the influence of several characteristics on yield strength, tensile strength, and elongation was obtained, such as... Figure 2 As shown in the figure. For yield strength, Zn, Mn, Y, extrusion temperature, and homogenization temperature are positively correlated with it, while solution temperature and solution time are negatively correlated. In particular, the addition of Zn and Mn significantly improves yield strength. For tensile strength, Y, Mn, Zn, Cu, extrusion temperature, and homogenization temperature are positively correlated with it, while solution temperature and solution time are negatively correlated. In particular, increasing Y, Mn, and Zn significantly improves tensile strength. For elongation, Zr, extrusion temperature, and extrusion ratio have a positive effect, while Zn, Ti, and aging temperature have a negative correlation.

[0041] Step 202: Analyze the factors affecting yield strength, including the content of Zn, Mn, and Y elements, extrusion temperature, homogenization temperature, solution temperature, and solution time. Linear regression analysis shows that increasing the content of Zn and Mn significantly improves yield strength, while excessively long solution temperatures and times decrease yield strength.

[0042] Step 203: Analyze the factors affecting tensile strength, including the content of Y, Mn, Zn, and Cu elements, extrusion temperature, homogenization temperature, solution temperature, and solution time. The results show that increasing the content of Y, Mn, and Zn significantly improves tensile strength, while solution temperature and solution time also have a significant impact on tensile strength.

[0043] Step 204: Analyze the factors affecting elongation after fracture, including the content of Gd, Y, and Zn elements, extrusion temperature, homogenization temperature, and aging temperature. Increases in Gd and Y will decrease elongation after fracture, while Zn content within a certain range will help increase elongation after fracture.

[0044] Step 205: By analyzing the importance of features, the gradient boosting decision tree algorithm is used to train the model and rank the key factors affecting mechanical properties. The results show that the contents of Gd, Y, Zn, and Zr have the greatest impact on mechanical properties, while process parameters such as solution temperature, solution time, and extrusion ratio also have a significant impact on mechanical properties.

[0045] Step 3: Construct a machine learning model based on the prediction performance of composition and process: Step 301: Select the gradient boosting decision tree algorithm as the machine learning algorithm to establish a preliminary prediction model. Use cross-validation to select the optimal model parameters and train an accurate prediction model.

[0046] Step 302: Optimize the model's hyperparameters to improve model accuracy. A grid search algorithm is used to scan the model's hyperparameter space to find the optimal combination of hyperparameters. After optimization, the actual and predicted values ​​of the yield strength, tensile strength, and elongation prediction models are compared as shown in the following figure. Figures 4 to 6 As shown, the R values ​​of the three models are... 2 The values ​​are 0.962, 0.977, and 0.677, respectively.

[0047] Step 303: Employ particle swarm optimization (PSO) for multi-objective optimization to achieve coordinated optimization of composition and process parameters. Initialize the population and iteratively calculate to find the optimal solution until convergence is achieved.

[0048] Step 4: Composition optimization based on predicted mechanical properties: Step 401: Within the range of Gd: 2~12, Y: 0~8, Zn: 0~4, Zr: 0~2, perform component optimization based on a genetic algorithm. The objective can be to maximize the yield strength, tensile strength, and elongation simultaneously. New individuals are generated through crossover and mutation operations, and surviving individuals are selected through competitive selection.

[0049] Step 402: Based on the series of alloy compositions obtained through optimization, their yield strength, tensile strength, and elongation are as follows: Figure 7As shown, referring to the composition and performance data in relevant patents and literature, the Mg-8.1Gd-3.9Y-1.0Zn-0.39Zr alloy was selected as the design composition. The target properties are yield strength ≥395 MPa, tensile strength ≥469 MPa, and elongation after fracture ≥13.5%.

[0050] Step 5: Optimize process parameters: Step 501: Optimize process parameters such as extrusion ratio, extrusion temperature, homogenization temperature, and solution temperature. Response surface methodology was used to analyze the impact of each parameter on mechanical properties, and the optimal combination of process parameters was determined to be: homogenization annealing temperature 520℃, homogenization annealing time 14 h, extrusion ratio of 9, extrusion temperature 500℃, solution temperature 500℃, and solution time 5 h.

[0051] Step 502: High-strength and tough alloys with different properties are prepared by using different aging temperatures and aging times. By controlling the temperature and time during the aging process, the microstructure of the material is adjusted. The optimal aging process combination is an aging temperature of 200℃ and an aging time of 72 h.

[0052] Step 503: The Mg-8.1Gd-3.9Y-1.0Zn-0.39Zr alloy prepared by the above process has the following mechanical properties: yield strength ≥399MPa, tensile strength ≥471 MPa, and elongation after fracture ≥13.6%.

[0053] Example 2 This embodiment provides a machine learning-based method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys. The specific steps are as follows: Step 1: Data Collection and Preprocessing Step 101: Collect composition, processing, and mechanical property data of Mg-Gd-Y-Zn magnesium alloys from literature to construct a database of high-strength forged rare-earth magnesium alloys. For example... Figure 1 As shown, the database specifically includes the composition content of Gd: 0~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%, Ce: 0~0.26 wt%, La: 0~0.3 wt%, Nd: 0~3.1 wt%, Ni: 0~1 wt%, Cu: 0~1 wt%, Mn: 0~2.56 wt%, Ti: 0~0.62 wt%, solution temperature: 500~600℃, solution time: 10~30 hours, extrusion ratio: 10~20, extrusion temperature: 400~500℃, and other process parameters, as well as mechanical property indicators such as yield strength: 250~550 MPa, tensile strength: 350~650 MPa, and elongation after fracture: 5~15%.

[0054] Step 102: Clean and standardize the collected data, remove outliers and duplicate data, and normalize the data of each indicator to make it range from 0 to 1.

[0055] Step 103: Extract key features and construct a feature matrix of components and process parameters. The contents of Gd, Y, Zn, and Zr are selected as component features, while solution temperature, solution time, extrusion ratio, and extrusion temperature are selected as process features.

[0056] Step 2, Analysis of Factors Affecting Mechanical Properties: Step 201: Through Pearson correlation coefficient analysis, the influence of several characteristics on yield strength, tensile strength, and elongation was obtained, such as... Figure 2 As shown in the figure. For yield strength, Zn, Mn, Y, extrusion temperature, and homogenization temperature are positively correlated with it, while solution temperature and solution time are negatively correlated. In particular, the addition of Zn and Mn significantly improves yield strength. For tensile strength, Y, Mn, Zn, Cu, extrusion temperature, and homogenization temperature are positively correlated with it, while solution temperature and solution time are negatively correlated. In particular, increasing Y, Mn, and Zn significantly improves tensile strength. For elongation, Zr, extrusion temperature, and extrusion ratio have a positive effect, while Zn, Ti, and aging temperature have a negative correlation.

[0057] Step 202: Analyze the factors affecting yield strength, including the content of Zn, Mn, and Y elements, extrusion temperature, homogenization temperature, solution temperature, and solution time. Linear regression analysis shows that increasing the content of Zn and Mn significantly improves yield strength, while excessively long solution temperatures and times decrease yield strength.

[0058] Step 203: Analyze the factors affecting tensile strength, including the content of Y, Mn, Zn, and Cu elements, extrusion temperature, homogenization temperature, solution temperature, and solution time. The results show that increasing the content of Y, Mn, and Zn significantly improves tensile strength, while solution temperature and solution time also have a significant impact on tensile strength.

[0059] Step 204: Analyze the factors affecting elongation after fracture, including the content of Gd, Y, and Zn elements, extrusion temperature, homogenization temperature, and aging temperature. Increases in Gd and Y will decrease elongation after fracture, while Zn content within a certain range will help increase elongation after fracture.

[0060] Step 205: By analyzing the importance of features, the gradient boosting decision tree algorithm is used to train the model and rank the key factors affecting mechanical properties. The results show that the contents of Gd, Y, Zn, and Zr have the greatest impact on mechanical properties, while process parameters such as solution temperature, solution time, and extrusion ratio also have a significant impact on mechanical properties.

[0061] Step 3: Construct a machine learning model based on the prediction performance of composition and process: Step 301: Select the gradient boosting decision tree algorithm as the machine learning algorithm to establish a preliminary prediction model. Use cross-validation to select the optimal model parameters and train an accurate prediction model.

[0062] Step 302: Optimize the model's hyperparameters to improve model accuracy. A grid search algorithm is used to scan the model's hyperparameter space to find the optimal combination of hyperparameters. After optimization, the actual and predicted values ​​of the yield strength, tensile strength, and elongation prediction models are compared as shown in the following figure. Figures 4 to 6 As shown, the R values ​​of the three models are... 2 The values ​​are 0.962, 0.977, and 0.677, respectively.

[0063] Step 303: Employ particle swarm optimization (PSO) for multi-objective optimization to achieve coordinated optimization of composition and process parameters. Initialize the population and iteratively calculate to find the optimal solution until convergence is achieved.

[0064] Step 4: Composition optimization based on predicted mechanical properties: Step 401: Within the range of Gd: 2~12, Y: 0~8, Zn: 0~4, Zr: 0~2, perform component optimization based on a genetic algorithm. The objective can be to maximize the yield strength, tensile strength, and elongation simultaneously. New individuals are generated through crossover and mutation operations, and surviving individuals are selected through competitive selection.

[0065] Step 402: Based on the series of alloy compositions obtained through optimization, their yield strength, tensile strength, and elongation are as follows: Figure 7 As shown, referring to the composition and performance data in relevant patents and literature, the Mg-8.1Gd-3.9Y-1.0Zn-0.39Zr alloy was selected as the design composition. The target properties are yield strength ≥395 MPa, tensile strength ≥469 MPa, and elongation after fracture ≥13.5%.

[0066] Step 403: Add 0.3 wt% Ce element to the design composition. The solute growth is restricted and the nucleation of the second phase is promoted to refine the as-cast structure. The solute is used to form pinned recrystallization grains by segregating at the grain boundaries.

[0067] Step 5: Optimize process parameters: Step 501: Optimize process parameters such as extrusion ratio, extrusion temperature, homogenization temperature, and solution temperature. Response surface methodology was used to analyze the impact of each parameter on mechanical properties, and the optimal combination of process parameters was determined to be: homogenization annealing temperature 520℃, homogenization annealing time 16 h, extrusion ratio 10, extrusion temperature 500℃, solution temperature 510℃, and solution time 6 h.

[0068] Step 502: High-strength and tough alloys with different properties are prepared by using different aging temperatures and aging times. By controlling the temperature and time during the aging process, the microstructure of the material is adjusted. The optimal aging process combination is an aging temperature of 200℃ and an aging time of 48 h.

[0069] Step 503: The Mg-8.1Gd-3.9Y-1.0Zn-0.39Zr-0.3Ce alloy prepared by the above process has the following mechanical properties: yield strength ≥406 MPa, tensile strength ≥486 MPa, and elongation after fracture ≥13.9%.

[0070] Example 3 This embodiment provides a machine learning-based method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys. The specific steps are as follows: Step 1: Data Collection and Preprocessing Step 101: Collect composition, processing, and mechanical property data of Mg-Gd-Y-Zn magnesium alloys from literature to construct a database of high-strength forged rare-earth magnesium alloys. For example... Figure 1 As shown, the database specifically includes the composition content of Gd: 0~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%, Ce: 0~0.26 wt%, La: 0~0.3 wt%, Nd: 0~3.1 wt%, Ni: 0~1 wt%, Cu: 0~1 wt%, Mn: 0~2.56 wt%, Ti: 0~0.62 wt%, solution temperature: 500~600℃, solution time: 10~30 hours, extrusion ratio: 10~20, extrusion temperature: 400~500℃, and other process parameters, as well as mechanical property indicators such as yield strength: 250~550 MPa, tensile strength: 350~650 MPa, and elongation after fracture: 5~15%.

[0071] Step 102: Clean and standardize the collected data, remove outliers and duplicate data, and normalize the data of each indicator to the range of [0, 1].

[0072] Step 103: Extract key features and construct a feature matrix of components and process parameters. The contents of Gd, Y, Zn, and Zr are selected as component features, and solution temperature, solution time, extrusion ratio, and extrusion temperature are selected as process features. The first seven principal components are extracted using principal component analysis.

[0073] Step 2, Analysis of Factors Affecting Mechanical Properties: Step 201: Through Pearson correlation coefficient analysis, the influence of several characteristics on yield strength, tensile strength, and elongation was obtained, such as... Figure 2As shown in the figure. For yield strength, Zn, Mn, Y, extrusion temperature, and homogenization temperature are positively correlated with it, while solution temperature and solution time are negatively correlated. In particular, the addition of Zn and Mn significantly improves yield strength. For tensile strength, Y, Mn, Zn, Cu, extrusion temperature, and homogenization temperature are positively correlated with it, while solution temperature and solution time are negatively correlated. In particular, increasing Y, Mn, and Zn significantly improves tensile strength. For elongation, Zr, extrusion temperature, and extrusion ratio have a positive effect, while Zn, Ti, and aging temperature have a negative correlation.

[0074] Step 202: Analyze the factors affecting yield strength. Multinomial regression analysis shows that a Zn content of 2–4 wt% significantly affects yield strength, a Y content of 3–7 wt% helps improve yield strength, and excessively high Gd content reduces yield strength. Excessively high solution temperature or excessively long solution time also reduces yield strength.

[0075] Step 203: Analyze the factors affecting tensile strength. Through multinomial regression analysis, it was found that the Y content (2-6 wt%) has a significant impact on tensile strength, while the Zn content (1-3 wt%) helps improve tensile strength. Increased Gd and Cu content also enhance tensile strength. Solution temperature and solution time also have a significant impact on tensile strength.

[0076] Step 204: Analyze the factors affecting elongation after fracture. Linear regression analysis shows that excessively high Gd content reduces elongation after fracture, while moderate increases in Y and Zn content help improve it. Excessively high extrusion temperature also reduces elongation after fracture, and homogenization temperature and aging temperature also affect it.

[0077] Step 205: By analyzing feature importance, the random forest algorithm is used to train the model, ranking the key factors affecting mathematical performance, such as... Figure 3 As shown in the figure, the results indicate that the contents of Gd, Y, Zn, and Zr have the greatest impact on mechanical properties, followed by process parameters such as solution temperature, solution time, and extrusion ratio.

[0078] Step 3: Construct a machine learning model based on the prediction performance of composition and process: Step 301: Select the Random Forest algorithm as the machine learning algorithm to establish a preliminary prediction model. Use 5-fold cross-validation to select the optimal model parameters and train an accurate prediction model.

[0079] Step 302: Optimize the model's hyperparameters to improve model accuracy. The model's hyperparameter space is scanned using the Grid Search algorithm to find the optimal hyperparameter combination, with a maximum depth of 5 and a maximum number of leaf nodes of 10. The comparison between the actual and predicted values ​​of the yield strength, tensile strength, and elongation prediction model after optimization is shown in the following figure. Figures 4 to 6 As shown, the R values ​​of the three models are... 2 The values ​​are 0.962, 0.977, and 0.677, respectively.

[0080] Step 303: Employ particle swarm optimization (PSO) for multi-objective optimization to achieve coordinated optimization of composition and process parameters. Initialize a population of 1000 and iteratively calculate to find the optimal solution until convergence is achieved.

[0081] Step 4: Composition optimization based on predicted mechanical properties: Step 401: Within the range of Gd: 2~12, Y: 0~8, Zn: 0~4, Zr: 0~2, perform component optimization based on genetic algorithm. The objective can be to maximize the yield strength, tensile strength and elongation simultaneously.

[0082] Step 402: Based on the series of alloy compositions obtained through optimization, their yield strength, tensile strength, and elongation are as follows: Figure 7 As shown, referring to the composition and performance data in relevant patents and literature, the Mg-8.7Gd-3.6Y-1.0Zn-0.24Zr alloy was selected as the design composition. The target properties are yield strength ≥380 MPa, tensile strength ≥465 MPa, and elongation after fracture ≥15.5%.

[0083] Step 403: Add 0.3 wt% Ce element to the design composition. The solute growth is restricted and the nucleation of the second phase is promoted to refine the as-cast structure. The solute is used to form pinned recrystallization grains by segregating at the grain boundaries.

[0084] Step 5: Optimize process parameters: Step 501: Optimize process parameters such as extrusion ratio, extrusion temperature, homogenization temperature, and solution temperature. Response surface methodology was used to analyze the impact of each parameter on mechanical properties, and the optimal combination of process parameters was determined to be: homogenization annealing temperature 510℃, homogenization annealing time 13 h, extrusion ratio 14, extrusion temperature 500℃, solution temperature 500℃, and solution time 14 h.

[0085] Step 502: High-strength and tough alloys with different properties are prepared by using different aging temperatures and aging times. By controlling the temperature and time during the aging process, the microstructure of the material is adjusted. The optimal aging process combination is an aging temperature of 175℃ and an aging time of 48h.

[0086] Step 503: The Mg-8.1Gd-3.9Y-1.0Zn-0.39Zr-0.3Ce alloy prepared by the above process has the following mechanical properties: yield strength ≥397 MPa, tensile strength ≥481 MPa, and elongation after fracture ≥15.8%.

[0087] It is evident that the alloy prepared by the method of the present invention, based on the predicted alloy composition and process parameters, possesses excellent mechanical properties, significantly improving the overall performance of rare earth magnesium alloys.

[0088] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare-earth magnesium alloys based on machine learning, characterized in that, Includes the following steps: S1. Data Collection and Preprocessing: Collect composition data, process parameter data and corresponding mechanical property data of Mg-Gd-Y-Zn magnesium alloys, construct an alloy database, and clean and standardize the data to construct a feature matrix containing compositional and process characteristics. S2. Analysis of factors affecting mechanical properties: Based on the feature matrix, machine learning methods are used to analyze the influence of the characteristics of each component and process on the preset mechanical performance indicators, and to determine the key factors affecting each mechanical performance indicator; S3. Construct a performance prediction model and perform collaborative optimization: Use component parameters and process parameters as input features and mechanical performance indicators as output to train a machine learning prediction model; An optimization algorithm is employed to perform collaborative multi-objective optimization of the composition and process parameters in order to obtain an optimized solution set that meets the preset performance objectives. S4. Composition optimization design based on prediction results: Within the preset composition range, composition optimization is performed based on the optimization algorithm, and based on the optimization results, an alloy composition that meets the target mechanical properties is designed. S5. Process Parameter Optimization: Based on the prediction model and optimization algorithm, optimize the key process parameters and output the designed alloy composition and its corresponding key process parameters.

2. The method according to claim 1, characterized in that, In step S1: The composition data includes at least the elemental contents of Gd, Y, Zn, and Zr; The process parameter data includes one or more of the following: solution temperature, solution time, extrusion ratio, and extrusion temperature. The mechanical properties include at least one of yield strength, tensile strength, and elongation after fracture.

3. The method according to claim 1 or 2, characterized in that, In step S1, the range of the composition data includes: Gd: 0~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%; the range of the process parameters includes: solution temperature: 500~600℃, solution time: 10~30 hours, extrusion ratio: 10~20, extrusion temperature: 400~500℃; the range of the mechanical properties includes: yield strength: 250~550 MPa, tensile strength: 350~650 MPa, elongation after fracture: 5~15%.

4. The method according to any one of claims 1 to 3, characterized in that, Step S2 involves identifying key influencing factors affecting various mechanical performance indicators by constructing a prediction model using a random forest algorithm or a gradient boosting decision tree algorithm, and ranking the features output by the model based on their importance to determine the key factors affecting mechanical performance.

5. The method according to any one of claims 1 to 4, characterized in that, In step S3: The machine learning prediction model is a model built on a tree ensemble model, and hyperparameter optimization is performed using cross-validation and parameter search. The optimization algorithm is a swarm intelligence optimization algorithm, used to perform collaborative multi-objective optimization of component parameters and process parameters; Preferably, the tree ensemble model is a random forest model, and the swarm intelligence optimization algorithm is a particle swarm optimization algorithm.

6. The method according to any one of claims 1 to 5, characterized in that, Step S3 includes: establishing and optimizing a machine learning prediction model; and using optimization algorithms to perform synergistic multi-objective optimization of composition and process parameters.

7. The method according to any one of claims 1 to 6, characterized in that, In step S4, the component optimization is performed within a preset component range based on a genetic algorithm; Preferably, the preset composition range is: Gd: 2~12 wt%, Y: 0~8 wt%, Zn: 0~4 wt%, Zr: 0~2 wt%, with the remainder being Mg and unavoidable impurities; Preferably, in step S4, the target mechanical properties of the optimized target composition alloy are: yield strength ≥ 395 MPa, tensile strength ≥ 469 MPa, and elongation after fracture ≥ 13.5%.

8. The method according to any one of claims 1 to 7, characterized in that, Step S5 includes: The effects of extrusion ratio, extrusion temperature, homogenization temperature, and solution temperature on mechanical properties were analyzed using process parameter optimization methods to determine the optimal combination of process parameters. Furthermore, the strength and toughness of the alloy were adjusted by controlling the aging temperature and time, as well as by inhibiting or promoting recrystallization mechanisms.

9. A high-strength wrought rare-earth magnesium alloy, characterized in that, The alloy is designed by the method of any one of claims 1 to 8, and its constituent elements and element mass percentages are as follows: Gd: 6-9%, Y: 4-6%, Zn: 0.5-3%, Zr: 0.2-0.5%, and the balance Mg.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the program is executed by the processor, it implements the steps of the method for optimizing the composition and predicting the mechanical properties of high-strength deformable rare earth magnesium alloys based on machine learning, as described in any one of claims 1 to 8.

Images