Alloy material component screening method, system, equipment and medium

Abstract

The invention relates to the technical field of alloy materials, and discloses an alloy material component screening method, system and equipment and a medium. The method comprises the following steps: constructing an alloy component performance data set; constructing a multi-model performance prediction model, and training the multi-model performance prediction model according to the alloy component performance data set to screen out a performance prediction reference model; performing global search according to a genetic algorithm and the performance prediction reference model to obtain an alloy component scheme and a corresponding performance prediction result; according to the alloy component scheme and the performance prediction result, weighted assignment sorting is carried out to screen out a target alloy combination; and performing simulation verification and process analysis on the target alloy combination to optimize the performance prediction reference model. According to the method, deviation of a single model is avoided through a multi-model performance prediction model, the genetic algorithm is introduced to ensure global search and avoid local optimization, the closed-loop process design is formed by combining simulation verification and process analysis, the target alloy combination meeting the requirements is rapidly and accurately screened, and the development efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of alloy materials technology, and in particular to a method, system, equipment and medium for screening alloy material components. Background Technology

[0002] In recent years, with the rapid development of high-performance fields such as aerospace, automobile manufacturing, and rail transportation, the demand for lightweight and high-strength materials has been increasing. Aluminum-based alloys, due to their advantages such as low density, high specific strength, good machinability, and strong corrosion resistance, have become a research hotspot for key structural materials.

[0003] The development of traditional Al-based alloy materials mainly relies on iterative experimental testing based on experience. However, due to the large number of elements and the complexity and nonlinearity of their interactions, there are often tens of thousands of alloy types composed of various alloy compositions. Repeated enumeration and experimental testing result in enormous time and cost losses, and the designed alloys often do not achieve ideal performance. Furthermore, traditional experimental methods are mostly single-objective optimization problems, making it difficult to simultaneously achieve multiple objectives such as low density, high strength, and high modulus. Summary of the Invention

[0004] The main objective of this invention is to provide a method, system, device, and medium for screening the composition of alloy materials, aiming to solve at least one of the aforementioned technical problems.

[0005] In a first aspect, embodiments of the present invention provide a method for screening the composition of alloy materials, comprising:

[0006] Construct an alloy composition and performance dataset;

[0007] A multi-model performance prediction model is constructed, and the multi-model performance prediction model is trained based on the alloy composition performance dataset to select a performance prediction benchmark model.

[0008] A global search is performed based on the genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results.

[0009] The alloy composition scheme and performance prediction results are weighted and ranked to select the target alloy combination;

[0010] The target alloy combination was simulated and the process was analyzed to optimize the performance prediction benchmark model.

[0011] In some embodiments, constructing the alloy composition performance dataset includes:

[0012] Collect data on the alloying element composition, mechanical properties, and heat treatment process of the target alloy;

[0013] The composition of alloying elements, mechanical properties, and heat treatment process data were analyzed to obtain element quality and optimization targets.

[0014] The quality of the elements and the optimization objectives are standardized.

[0015] An alloy composition performance dataset was constructed based on the standardized element quality and optimization objectives.

[0016] In some embodiments, the step of constructing a multi-model performance prediction model, training the multi-model performance prediction model based on the alloy composition performance dataset, and selecting a performance prediction benchmark model includes:

[0017] A multi-model performance prediction model is constructed based on several machine learning models;

[0018] The multi-model performance prediction model is trained based on the alloy composition and performance dataset to obtain the trained model.

[0019] Evaluation indicators are constructed based on mean absolute error, mean square error, and coefficient of determination.

[0020] The trained models are selected based on the evaluation metrics to obtain the performance prediction benchmark model.

[0021] In some embodiments, the multi-model performance prediction model includes a backpropagation neural network, a support vector machine model, and a random forest model; the performance prediction benchmark model includes a backpropagation neural network.

[0022] In some embodiments, the step of performing a global search based on a genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results includes:

[0023] Genetic algorithms are used to set chromosome encoding methods, determine component search boundaries, and configure genetic algorithm control parameters.

[0024] Using the performance prediction benchmark model as an evaluation tool, the predicted performance of each alloy individual composition is calculated, and a multi-objective comprehensive score function is constructed as the fitness function of the genetic algorithm.

[0025] Perform selection, crossover, and mutation genetic operations;

[0026] Based on the genetic algorithm, an iterative search is performed to obtain the alloy composition scheme and corresponding performance prediction results for each individual.

[0027] In some embodiments, the step of weighting and ranking the alloy composition scheme and performance prediction results to select target alloy combinations includes:

[0028] The performance prediction results are normalized to obtain the normalized index score;

[0029] Obtain the weights of each performance metric;

[0030] The weighted score is calculated based on the normalized index score and weight;

[0031] The alloy composition schemes are sorted from high to low based on the weighted scores to select target alloy combinations.

[0032] In some embodiments, calculating the weighted score based on the normalized indicator score and weights includes:

[0033] The density score, Young's modulus score, and tensile strength score are determined based on the normalized index scores.

[0034] The density weight, Young's modulus weight, and tensile strength weight are determined based on the aforementioned weights.

[0035] The weighted score is obtained by summing the products of the density score and density weight, the Young's modulus score and Young's modulus weight, and the tensile strength score and tensile strength weight.

[0036] Secondly, embodiments of the present invention provide an alloy material composition screening system, comprising:

[0037] The data acquisition module is used to build alloy composition and performance datasets;

[0038] The model building module is used to build a multi-model performance prediction model and train the multi-model performance prediction model based on the alloy composition performance dataset to select the performance prediction benchmark model.

[0039] The search optimization module is used to perform a global search based on the genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results.

[0040] The weighted scoring module is used to perform weighted scoring and sorting based on the alloy composition scheme and performance prediction results in order to select the target alloy combination;

[0041] The simulation verification module is used to perform simulation verification and process analysis on the target alloy combination in order to optimize the performance prediction benchmark model.

[0042] Thirdly, embodiments of the present invention provide an electronic device, including:

[0043] One or more processors;

[0044] Memory, used to store one or more programs;

[0045] When the one or more programs are executed by the one or more processors, the one or more processors implement any of the methods described above.

[0046] Fourthly, embodiments of the present invention provide a computer-readable medium on which a computer program is stored, the computer program being executed by a processor to implement the steps of any of the methods described above.

[0047] This invention provides a method for screening alloy material compositions, comprising: constructing an alloy composition performance dataset; constructing a multi-model performance prediction model, training the multi-model performance prediction model based on the alloy composition performance dataset to screen out a performance prediction benchmark model; performing a global search based on a genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results; performing weighted scoring and ranking based on the alloy composition schemes and performance prediction results to screen out target alloy combinations; and performing simulation verification and process analysis on the target alloy combinations to optimize the performance prediction benchmark model. This invention constructs an alloy composition performance dataset based on a large amount of publicly available alloy composition and its corresponding mechanical property data. This dataset has a relatively low requirement in terms of quantity. By constructing a multi-model performance prediction model, biases arising from a single model are avoided. A genetic algorithm is introduced to ensure a global search, avoiding local optimization. Combined with simulation verification and process analysis, a closed-loop process design is formed, deeply exploring the nonlinear and complex mapping relationship between "composition-performance". This allows for the rapid and accurate screening of target alloy composition combinations, such as aluminum alloy composition combinations, that meet the requirements of low density, high modulus, and high strength, greatly improving the efficiency of developing new lightweight high-strength aluminum alloys. Attached Figure Description

[0048] Figure 1 A schematic flowchart of an alloy material composition screening method provided in an embodiment of the present invention;

[0049] Figure 2 This is a flowchart illustrating the implementation process of the embodiments of the present invention;

[0050] Figure 3 A structural block diagram of an alloy material composition screening system provided in an embodiment of the present invention;

[0051] Figure 4 This is a structural block diagram of an electronic device provided in an embodiment of the present invention.

[0052] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0053] To enable those skilled in the art to better understand the technical solutions of the present invention, exemplary embodiments of the present invention are described below in conjunction with the accompanying drawings, including various details of the embodiments of the present invention to aid understanding. These should be considered merely exemplary. Therefore, those skilled in the art should recognize that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present invention. Similarly, for clarity and brevity, descriptions of well-known functions and structures are omitted in the following description.

[0054] Where there is no conflict, the various embodiments of the present invention and the features thereof may be combined with each other.

[0055] As used herein, the term “and / or” includes any and all combinations of one or more related enumerated entries.

[0056] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used herein, the singular forms “a” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that when the terms “comprising” and / or “made of” are used in this specification, the presence of the stated feature, integral, step, operation, element, and / or component is specified, but the presence or addition of one or more other features, integrals, steps, operations, elements, components, and / or groups thereof is not excluded. Terms such as “connected” or “linked” are not limited to physical or mechanical connections but can include electrical connections, whether direct or indirect.

[0057] Unless otherwise specified, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art. It will also be understood that terms such as those defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art and the invention, and will not be interpreted as having an idealized or overly formal meaning unless expressly so defined herein.

[0058] In the technical solution of this invention, the collection, storage, use, processing, transmission, provision, and disclosure of user personal information all comply with relevant laws and regulations and do not violate public order and good morals. The use of user data in this technical solution follows relevant national laws and regulations (e.g., the "Information Security Technology - Personal Information Security Specification"). For example: appropriate measures are taken for personal information access control; restrictions are imposed on the display of personal information; the purpose of using personal information does not exceed the scope of direct or reasonable association; and explicit identity targeting is eliminated when using personal information to avoid precisely locating a specific individual.

[0059] In related technologies, to overcome the problems in the development of traditional Al-based alloy materials, material design methods based on machine learning (ML) have become a breakthrough direction in the field of materials genome engineering in recent years. Among them, data-driven modeling and high-throughput simulation can predict material properties in a short time and assist in composition optimization, and have been initially applied in multiple material systems such as amorphous alloys, high-temperature alloys, and functional materials.

[0060] Current technologies offer diverse solutions for structure-performance mapping, material composition prediction, and macroscopic material service prediction, but they generally suffer from the following shortcomings: machine learning models require large datasets and often rely on single-model predictions; furthermore, they lack the ability to perform weighted optimization on multiple performance targets (such as alloy density, Young's modulus, and tensile strength), and the inconsistent units between these targets prevent direct comparison of single-weighted processing, resulting in excessive manual intervention in the output results. Even when machine learning models design one or more optimized alloy compositions, there is still a lack of testing and verification of their alloy properties and post-heat treatment, and a closed-loop "design-screening-verification" process has not yet been established, making it difficult to quickly evaluate candidate alloy compositions.

[0061] For example, a general reverse computation method applied to the design optimization of materials for high-end equipment uses AI machine learning algorithms to predict the service life of materials under different operating environments, significantly reducing experimental costs and improving material characterization efficiency. It can also predict the service life and performance after processing and assembly during the equipment design phase. However, this method cannot solve multi-objective optimization problems, and the computational model requires a large amount of data, has poor adaptability to small sample scenarios, and does not incorporate the correlation between process parameters and performance, nor does it analyze the impact of process deviations on performance.

[0062] For example, a novel amorphous alloy design method based on machine learning uses a random forest model to train and test a divided dataset to obtain a final prediction model. This model is then used to predict the glass-forming ability of the alloy system, thereby significantly shortening the design cycle and reducing R&D costs. However, this approach is not suitable for multi-objective optimization problems, requires a large dataset, and lacks explicit composition and process predictions. It is only applicable to amorphous alloys, limiting its applicability.

[0063] For example, a method for designing and preparing high-performance titanium alloys involves data screening using Pearson correlation screening, recursive elimination, and feature importance ranking. A multi-objective optimization strategy using a non-dominated sorting genetic evolution algorithm is employed to collaboratively optimize multiple alloy properties. Weight coefficients are introduced into the NSGA-II algorithm to represent the importance of alloy properties, adjusting the weights of optimization objectives to adjust the optimization focus. Ultimately, a titanium alloy composition with optimal overall performance is obtained, and experimental verification and iterative optimization are performed. While this scheme includes weights, it lacks normalization and scoring calculations, increasing computation time in multi-objective optimization of multivariate high-dimensional compositions. Furthermore, the subsequent heat treatment process for the designed titanium alloy lacks theoretical analysis and still relies on experimental testing, leading to significant time and testing costs.

[0064] For example, a multi-component alloy composition design method based on machine learning and oriented towards performance requirements can leverage existing data on alloy composition and properties to unlock the implicit and complex relationship between composition and performance using machine learning techniques, achieving the goal of quickly and accurately designing alloy compositions according to performance requirements. However, this method requires a large amount of machine learning data, and the steepest descent method used is prone to local optima, thus missing the optimal composition and performance. Furthermore, the gradient calculation is extremely large when predicting multi-dimensional, high-dimensional alloy composition and performance, and conflicts may occur when optimizing target strength and conductivity, making it impossible to guarantee the optimization order.

[0065] Therefore, there is an urgent need to propose an intelligent design method that integrates machine learning, weighted scoring, multi-objective optimization algorithms, and heat treatment simulation tools to achieve efficient screening, performance prediction, and precise control of aluminum-based alloy compositions. This invention represents a systematic innovation addressing this gap in the field, possessing significant theoretical value and promising engineering applications.

[0066] To address at least one of the technical problems existing in the aforementioned related technologies, the present invention provides a method for screening the composition of alloy materials. Figure 1 This is a flowchart illustrating a method for screening the composition of alloy materials according to an embodiment of the present invention.

[0067] As an embodiment of the present invention, such as Figure 1 As shown, the alloy material composition screening method includes:

[0068] Step S100: Construct an alloy composition and performance dataset;

[0069] Step S200: Construct a multi-model performance prediction model, train the multi-model performance prediction model based on the alloy composition performance dataset, and select a performance prediction benchmark model;

[0070] Step S300: Perform a global search based on the genetic algorithm and the performance prediction benchmark model to obtain the alloy composition scheme and the corresponding performance prediction results;

[0071] Step S400: Perform weighted scoring and sorting based on the alloy composition scheme and performance prediction results to select the target alloy combination;

[0072] Step S500: Perform simulation verification and process analysis on the target alloy combination to optimize the performance prediction benchmark model.

[0073] It should be noted that the execution subject in this embodiment can be an electronic device, which can be a computer device with data processing function, or other devices that can achieve the same or similar functions. This embodiment does not limit this. In this embodiment, a computer device is used as an example for explanation.

[0074] Understandably, in response to the problems in related technologies, such as the difficulty in simultaneously optimizing multi-objective performance, the lack of performance weighting and scoring, and the poor performance of composition-performance prediction on small datasets, the method proposed in this embodiment is an intelligent design method that integrates machine learning, weighted scoring, multi-objective optimization algorithms, and heat treatment simulation tools. By searching small datasets, the machine learning model can be trained, and multi-objective performance can be weighted, ranked, and predicted. It can be applied to the prediction of various alloy compositions and performances, and can simulate and calculate alloy compositions and performances, realizing a closed-loop process of "design-screening-verification". This enables efficient screening, performance prediction, and precise control of aluminum-based alloy compositions. The following describes the specific steps.

[0075] For example, this embodiment uses an aluminum-based alloy as an example for illustration.

[0076] In one embodiment, constructing an alloy composition and performance dataset includes: collecting data on the alloying element composition, mechanical properties, and heat treatment process of a target alloy; analyzing the alloying element composition, mechanical properties, and heat treatment process data to obtain element quality and optimization targets; standardizing the element quality and optimization targets; and constructing the alloy composition and performance dataset based on the standardized element quality and optimization targets.

[0077] Specifically, such as Figure 2As shown, S1: Collect Al alloy composition-performance data, establish an alloy composition-performance dataset, and determine optimization objectives. Through extensive research and literature review, collect typical alloying element compositions, corresponding mechanical properties, and heat treatment process data for aluminum-based alloys. The dataset can include two parts: model input features (elemental mass) and output objectives (performance indicators). A random partitioning method is used: training set (approximately 80%) and test set (approximately 20%). To ensure the target requirements of lightweight and high strength of the alloy, it is necessary to include the matrix composition and lightweight alloy composition (Al, Mg, Li) of Al-based alloys, at least three strengthening phase elements (e.g., Cu, Zn, Mn, etc.), and at least one grain refinement component (e.g., Sc, Zr, Ti, etc.). In this embodiment, the dataset used includes the above-mentioned main components, and some components also include Si. The optimization objectives include alloy density (ρ), Young's modulus (E), and tensile strength (σ). b The input feature (elemental mass) can include the mass percentage of various elements in the alloy, such as the mass percentage of Al, Mg, Li, Cu, Zn, Mn, Sc, Zr, Ti, Si, etc.

[0078] For example, the dataset (alloy composition performance dataset) can consist of only 50-100 sets, with the following composition ranges: Mg 0-14wt%, Li 0-3wt%, Cu 0-6wt%, Mn 0-5.5wt%, Zn 0-9wt%, Sc 0-1wt%, Zr 0-1.2wt%, Ti 0-0.6wt%, and Si 0-4.8wt%.

[0079] Specifically, such as Figure 2 As shown, S2: Optimization target normalization processing to eliminate differences in data dimensions. To eliminate differences in dimensions and magnitudes among elements and performance indicators in the data, this embodiment performs Min-Max standardization on all input and output data.

[0080] For example, Min-Max standardization is performed using the following formula:

[0081]

[0082] in, The normalized value; x i This is the original data; x max x min These are the maximum and minimum values ​​of this feature in the dataset.

[0083] It should be noted that most alloy design methods in related technologies focus on a single performance (such as tensile strength or hardness) as the optimization objective, neglecting the nonlinear correlation and coupling effects between various performance indicators (such as density, Young's modulus, and tensile strength). This lack of a systematic and comprehensive balancing method results in the designed material failing to simultaneously meet targets across multiple performance dimensions. To address the difficulty of synergistic optimization among multiple performance indicators, this embodiment establishes a strategy for unified expression and optimization of multiple performance indicators. It introduces performance normalization and weighted scoring methods to achieve synergistic design among objectives such as minimizing density, maximizing modulus, and maximizing strength, thereby improving the material's overall performance adaptability in practical engineering applications.

[0084] In one embodiment, a multi-model performance prediction model is constructed, and the multi-model performance prediction model is trained based on the alloy composition and performance dataset to select a performance prediction benchmark model. This includes: constructing a multi-model performance prediction model based on several machine learning models; training the multi-model performance prediction model based on the alloy composition and performance dataset to obtain a trained model; constructing evaluation indicators based on mean absolute error, mean square error, and coefficient of determination; and selecting the trained model based on the evaluation indicators to obtain a performance prediction benchmark model.

[0085] In one embodiment, the multi-model performance prediction model includes a backpropagation neural network, a support vector machine model, and a random forest model; the performance prediction benchmark model includes a backpropagation neural network.

[0086] Specifically, such as Figure 2 As shown in S3: Based on a standardized dataset, multiple machine learning models, including Backpropagation Neural Network (BPNN), Support Vector Machine (SVM), and Random Forest (RF), are created to construct a performance prediction model. This embodiment builds a performance prediction model (multi-model performance prediction model) based on a standardized dataset (alloy composition performance dataset). Model training can be performed using Mlatlab software and machine learning algorithms. Backpropagation Neural Network (BPNN), Support Vector Machine (SVM), and Random Forest (RF) models are constructed and trained, and their performance on the training and test sets is recorded. The optimal prediction model (performance prediction benchmark model) is selected through systematic evaluation.

[0087] For example, such as Figure 2 As shown, S4: Optimize the model hyperparameters using coefficients of determination (R²). 2The model performance is evaluated using two metrics: mean absolute error (MAE) and mean squared error (MSE). The model is trained on the training set, and hyperparameters are adjusted. During training, hyperparameters (e.g., the number of BPNN layers and neurons, SVM kernel type, number of RF trees, etc.) are adjusted based on cross-validation. The number of training iterations can be set to 1000, the learning efficiency can be set to 0.01, and the training objective is a minimum error of 0.000001. The mean absolute error (MAE), mean squared error (RMSE), and coefficient of determination (R²) are used to evaluate the model performance. 2 ), R 2 For models ∈ [0, 1], three evaluation metrics can be used to assess their performance. A smaller mean absolute error (MAE) indicates a closer approximation to the true value; a lower mean squared error (RMSE) indicates a smaller prediction error; and the coefficient of determination (R²) indicates a smaller prediction error. 2 A value closer to 1 indicates a better model, demonstrating a better fit. A horizontal comparison of the prediction performance of the three models on the training and test sets was conducted, selecting models with low MAE, low RMSE, and low R-value. 2 The high-performing model serves as the performance prediction benchmark model for subsequent optimization and scoring processes.

[0088] For example, the evaluation indicators used are the mean absolute error (MAE), the mean squared error (MSE), and the coefficient of determination (R²). 2 The three evaluation indicators are as follows:

[0089]

[0090] in, This represents the i-th true value; This represents the i-th predicted value; represents the mean of the true values; n represents the sample size.

[0091] In one example, the training results of three machine learning models, BPNN, SVM, and RF, are shown in Table 1:

[0092] Table 1

[0093]

[0094] Based on the comparison of the above training results, it can be seen that the BPNN model has the best training effect and accuracy. Therefore, this embodiment will use the BPNN machine learning model for training and prediction.

[0095] In one embodiment, a global search is performed based on a genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results. This includes: setting chromosome encoding methods, determining composition search boundaries, and configuring genetic algorithm control parameters using a genetic algorithm; using the performance prediction benchmark model as an evaluation tool to calculate the predicted performance of each alloy individual composition, and constructing a multi-objective comprehensive score function as the fitness function of the genetic algorithm; performing selection, crossover, and mutation genetic operations; and performing an iterative search based on the genetic algorithm to obtain the alloy composition scheme and corresponding performance prediction results for each individual.

[0096] It is understandable that aluminum-based alloys typically contain 5-10 or more alloying elements, resulting in a complex composition space and high search dimensionality. Related technologies using experimental or enumeration methods are extremely inefficient and cannot quickly locate high-performance compositional combinations. To address the issues of high dimensionality in the compositional search space and low optimization efficiency, this embodiment introduces a Genetic Algorithm (GA) intelligent optimization method. Under constraints on the compositional range, search direction, and step size, this method achieves efficient and stable alloy combination searching, reducing the design space and improving screening efficiency.

[0097] Specifically, such as Figure 2 As shown, S5: Introducing a Genetic Algorithm (GA) for search optimization. Using a BPNN machine learning model as the performance prediction benchmark, GA is employed for intelligent search optimization to clarify the types of aluminum-based alloy elements and the search range. Through iterative searching using GA, the element search space and process constraints, fitness function design, and genetic operator steps are progressively implemented to construct an intelligent alloy composition search system based on a genetic algorithm, specifically including:

[0098] 1. Set chromosome encoding method: encode the mass percentage of each element in the alloy as a continuous variable using floating-point numbers, with each individual representing a specific Al-based alloy composition combination.

[0099] 100 random gold composition combinations are generated, each requiring the sum of all element contents to equal 100% (Al is automatically filled in), and the percentage signs of the elements are encoded using floating-point numbers, as follows:

[0100] |---|---|---|---|---|---|---|---|---|---|---|

[0101] |1|5.0|1.5|3.0|2.0|4.0|0.3|0.4|0.2|1.0|82.6|

[0102] |2|8.5|2.2|4.8|1.5|6.0|0.5|0.3|0.1|0.8|75.3|

[0103] |3|12.0|3.5|2.5|3.0|2.0|0.2|0.6|0.3|0.5|75.4|

[0104] …

[0105] 2. Determine the composition search boundary: Based on literature data and industrial experience, set the upper and lower limits of each element (e.g., Mg, Li, Cu, Zn, Zr, etc.) to ensure the physical rationality and metallurgical feasibility of the composition space, and set the Al content to automatically complete so that the total is 100%.

[0106] In this embodiment, considering the limitations of dataset search, the search range of alloy composition is appropriately adjusted. Specifically, it is Mg (0-10wt%, to avoid generating a large amount of β-Al3Mg2 which affects alloy toughness), Li (0-4.0wt%, excessive addition easily forms coarse β-LiAl phase), Cu (0-4wt%, excessive addition easily reduces alloy toughness), Mn (0-4.0wt%, to avoid exceeding the solid solution limit too much, and the Al6Mn phase precipitated after excess is easy to coarse), Zn (0-6wt%, excessive content affects alloy density and toughness), Sc (0-1.0wt%, to avoid oversaturation leading to coarse grains), Zr (0-1.0wt%, to avoid oversaturation leading to coarse grains), Ti (0-0.6wt%, excessive addition of TiAl3 phase easily coarsens), and Si (0-2.0wt%, excessive addition easily reduces alloy strength).

[0107] 3. Configure genetic algorithm control parameters: For example, parameters such as population size (n=100), number of iterations (G=200), crossover rate (Pc=0.8), mutation rate (Pm=0.1), and component search step size (e.g., Δ=0.1%) can all be adjusted according to the target accuracy.

[0108] 4. Construct the fitness function: Using the performance prediction model (backpropagation neural network BPNN) selected in step S4 above as the evaluation tool, calculate the predicted performance of each individual alloy composition, and construct the following multi-objective comprehensive score function as the fitness.

[0109] The fitness function design in the genetic algorithm includes normalization and a comprehensive scoring function. These functions serve as an internal real-time evaluation function during the genetic algorithm's iteration process, evaluating the quality of each individual's alloy composition. In this embodiment, using normalization and a fitness function improves search efficiency and avoids blind searching.

[0110]

[0111] Where v is a performance prediction value; s v The normalized score of a certain performance index v; v minv max The minimum and maximum values ​​in the training set; w i The importance weight of this indicator, The weights of each performance metric are determined to satisfy the following conditions: .

[0112] The lower the data, the better; therefore, in this embodiment, the alloy density can be expressed using the formula... For inverse normalization, higher Young's modulus and tensile strength are better; therefore, the formula can be used. Forward normalization. For example, the weighting coefficients can be... .

[0113] Taking any individual in this embodiment as an example, the density ρ predicted by the backpropagation neural network (BPNN) is 2.65 g / cm³. 3 Young's modulus E = 72 GPa, tensile strength σ b =480MPa. v of the training set density ρ min =2.4g / cm 3 v max =2.9g / cm 3 Young's modulus E v min =65GPa, v max =100GPa, tensile strength σ b v min =400MPa, v max =720MPa.

[0114] After normalization, Fitness is calculated as .

[0115] 5. Select genetic operators: including standard genetic operations such as roulette wheel selection, arithmetic crossover (blend crossover), and Gaussian mutation, to ensure population diversity and the ability to search for the global optimum.

[0116] For example, since the alloy composition in this embodiment is a continuous variable (mass percentage), arithmetic cross can generate new individuals through linear combination, ensuring that the composition of the offspring is still within a reasonable continuous range, avoiding composition jumps caused by discrete cross (such as single-point cross) and non-compliance with metallurgical constraints (avoiding Mg > 14wt%, Li > 4.2wt%, etc.).

[0117] For example, arithmetic crossover can preserve the superior genes of the parent individuals (such as the Cu content range corresponding to high tensile strength) while increasing the diversity of offspring through parameter adjustment, thus balancing "local search accuracy" and "global exploration capability".

[0118] From the selected population, 50 pairs were randomly paired, with each pair representing the parent individual. and (k represents the type of alloying element, such as Al, Mg, Li, Cu, Mn, Zn, Sc, Zr, Ti, and Si, totaling 10 types).

[0119] Set the crossover coefficient α (α=0.5 is recommended, but can be adjusted according to the search precision), and generate two offspring individuals according to the following formula:

[0120]

[0121] Where j represents the j-th element, for example, j=1 corresponds to the Mg content, j=2 corresponds to the Li content, etc.

[0122] Check if the offspring composition meets metallurgical constraints (e.g., Li ≤ 3.5 wt%, Cu ≤ 4.0 wt%, Mn ≤ 4.0 wt%). If it exceeds these limits, truncate the content of that element to the upper / lower limit to ensure composition feasibility. If the content of a certain element is the same after crossover among selected individuals, Gaussian mutation can be applied to add a small perturbation to the content of that element in the selected individuals (e.g., Mg content ± 0.1 wt% based on 6.75 wt%) to ensure population diversity.

[0123] It should be noted that some data-driven design methods in related technologies only consider the accuracy of model predictions and ignore the metallurgical compatibility of alloys and actual manufacturing feasibility. For example, excessively high Li or Mg content may cause hot cracking, coarse grains, or deterioration of mechanical properties. To address the lack of clear manufacturability constraints in the composition design results, this embodiment sets metallurgical behavior and processing constraint rules during the composition search process (e.g., alloy components that significantly exceed the solid solution limit should be screened out) to ensure that the optimized alloy scheme has actual manufacturability.

[0124] In one example, the genetic algorithm optimization involves setting a chromosome encoding method, using the mass percentage of each element in the alloy as a continuous variable encoded as a floating-point number. Each individual represents a specific Al-based alloy composition combination, and a certain number of individuals are randomly generated to ensure population diversity. Based on literature data and industrial experience, upper and lower limits for the content of each element are set to ensure the physical rationality and metallurgical feasibility of the composition space. Genetic algorithm control parameters are configured. A selected performance prediction model (backpropagation neural network BPNN) is used as the evaluation tool to calculate the predicted performance of each alloy individual composition. A multi-objective comprehensive score function is constructed as the fitness, for example: density is inversely normalized because lower density is better; Young's modulus and tensile strength are forward normalized because higher values ​​of these two indicators are better; weights are assigned to each performance indicator, for example: density weight is 0.4, Young's modulus weight is 0.2, and tensile strength weight is 0.4. The fitness score of each individual is calculated; a higher score indicates a better alloy combination. Genetic selection operations include: Selection, using roulette wheel selection, which selects individuals based on their fitness scores; higher fitness increases the probability of selection. Crossover, using arithmetic crossover, generates new individuals through linear combinations, preserving superior genes from parents and increasing offspring diversity. Mutation, using Gaussian mutation, adds a small perturbation to the content of a certain element in selected individuals to ensure population diversity.

[0125] In one embodiment, the process of weighted scoring and ranking based on the alloy composition scheme and performance prediction results to select target alloy combinations includes: normalizing the performance prediction results to obtain normalized index scores; obtaining the weights of each performance index; calculating a weighted score based on the normalized index scores and weights; and ranking the alloy composition schemes from high to low based on the weighted scores to select target alloy combinations.

[0126] In one embodiment, calculating a weighted score based on the normalized index score and weights includes: determining a density score, a Young's modulus score, and a tensile strength score based on the normalized index score; determining a density weight, a Young's modulus weight, and a tensile strength weight based on the weights; and summing the products of the density score and density weight, the Young's modulus score and Young's modulus weight, and the tensile strength score and tensile strength weight to obtain the weighted score.

[0127] Specifically, such as Figure 2 As shown, S6: Select a weighted scoring method to ensure that the optimization target weight is controllable. A dual mechanism combining subjective experience and objective entropy weighting is used. To avoid the subjectivity of traditional single manual weighting methods, this embodiment introduces a dual weighting mechanism combining subjective weighting and objective entropy weighting.

[0128] For example, the subjective weighting method sets initial weight reference values ​​based on actual engineering needs. The objective entropy weighting method automatically assigns weights based on the dispersion of each performance index in the data. The smaller the entropy weight, the greater the difference in the index, indicating that the performance plays a stronger role in the scheme judgment and should be given a higher weight.

[0129]

[0130] Among them, e j Let m be the entropy value of the j-th indicator; m be the sample size; p ij w represents the proportion of the i-th sample in the j-th indicator; j Let be the weight of the j-th indicator; n be the total number of indicators. Entropy value e j The smaller the value, the closer the coefficient of variation is to 1; e j The larger the value, the higher the weight w. j The higher.

[0131] In this embodiment, due to the clear requirements for lightweighting and high strength, the subjective weighting method is selected. The weight selection can be consistent with step S5 above, and the weight coefficient can be... .

[0132] Specifically, such as Figure 2 As shown, S7: Prediction performance normalization and weighted scoring ranking. To achieve comprehensive evaluation and objective ranking of multi-objective performance, this embodiment, after completing the genetic algorithm search, reprocesses the performance data of all output alloy combinations, and comprehensively evaluates and ranks all output alloy composition schemes. Specifically:

[0133] Indicator normalization method: For performance indicators (E, σ) where larger values ​​are better, b ), using standard Min-Max normalization:

[0134]

[0135] For performance metrics where smaller values ​​are preferred (such as density ρ), inverse normalization is performed:

[0136]

[0137] Weighted scoring function:

[0138]

[0139] in, The score is the normalized indicator score; The weights of each performance indicator are 1, and the specific values ​​can be adjusted according to actual needs. This embodiment does not impose any restrictions on this.

[0140] For example, the scoring results are sorted from highest to lowest based on the comprehensive score. The top-3 optimal alloy combinations are selected for the next simulation verification stage. The three formulas used in the above index normalization method are then re-normalized and weighted for ranking, reducing extreme biases in the prediction process. In one example, after 200 iterations of the genetic algorithm, the top 3 alloy compositions and properties in the weighted ranking are as follows:

[0141] TOP1: Al-8.8Mg-2.8Li-2.8Cu-2.5Mn-2.5Zn-1.0Sc-0.8Zr-0.6Ti-0.5Si, density 2.54g / cm 3 Young's modulus is 102.5 GPa, and tensile strength is 768 MPa.

[0142] TOP2: Al-7.8-2.6Li-3.2Cu-2.8Mn-2.8Zn-1.0Sc-0.7Zr-0.6Ti-0.4Si, density 2.58g / cm 3 Young's modulus 96.8 GPa, tensile strength 761 MPa.

[0143] TOP3: Al-8.5-2.5Li-3.5Cu-2.5Mn-2.4Zn-0.8Sc-0.8Zr-0.5Ti-0.3Si, density 2.5g / cm 3 Young's modulus is 93.6 GPa and tensile strength is 736 MPa.

[0144] In this embodiment, one or more optimal alloy composition combinations are obtained through iterative search using a genetic algorithm. These combinations, while satisfying physical and metallurgical constraints, can achieve multi-objective optimization, such as reducing density, increasing Young's modulus, and tensile strength. A selected performance prediction model (e.g., BPNN) can be used to predict the performance of the optimal alloy composition combinations, obtaining corresponding performance indicators such as density, Young's modulus, and tensile strength. These performance prediction results can serve as a reference for subsequent simulation verification and process analysis.

[0145] In some embodiments, the target alloy combination is subjected to simulation verification and process analysis to optimize the performance prediction benchmark model.

[0146] Specifically, such as Figure 2 As shown, S8: Composition and performance simulation verification, process analysis and model evaluation. Simulation verification involves inputting the selected optimal combination of alloy composition, such as three optimal combinations, into JMatPro software for performance simulation calculations. The output includes density, Young's modulus, and Poisson's ratio. Simulation calculations are also performed on the heat treatment process to complete the closed-loop design of composition-performance prediction and processing technology. The model accuracy is then judged based on the tensile strength of the alloy after heat treatment.

[0147] For TOP1: Al-8.8Mg-2.8Li-2.8Cu-2.5Mn-2.5Zn-1.0Sc-0.8Zr-0.6Ti-0.5Si. The process involves hot forging after casting, with ingot homogenization treatment: 420℃×24h, followed by air cooling. This is mainly to eliminate casting segregation (high Mg easily leads to dendritic segregation) and ensure uniform distribution of elements such as Sc and Zr, laying the foundation for subsequent forging and precipitation strengthening. The initial forging temperature during the hot forging stage is 450℃, lower than the Mg-Li eutectic temperature (approximately 470℃), to avoid hot brittleness; the final forging temperature is 380℃, and the deformation is 70% (multi-pass forging, with 20-30% deformation per pass). A segmented solution treatment is used: 375℃×5h + 460℃×2h, followed by water quenching (cooling rate ≥50℃ / s). Aging treatment: 120℃×4h (pre-aging) + 175℃×16h (peak aging), air cooling. The simulated alloy density is 2.5 g / cm³. 3 The model has a Young's modulus of 99.4 GPa and a tensile strength of 753 MPa, with an overall error within 5%, indicating that the model has high accuracy.

[0148] For TOP2: Al-7.8-2.6Li-3.2Cu-2.8Mn-2.8Zn-1.0Sc-0.7Zr-0.6Ti-0.4Si. It employs a casting followed by hot forging process, with ingot homogenization treatment: 430℃ × 20h, followed by air cooling. 430℃ promotes Cu diffusion homogenization, while Mn dissolves to form the Al6Mn precursor. The initial forging temperature during the hot forging stage is 470℃, which can be increased to 470℃ (higher than TOP1) to improve the alloy's high-temperature stability, increase plasticity, and inhibit grain boundary migration. The final forging temperature is 400℃, with a deformation rate of 65% (two forging passes, 30-35% per pass). The final forging temperature of 400℃ ensures the precipitation of the Al6Mn phase during deformation, aiding in grain refinement, while the Sc / Zr phase remains stable at 470℃. Solution treatment: 500℃×3h, water quenching (cooling rate ≥60℃ / s); Aging treatment: 160℃×20h, air cooling. The simulated alloy density is 2.55 g / cm³. 3 The model has a Young's modulus of 98.5 GPa and a tensile strength of 748 MPa, with an overall error within 5%, indicating that the model has high accuracy.

[0149] For the TOP3 alloy: Al-8.5-2.5Li-3.5Cu-2.5Mn-2.4Zn-0.8Sc-0.8Zr-0.5Ti-0.3Si, a hot forging process following casting is employed. The ingot is homogenized at 440℃ for 18 hours, followed by air cooling. The high Cu and low Si content reduces low-melting-point phases (e.g., Al-Cu-Si), increasing the homogenization temperature to 440℃ and accelerating Cu diffusion. Simultaneously, Ti forms Al3Ti, refining the as-cast grains. The initial forging temperature during hot forging is 480℃, and the final forging temperature is 410℃. The deformation is 60% (large deformation in a single pass, utilizing the Sc / Zr phase to suppress recrystallization grain growth). Solution treatment is performed at 510℃ for 2 hours, followed by water quenching (cooling rate ≥55℃ / s). Aging treatment is performed at 180℃ for 12 hours, followed by air cooling. The simulated alloy density is 2.54 g / cm³. 3 The model has a Young's modulus of 97.7 GPa and a tensile strength of 721 MPa, with an overall error within 5%, indicating that the model has high accuracy.

[0150] Understandably, even if the model outputs multiple alloy combinations that may meet the performance requirements, the related technologies lack a unified comprehensive scoring mechanism to evaluate and rank these combinations, making decision-making reliant on human experience and highly subjective. To address the problem of not being able to uniformly and objectively rank candidate alloys for final performance, this embodiment introduces a multi-performance normalization and weighted scoring system to comprehensively rank all candidate schemes using deterministic indicators, identify the optimal combination, and improve the objectivity of the final design results and the efficiency of decision-making.

[0151] It should be noted that the method proposed in this embodiment is a multi-objective optimization weighted scoring ranking material-performance design method. It is based on a large amount of publicly available data on aluminum-based alloy compositions and their corresponding mechanical properties (density, Young's modulus, tensile strength). This dataset has a relatively low requirement in terms of quantity. By constructing a multi-model machine learning prediction system, biases arising from a single model are avoided. A GA genetic algorithm is introduced to ensure global search and avoid local optimization. Combined with heat treatment simulation tools, a closed-loop process design is formed, deeply exploring the nonlinear and complex mapping relationship between "composition and performance." This allows for the rapid and accurate selection of aluminum-based alloy composition combinations that meet the requirements of low density, high modulus, and high strength, greatly improving the efficiency of developing new lightweight high-strength aluminum alloys. The beneficial effects of this embodiment include:

[0152] (1) By introducing standard Min-Max performance normalization and weighted scoring mechanism, the dimensional differences of multiple objectives such as density, Young's modulus and tensile strength are eliminated by standard normalization to eliminate the positive data differences of the "larger is better" performance (Young's modulus and tensile strength), and by reverse normalization to eliminate the negative data differences of the "smaller is better" performance (density). This makes the optimization objectives more uniform and convenient to sort in the subsequent weighted scoring without manual intervention.

[0153] (2) The closed-loop design process of "machine learning + genetic algorithm + heat treatment simulation" solves the problem of missing manufacturability and verification of design results. Manufacturability constraint optimization of genetic algorithm: metallurgical constraint rules are set in the GA search stage, and arithmetic crossover (linear combination of parent components) and Gaussian mutation (small perturbation of element content) are adopted to ensure that the generated alloy composition meets the actual metallurgical feasibility and avoids the prediction problem of local optima. At the same time, it solves the problem of "only focusing on prediction accuracy and ignoring manufacturability" in related technologies. The closed-loop design of "data acquisition-model prediction-intelligent search-process simulation-accuracy verification" in this embodiment can evaluate the actual performance and processing feasibility of candidate alloys without relying on physical experiments, which greatly shortens the R&D cycle.

[0154] (3) A multi-model optimization mechanism adapted to small samples, and the method in this embodiment can be applied to multi-objective optimization problems of other alloy systems, solving the problem of performance prediction accuracy under small datasets. Three machine learning models, BPNN, SVM, and RF, are constructed, and MAE, MSE, and R are used. 2 Three indicators (e.g., BPNN density prediction R) 2 A horizontal comparison of the BPNN model (0.978 for BPNN and 0.9356 for SVM) revealed that it achieved the best accuracy with a small sample size, serving as the prediction benchmark. This embodiment exhibits good scalability and can also be applied to the intelligent design needs of other metal alloy systems, while requiring a relatively small dataset.

[0155] This embodiment provides a method for screening alloy material compositions, including: constructing an alloy composition performance dataset; constructing a multi-model performance prediction model, training the multi-model performance prediction model based on the alloy composition performance dataset to screen out a performance prediction benchmark model; performing a global search based on a genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results; performing weighted scoring and ranking based on the alloy composition schemes and performance prediction results to screen out target alloy combinations; and performing simulation verification and process analysis on the target alloy combinations to optimize the performance prediction benchmark model. This embodiment constructs an alloy composition performance dataset based on a large amount of publicly available alloy composition and its corresponding mechanical property data. This dataset has a relatively low requirement in terms of quantity. By constructing a multi-model performance prediction model, biases arising from a single model are avoided. A genetic algorithm is introduced to ensure a global search, avoiding local optimization. Combined with simulation verification and process analysis, a closed-loop process design is formed, deeply exploring the nonlinear and complex mapping relationship between "composition-performance". This allows for the rapid and accurate screening of target alloy composition combinations, such as aluminum alloy composition combinations, that meet the requirements of low density, high modulus, and high strength, greatly improving the efficiency of developing new lightweight high-strength aluminum alloys.

[0156] Reference Figure 3 , Figure 3 This is a structural block diagram of an embodiment of the alloy material composition screening system of the present invention. Figure 3 As shown, the alloy material composition screening system includes:

[0157] Data acquisition module 10 is used to construct alloy composition and performance datasets;

[0158] The model building module 20 is used to build a multi-model performance prediction model and train the multi-model performance prediction model based on the alloy composition performance dataset to select a performance prediction benchmark model.

[0159] The search optimization module 30 is used to perform a global search based on the genetic algorithm and the performance prediction benchmark model to obtain the alloy composition scheme and the corresponding performance prediction results.

[0160] The weighted scoring module 40 is used to perform weighted scoring and sorting based on the alloy composition scheme and performance prediction results in order to screen out the target alloy combination;

[0161] The simulation verification module 50 is used to perform simulation verification and process analysis on the target alloy combination in order to optimize the performance prediction benchmark model.

[0162] For example, this embodiment uses aluminum-based alloys as an example for illustration. The system proposed in this embodiment adopts a multi-objective optimization weighted scoring ranking of material-property for aluminum-based alloy design. It is based on a large amount of publicly available data on aluminum-based alloy composition and its corresponding mechanical properties (density, Young's modulus, tensile strength). The data set requirement is relatively low. By constructing a multi-model machine learning prediction system, the bias of a single model is avoided. A GA genetic algorithm is introduced to ensure global search and avoid local optimization. Combined with heat treatment simulation tools, a closed-loop process design is formed. The nonlinear complex mapping relationship between "composition-property" is deeply explored, which can quickly and accurately screen aluminum-based alloy composition combinations that meet the requirements of low density, high modulus, and high strength, greatly improving the efficiency of developing new lightweight high-strength aluminum alloys.

[0163] The alloy material composition screening system provided in this embodiment constructs an alloy composition performance dataset based on a large amount of publicly available alloy composition and its corresponding mechanical property data. This dataset has a low requirement for quantity. By constructing a multi-model performance prediction model, biases caused by a single model are avoided. A genetic algorithm is introduced to ensure global search and avoid local optimization. Combined with simulation verification and process analysis, a closed-loop process design is formed. The system deeply explores the nonlinear and complex mapping relationship between "composition-performance". It can quickly and accurately screen target alloy composition combinations that meet the requirements of low density, high modulus, and high strength, such as aluminum alloy composition combinations, which greatly improves the efficiency of developing new lightweight and high-strength aluminum alloys.

[0164] It should be noted that technical details not described in detail in the embodiments of this alloy material composition screening system can be found in any embodiment of the present invention applied to the alloy material composition screening method as described above, and will not be repeated here.

[0165] Based on the same inventive concept, embodiments of the present invention also provide an electronic device. Figure 4 This is a structural block diagram of an electronic device provided in an embodiment of the present invention. Figure 4 As shown, an embodiment of the present invention provides an electronic device including: one or more processors 101, a memory 102, and one or more I / O interfaces 103. The memory 102 stores one or more programs, which, when executed by the one or more processors, cause the one or more processors to implement any of the alloy material composition screening methods described in the above embodiments; the one or more I / O interfaces 103 are connected between the processor and the memory, configured to enable information interaction between the processor and the memory.

[0166] The processor 101 is a device with data processing capabilities, including but not limited to a central processing unit (CPU); the memory 102 is a device with data storage capabilities, including but not limited to random access memory (RAM, more specifically SDRAM, DDR, etc.), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), and flash memory (FLASH); the I / O interface (read / write interface) 103 is connected between the processor 101 and the memory 102, and can realize information interaction between the processor 101 and the memory 102, including but not limited to a data bus (Bus).

[0167] In some embodiments, the processor 101, memory 102, and I / O interface 103 are interconnected via bus 104, and thus connected to other components of the computing device.

[0168] In some embodiments, the one or more processors 101 include a field-programmable gate array.

[0169] This invention also provides a computer-readable medium. The computer-readable medium stores a computer program, which, when executed by a processor, implements the steps in any of the alloy material composition screening methods described in the above embodiments. The computer-readable storage medium can be volatile or non-volatile.

[0170] This invention also provides a computer program product, including computer-readable code, or a non-volatile computer-readable storage medium carrying computer-readable code. When the computer-readable code is run in the processor of an electronic device, the processor in the electronic device executes the above-described alloy material composition screening method.

[0171] Those skilled in the art will understand that all or some of the steps, systems, and apparatuses disclosed above, and their functional modules / units, can be implemented as software, firmware, hardware, or suitable combinations thereof. In hardware implementations, the division between functional modules / units mentioned above does not necessarily correspond to the division of physical components; for example, a physical component may have multiple functions, or a function or step may be performed collaboratively by several physical components. Some or all physical components may be implemented as software executed by a processor, such as a central processing unit, digital signal processor, or microprocessor, or as hardware, or as an integrated circuit, such as an application-specific integrated circuit (ASIC). Such software can be distributed on a computer-readable storage medium, which may include computer storage media (or non-transitory media) and communication media (or transient media).

[0172] As is known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable program instructions, data structures, program modules, or other data. Computer storage media includes, but is not limited to, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), static random access memory (SRAM), flash memory or other memory technologies, portable compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical disc storage, magnetic cartridges, magnetic tape, disk storage or other magnetic storage devices, or any other medium that can be used to store desired information and is accessible to a computer. Furthermore, it is known to those skilled in the art that communication media typically contain computer-readable program instructions, data structures, program modules, or other data in modulated data signals such as carrier waves or other transmission mechanisms, and may include any information delivery medium.

[0173] The computer-readable program instructions described herein can be downloaded from computer-readable storage media to various computing / processing devices, or downloaded via a network, such as the Internet, local area network, wide area network, and / or wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmission, wireless transmission, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the computer-readable program instructions from the network and forwards them to the computer-readable storage media in the respective computing / processing device.

[0174] The computer program instructions used to perform the operations of this invention may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The computer-readable program instructions may be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the computer-readable program instructions. This electronic circuitry can execute the computer-readable program instructions to implement various aspects of the invention.

[0175] The computer program product described herein can be implemented specifically through hardware, software, or a combination thereof. In one alternative embodiment, the computer program product is specifically embodied in a computer storage medium; in another alternative embodiment, the computer program product is specifically embodied in a software product, such as a software development kit (SDK), etc.

[0176] Various aspects of the present invention are described herein with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It should be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer-readable program instructions.

[0177] These computer-readable program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing apparatus to produce a machine such that, when executed by the processor of the computer or other programmable data processing apparatus, they create means for implementing the functions / actions specified in one or more blocks of the flowchart and / or block diagram. These computer-readable program instructions can also be stored in a computer-readable storage medium that causes a computer, programmable data processing apparatus, and / or other device to operate in a particular manner; thus, the computer-readable medium storing the instructions comprises an article of manufacture that includes instructions for implementing aspects of the functions / actions specified in one or more blocks of the flowchart and / or block diagram.

[0178] Computer-readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable data processing apparatus, or other device to produce a computer-implemented process, thereby causing the instructions executed on the computer, other programmable data processing apparatus, or other device to perform the functions / actions specified in one or more boxes of a flowchart and / or block diagram.

[0179] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of an instruction, which contains one or more executable instructions for implementing a specified logical function. In some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0180] Example embodiments have been disclosed herein, and while specific terminology has been used, it is for illustrative purposes only and should be construed as such, and is not intended to be limiting. In some instances, it will be apparent to those skilled in the art that features, characteristics, and / or elements described in conjunction with particular embodiments may be used alone, or in combination with features, characteristics, and / or elements described in conjunction with other embodiments, unless otherwise expressly indicated. Therefore, those skilled in the art will understand that various changes in form and detail may be made without departing from the scope of the invention as set forth in the appended claims.

Claims

1. A method for screening the composition of alloy materials, characterized in that, include: Construct an alloy composition and performance dataset; A multi-model performance prediction model is constructed, and the multi-model performance prediction model is trained based on the alloy composition performance dataset to select a performance prediction benchmark model. A global search is performed based on the genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results. The alloy composition scheme and performance prediction results are weighted and ranked to select the target alloy combination; The target alloy combination was simulated and the process was analyzed to optimize the performance prediction benchmark model.

2. The method as described in claim 1, characterized in that, The construction of the alloy composition performance dataset includes: Collect data on the alloying element composition, mechanical properties, and heat treatment process of the target alloy; The composition of alloying elements, mechanical properties, and heat treatment process data were analyzed to obtain element quality and optimization targets. The quality of the elements and the optimization objectives are standardized. An alloy composition performance dataset was constructed based on the standardized element quality and optimization objectives.

3. The method as described in claim 1, characterized in that, The construction of the multi-model performance prediction model, which involves training the multi-model performance prediction model based on the alloy composition performance dataset to select a performance prediction benchmark model, includes: A multi-model performance prediction model is constructed based on several machine learning models; The multi-model performance prediction model is trained based on the alloy composition and performance dataset to obtain the trained model. Evaluation indicators are constructed based on mean absolute error, mean square error, and coefficient of determination. The trained models are selected based on the evaluation metrics to obtain the performance prediction benchmark model.

4. The method as described in claim 1, characterized in that, The multi-model performance prediction model includes a backpropagation neural network, a support vector machine model, and a random forest model; the performance prediction benchmark model includes a backpropagation neural network.

5. The method as described in claim 1, characterized in that, The step of performing a global search based on the genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results includes: Genetic algorithms are used to set chromosome encoding methods, determine component search boundaries, and configure genetic algorithm control parameters. Using the performance prediction benchmark model as an evaluation tool, the predicted performance of each alloy individual composition is calculated, and a multi-objective comprehensive score function is constructed as the fitness function of the genetic algorithm. Perform selection, crossover, and mutation genetic operations; Based on the genetic algorithm, an iterative search is performed to obtain the alloy composition scheme and corresponding performance prediction results for each individual.

6. The method as described in claim 1, characterized in that, The step of weighting and ranking the alloy composition scheme and performance prediction results to select the target alloy combination includes: The performance prediction results are normalized to obtain the normalized index score; Obtain the weights of each performance metric; The weighted score is calculated based on the normalized index score and weight; The alloy composition schemes are sorted from high to low based on the weighted scores to select target alloy combinations.

7. The method as described in claim 6, characterized in that, The calculation of the weighted score based on the normalized indicator score and weights includes: The density score, Young's modulus score, and tensile strength score are determined based on the normalized index scores. The density weight, Young's modulus weight, and tensile strength weight are determined based on the aforementioned weights. The weighted score is obtained by summing the products of the density score and density weight, the Young's modulus score and Young's modulus weight, and the tensile strength score and tensile strength weight.

8. A system for screening the composition of alloy materials, characterized in that, include: The data acquisition module is used to build alloy composition and performance datasets; The model building module is used to build a multi-model performance prediction model and train the multi-model performance prediction model based on the alloy composition performance dataset to select the performance prediction benchmark model. The search optimization module is used to perform a global search based on the genetic algorithm and the performance prediction benchmark model to obtain alloy composition schemes and corresponding performance prediction results. The weighted scoring module is used to perform weighted scoring and sorting based on the alloy composition scheme and performance prediction results in order to select the target alloy combination; The simulation verification module is used to perform simulation verification and process analysis on the target alloy combination in order to optimize the performance prediction benchmark model.

9. An electronic device, characterized in that, include: One or more processors; Memory, used to store one or more programs; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1 to 7.

10. A computer-readable medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method as described in any one of claims 1 to 7.

Images