Alloy formula determination method and related device, alloy formula and related product

Abstract

The invention provides an alloy formula determination method, a related device, an alloy formula and a related product, and relates to the technical field of materials, the method comprises the following steps: obtaining a to-be-verified formula of a target alloy; the to-be-verified formula is input into a pre-trained performance prediction model to obtain predicted performance data, and the performance prediction model is used for determining the predicted performance data according to the to-be-verified formula and the mapping relation between the formula parameters and the performance parameters; and performing iterative optimization on the formula to be verified according to the predicted performance data and a target performance requirement corresponding to the target alloy, and determining a target formula. Through the performance prediction model, the performance of the alloy prepared by the formula is predicted, and the formula is dynamically optimized based on the performance requirements, so that the final formula of the alloy can meet the performance requirements, the trial and error times can be reduced, the research and development success rate is improved, the research and development period is greatly shortened, and meanwhile, the method is suitable for formula research and development of various alloys; and a universal framework is provided for development of new materials.

Description

Technical Field

[0001] The present disclosure relates to the field of material technology, and in particular to a method for determining an alloy formula and related devices, an alloy formula, and related products. Background Art

[0002] Traditional alloy formula development mainly relies on trial and error and expert experience. Researchers need to repeatedly prepare samples and test performance, which has problems such as long cycle, high cost and low efficiency. Summary of the Invention

[0003] In order to overcome the problems existing in the related art, the present disclosure provides a method for determining an alloy formula and related devices, an alloy formula and related products.

[0004] According to a first aspect of an embodiment of the present disclosure, a method for determining an alloy formula is provided, comprising: Obtain the target alloy formula to be verified; Inputting the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data, wherein the performance prediction model is used to determine the predicted performance data based on the recipe to be verified and a mapping relationship between recipe parameters and performance parameters; According to the predicted performance data and the target performance requirements corresponding to the target alloy, the formulation to be verified is iteratively optimized to determine the target formulation.

[0005] In this way, through the performance prediction model, the performance of the alloy that can be made from the formula is predicted, and the formula is dynamically optimized based on the performance requirements, which can ensure that the final formula of the alloy meets the performance requirements, reduce the number of trial and error and improve the success rate of R&D, greatly shorten the R&D cycle, and is suitable for the formula development of various alloys, providing a general framework for the development of new materials.

[0006] In some possible implementations, the training of the performance prediction model includes: Obtaining a sample formula of the target alloy and actual performance data corresponding to the sample formula; Determining sample predicted performance data based on the sample formula; The model parameters of the performance prediction model are updated according to the sample predicted performance data and the actual performance data.

[0007] In this way, the performance prediction model is trained using the sample formula and actual performance data of the target alloy, enabling the model to learn and adapt to the characteristics of the specific alloy, enhancing the applicability and pertinence of the model to the alloy system, and enabling it to better fit the actual mapping relationship between the formula and performance, thereby improving the model's prediction accuracy for alloy performance and reducing the deviation between the predicted results and actual performance.

[0008] In some possible implementations, updating the model parameters of the performance prediction model according to the sample predicted performance data and the actual performance data includes: determining a prediction deviation based on the sample predicted performance data and the actual performance data; According to the prediction deviation, the model parameters of the performance prediction model are updated.

[0009] In this way, by adjusting the prediction deviation and updating the model parameters, the performance prediction model can gradually learn a more accurate mapping relationship between alloy formula and performance, thereby improving the model's prediction accuracy for alloy performance and reducing the error between the prediction results and actual performance.

[0010] In some possible implementations, iteratively optimizing the formulation to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine a target formulation includes: In response to the predicted performance data not meeting the target performance requirement, adjusting the recipe to be verified to obtain an updated recipe; The updated formula is used as a new formula to be verified, and the step of inputting the formula to be verified into a pre-trained performance prediction model to obtain predicted performance data is repeated until the predicted performance data of the formula to be verified meets the target performance requirements, and the target formula is determined.

[0011] In this way, through iterative optimization, adjustments can be made gradually according to the formula to be verified, and the alloy formula that meets the target performance requirements can be determined, avoiding the large number of trial and error experiments in traditional testing methods, reducing the number of experiments and resource consumption, shortening the R&D cycle, and reducing costs.

[0012] In some possible implementations, obtaining a target alloy formula to be verified includes: Determining a target alloy formulation to be verified from within a range of formulation parameters to be optimized; Iteratively optimize the formulation to be verified to determine the target formulation, including: Iteratively optimizing the formulation to be verified to obtain an optimized formulation parameter range that meets the target performance requirements; The target formulation is determined according to the optimized formulation parameter range.

[0013] By providing a range of formulation parameters to be optimized and narrowing the range, the number of iterative optimization and experimental verification steps can be further reduced, saving computing resources and improving alloy R&D efficiency. Furthermore, the optimized formulation parameter range obtained through iterative optimization can also provide a formulation parameter range that meets performance requirements, allowing R&D personnel to flexibly adjust the formulation based on actual needs, improving the formulation's adaptability and practicality.

[0014] In some possible implementations, determining the target formulation according to the optimized formulation parameter range includes: Determining a candidate formulation based on the optimized formulation parameter range; Obtaining measured performance data corresponding to the candidate formula; Based on the measured performance data and the target performance requirements, the candidate formulations are verified to obtain the target formulation.

[0015] In this way, by verifying the candidate formulas obtained after iterative optimization of the model through actual test data, it can be ensured that the target formula can truly meet the target performance requirements, avoiding the errors and uncertainties that may be caused by relying solely on the prediction model, and effectively improving the performance stability and reliability of the target formula in actual production applications.

[0016] In some possible implementations, verifying the candidate formulation based on the measured performance data and the target performance requirements to obtain the target formulation includes: In response to the measured performance data not meeting the target performance requirement, iteratively training the performance prediction model using the candidate recipe and the measured performance data to obtain an updated performance prediction model; The updated performance prediction model is used as the performance prediction model, and the step of inputting the recipe to be verified into the pre-trained performance prediction model to obtain predicted performance data is repeatedly performed to iteratively optimize the recipe to be verified to obtain a new candidate recipe until the test performance data corresponding to the candidate recipe meets the target performance requirements.

[0017] According to a second aspect of the embodiments of the present disclosure, an alloy formula is provided, which is obtained by the alloy formula determination method described in the first aspect of the embodiments of the present disclosure.

[0018] In some possible implementations, the alloy is high-strength steel, and the components and their mass percentages in the formula include: Carbon: 0.29~0.44%; Silicon: 0.10~1.0%; Manganese: 0.30~1.5%; Chromium: 0.1~1%; Boron: 0.0005~0.005%; Aluminum: 0.10~0.60%; Niobium: 0.02~0.06%; Vanadium: 0.05~0.20%; Cobalt: 0.001~0.08%; The balance is iron.

[0019] In some possible embodiments, the ingredients in the formula and their mass percentages include: Carbon: 0.33~0.40%; Silicon: 0.2~0.5%; Manganese: 0.5~0.8%; Chromium: 0.1~0.5%; Boron: 0.001~0.01%; Aluminum: 0.10~0.40%; Niobium: 0.02~0.05%; Vanadium: 0.05~0.20%; Cobalt: 0.004~0.06%; The balance is iron.

[0020] In some possible embodiments, the ingredients in the formula and their mass percentages include: Carbon: 0.36%; Silicon: 0.33%; Manganese: 0.71%; Chromium: 0.36%; Boron: 0.0025%; Aluminum: 0.33%; Niobium: 0.024%; Vanadium: 0.13%; Cobalt: 0.005%; The balance is iron.

[0021] In some possible embodiments, the process parameters in the formula include an autotempering temperature in the heat treatment parameters, and the autotempering temperature is determined based on multiple components in the formula and their mass percentages, and the relationship between the multiple components and the autotempering temperature; wherein the autotempering temperature is greater than or equal to the set temperature.

[0022] In some possible implementations, the auto-tempering temperature is greater than or equal to 340 degrees Celsius.

[0023] According to a third aspect of an embodiment of the present disclosure, there is provided a device for determining an alloy formula, comprising: an acquisition module, configured to acquire a target alloy formula to be verified; an acquisition module configured to input the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data, wherein the performance prediction model is used to determine the predicted performance data based on the recipe to be verified and a mapping relationship between recipe parameters and performance parameters; The determination module is configured to iteratively optimize the formula to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine the target formula.

[0024] In some possible implementations, the training of the performance prediction model includes: Obtaining a sample formula of the target alloy and actual performance data corresponding to the sample formula; Determining sample predicted performance data based on the sample formula; The model parameters of the performance prediction model are updated according to the sample predicted performance data and the actual performance data.

[0025] In some possible implementations, the determining module is further configured to: In response to the predicted performance data not meeting the target performance requirement, adjusting the recipe to be verified to obtain updated recipe data; The updated recipe data is used as a new recipe to be verified, and the step of inputting the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data is repeated until the predicted performance data meets the target performance requirements, thereby determining the target recipe.

[0026] According to a fourth aspect of the embodiments of the present disclosure, an automobile component is provided, wherein the automobile component includes an alloy made according to the alloy formula described in the second aspect of the embodiments of the present disclosure.

[0027] According to a fifth aspect of the embodiment of the present disclosure, a car is provided, comprising the car component described in the fourth aspect of the embodiment of the present disclosure.

[0028] According to a sixth aspect of an embodiment of the present disclosure, there is provided an electronic device, including: processor; a memory for storing processor-executable instructions; The processor is configured to execute the method for determining the alloy formula described in the first aspect of the embodiment of the present disclosure.

[0029] According to a seventh aspect of the embodiments of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the method for determining the alloy formula described in the first aspect of the embodiments of the present disclosure is implemented.

[0030] According to an eighth aspect of the embodiments of the present disclosure, a computer program product is provided, comprising a computer program, which, when executed by a processor, implements the method for determining the alloy formula according to the first aspect of the embodiments of the present disclosure.

[0031] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects: The present disclosure obtains a target alloy formula to be verified; inputs the formula to be verified into a pre-trained performance prediction model to obtain predicted performance data. The performance prediction model is used to determine the predicted performance data based on the formula to be verified and the mapping relationship between the formula parameters and the performance parameters; and iteratively optimizes the formula to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine the target formula. In this way, the performance of the alloy that can be made from the formula is predicted through the performance prediction model, and the formula is dynamically optimized based on the performance requirements. This ensures that the final formula of the alloy meets the performance requirements, can reduce the number of trial and error times, and improve the success rate of research and development, greatly shortening the research and development cycle. At the same time, it is applicable to the formulation research and development of various alloys, providing a general framework for the development of new materials.

[0032] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0034] Figure 1 The figure is a flow chart showing a method for determining an alloy formula according to an exemplary embodiment.

[0035] Figure 2 FIG. 1 is a parallel coordinate diagram corresponding to a recipe of high-strength steel according to an exemplary embodiment.

[0036] Figure 3 is a stress-strain curve diagram of high-strength steel according to an exemplary embodiment.

[0037] Figure 4 is a schematic diagram showing the metallographic structure of a high-strength steel according to an exemplary embodiment.

[0038] Figure 5 The figure is a block diagram of a device for determining an alloy formula according to an exemplary embodiment.

[0039] Figure 6 It is a block diagram of an electronic device according to an exemplary embodiment. DETAILED DESCRIPTION

[0040] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.

[0041] As mentioned in the background, traditional alloy formulation development relies primarily on trial and error and expert experience. Researchers must repeatedly prepare samples and test performance, resulting in long cycles, high costs, and low efficiency. Computer-aided design or machine learning can be partially applied to materials research and development, such as thermodynamic calculations. However, these applications are typically used to predict performance results or trends, and the performance prediction accuracy for complex multi-component alloys is limited. Furthermore, it is difficult to dynamically optimize formulation parameters, such as annealing temperature, heating zone dew point, and other core process parameters.

[0042] Take the automotive industry, for example. As vehicle crash safety requirements increase, so too does the demand for crashworthiness. Automotive bodies utilize hot-formed steel, but traditional hot-formed steel typically has a strength of 1500 MPa. This falls short of the desired vehicle crash safety requirements, as its deformation resistance and energy absorption capacity are insufficient. Consequently, automakers and steel companies are developing and mass-producing materials with strengths of 2200 MPa and even higher.

[0043] While improving material strength, it's also crucial to maintain formability and toughness. Additional alloying and heat treatment process optimization can further improve the steel's yield strength ratio, elongation, and fatigue resistance. Currently, the industry typically uses traditional alloying combined with heat treatment to increase steel's ultimate strength. For example, by adding various elements to existing 1500MPa steel and optimizing the quenching / tempering process, material properties can be increased to 1700-1800MPa.

[0044] However, for high-strength steels above 2200 MPa, the currently available formulas for their composition and content are incomplete. Traditional R&D methods rely heavily on experience and extensive trial and error, resulting in lengthy development cycles and an inability to quickly identify optimal composition and heat treatment parameters. This makes it difficult to achieve the optimal balance between strength and toughness.

[0045] Reference Figure 1 , Figure 1 is a flow chart showing a method for determining an alloy formula according to an exemplary embodiment. Figure 1 As shown, the method for determining the alloy formula includes the following steps.

[0046] In step S101, the target alloy formula to be verified is obtained.

[0047] In step S102, the recipe to be verified is input into a pre-trained performance prediction model to obtain predicted performance data. The performance prediction model is used to determine the predicted performance data based on the recipe to be verified and the mapping relationship between recipe parameters and performance parameters.

[0048] In step S103, the recipe to be verified is iteratively optimized based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine the target recipe.

[0049] Exemplarily, the target alloy is an alloy to be developed. Among them, the target alloy can be classified according to the main elements, for example, it can be an iron alloy, an aluminum alloy, a copper alloy, a titanium alloy, a zinc alloy, a lead-tin alloy, and a high-temperature alloy. The target alloy can also be classified according to different metallographic structures, wherein the metallographic structure is used to reflect the specific morphology of the metal phase, characterize the chemical composition of the metal or alloy and the physical and chemical state of various components in the alloy. For example, the morphology of the target alloy is martensite, austenite, ferrite, pearlite, etc. This is not limited here.

[0050] For example, a recipe refers to the preparation method and proportion relationship formed by combining specific substances. Here, the recipe includes the component contents and process parameters. The process parameters can be determined based on the specific preparation process of the target alloy. For example, the process parameters may include smelting process parameters, heat treatment parameters, casting process parameters, cold working parameters, etc. Among them, heat treatment parameters also include annealing temperature, heat treatment temperature range, holding time, etc.

[0051] Exemplarily, the formulation to be verified is a set of initially set alloy element compositions and ratios, as well as process parameters. The formulation to be verified can be determined from the approximate range of alloy formulations and process parameters already available in the relevant art, or by combining relevant public literature, patents, and experimental data.

[0052] For example, the performance data of alloy materials mainly include hardness, strength, plasticity, toughness, corrosion resistance, heat resistance, formability, etc. Performance data is crucial for evaluating the application range and service life of alloy materials.

[0053] For example, target performance requirements are performance constraints that the target alloy must meet. Different materials have different target performance requirements. For example, for high-strength steel, the strength requirement may be a tensile strength greater than 2200 MPa, and the formability requirement may be an elongation at break greater than or equal to 19%.

[0054] For example, a performance prediction model is a mathematical model or neural network model established through machine learning that can predict the performance data corresponding to the alloy based on the input alloy formula parameters, such as random forest, neural network, XGBoost algorithm, Catboost algorithm, etc.

[0055] For example, a multidimensional database can be established based on historical experimental data, public literature, patent literature, etc., and a performance prediction model can be obtained by training through the multidimensional database, wherein the historical experimental data, public literature, and patent literature include known alloy formulas and their corresponding actual performance data. During the training process, the model will learn the correlation and rules between the alloy formula and the corresponding performance parameters, so that it can be applied to the performance prediction of the new formula to be verified.

[0056] For example, based on the predicted performance data and the target performance requirements corresponding to the target alloy, an optimization algorithm can be set to iteratively optimize the formulation to be verified, and the formulation parameters can be automatically adjusted. For example, an optimization algorithm such as a genetic algorithm, Bayesian optimization, gradient descent, etc. can be used to search for an optimal solution that meets the target performance requirements within a set parameter space.

[0057] For example, a basic parameter space can be provided based on experience, literature research, or preliminary experimental data. For example, the initial range of the ingredient content in the formula, as well as the initial range of process parameters such as heating temperature and holding time, can be used to determine the initial verification formula of the target alloy. This initial verification formula is input into a pre-trained performance prediction model. The model will output the predicted performance data corresponding to the formula based on the previously learned mapping relationship between formula and performance. These predicted performance data can initially reflect the mechanical properties characteristics that the alloy with this formula may possess.

[0058] For example, by comparing the predicted performance data with the target performance requirements of the target alloy, gaps and deficiencies in the initial formulation to be verified can be identified. Based on this comparison, the current formulation to be verified can be adjusted and optimized according to specific optimization rules, such as adjusting the element content ratio to increase strength or changing the ratio of certain elements to improve toughness, to generate a new formulation to be verified. The new formulation to be verified can be re-entered into the performance prediction model and the above process of performance prediction, comparative analysis, and optimization adjustment can be repeated. Alternatively, the actual performance of the new formulation to be verified can be tested in small-scale trials to verify the effectiveness of the optimization process. After multiple iterations, a final target formulation that meets the performance requirements of the target alloy can be determined.

[0059] The present disclosure obtains a target alloy formula to be verified; inputs the formula to be verified into a pre-trained performance prediction model to obtain predicted performance data. The performance prediction model is used to determine the predicted performance data based on the formula to be verified and the mapping relationship between the formula parameters and the performance parameters; and iteratively optimizes the formula to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine the target formula. In this way, the performance of the alloy that can be made from the formula is predicted through the performance prediction model, and the formula is dynamically optimized based on the performance requirements. This ensures that the final formula of the alloy meets the performance requirements, can reduce the number of trial and error times, and improve the success rate of research and development, greatly shortening the research and development cycle. At the same time, it is applicable to the formulation research and development of various alloys, providing a general framework for the development of new materials.

[0060] In some possible implementations, the training of the performance prediction model includes: Obtaining a sample formula of the target alloy and actual performance data corresponding to the sample formula; Determining sample predicted performance data based on the sample formula; The model parameters of the performance prediction model are updated according to the sample predicted performance data and the actual performance data.

[0061] For example, the sample formula and the actual performance data corresponding to the sample formula are used to train the performance prediction model and can be used to learn the relationship between formula parameters and performance parameters. The sample formula includes different elemental components and their contents, and the actual performance data is the alloy performance data actually measured for the sample formula.

[0062] Exemplarily, different sample formulations of the target alloy may have different formulation parameter values, such as different component contents of the target alloy and / or different process parameters of the target alloy. In one example, different sample formulations of the target alloy may have the same elemental composition and process flow, but different values ​​corresponding to the component contents and process parameters. This allows the model to more accurately learn the mapping relationship between the formulation parameters and performance parameters corresponding to the target alloy.

[0063] For example, a sample formula of the target alloy and its corresponding actual performance data are obtained as the dataset required for training a performance prediction model, where the sample formula serves as input data and the actual performance data serves as output labels. The sample formula is input into the performance prediction model, which then predicts the sample performance data based on the current formula parameters and the learned mapping relationship. Comparing the sample predicted performance data with the actual performance data can be used to evaluate the accuracy of the model's predictions and to update the model parameters.

[0064] In this way, the performance prediction model is trained using the sample formula and actual performance data of the target alloy, enabling the model to learn and adapt to the characteristics of the specific alloy, enhancing the applicability and pertinence of the model to the alloy system, and enabling it to better fit the actual mapping relationship between the formula and performance, thereby improving the model's prediction accuracy for alloy performance and reducing the deviation between the predicted results and actual performance.

[0065] In some possible implementations, updating the model parameters of the performance prediction model according to the sample predicted performance data and the actual performance data includes: determining a prediction deviation based on the sample predicted performance data and the actual performance data; According to the prediction deviation, the model parameters of the performance prediction model are updated.

[0066] For example, prediction deviation refers to the difference between sample predicted performance data and the corresponding actual performance data. Prediction deviation is used to characterize the degree of deviation between the model's prediction and the true value. Prediction deviation can be expressed in various forms, such as absolute error, relative error, and mean square error. Here, the prediction deviation between sample predicted performance data and actual performance data can be determined using a mean square error loss function.

[0067] For example, backpropagation can be used to determine the gradient of the prediction bias with respect to the weights of each layer in the network, such as the gradient of parameters such as the weights and biases of a neural network. The gradient represents the degree and direction of the impact of parameter changes on the loss function. Gradient descent can be used to update the network's weight parameters to minimize the prediction bias.

[0068] For example, a sample recipe is input into the performance prediction model to obtain sample predicted performance data. The predicted performance data is then compared with the corresponding actual performance data. The error between the two can be calculated using a mean square error loss function to obtain the prediction deviation. Based on the prediction deviation, the parameters of the performance prediction model can be updated and adjusted using a specific optimization algorithm, such as gradient descent, momentum gradient descent, Adam optimization, or other optimization algorithms, thereby gradually improving the consistency between its predicted performance and actual performance.

[0069] For example, the model can be trained and optimized by repeating the above process of determining the prediction bias, calculating the gradient through backpropagation, and updating the model parameters. As the number of iterations increases, the model's prediction bias gradually decreases, and the performance prediction capability continues to improve until a predetermined training stop condition is met, such as when the training reaches the maximum number of iterations or the prediction bias falls below a set threshold, ultimately resulting in a trained performance prediction model.

[0070] In this way, by adjusting the prediction deviation and updating the model parameters, the performance prediction model can gradually learn a more accurate mapping relationship between alloy formula and performance, thereby improving the model's prediction accuracy for alloy performance and reducing the error between the prediction results and actual performance.

[0071] In some possible implementations, iteratively optimizing the formulation to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine a target formulation includes: In response to the predicted performance data not meeting the target performance requirement, adjusting the recipe to be verified to obtain an updated recipe; The updated formula is used as a new formula to be verified, and the step of inputting the formula to be verified into a pre-trained performance prediction model to obtain predicted performance data is repeated until the predicted performance data of the formula to be verified meets the target performance requirements, and the target formula is determined.

[0072] For example, the unverified formulation is an alloy formulation initially determined based on prior knowledge or experience, serving as the starting point for optimization. The updated formulation is a new formulation resulting from adjustments to the initial unverified formulation or the previously-previously-verified formulation during the iterative optimization process, based on the discrepancy between predicted performance data and target performance requirements.

[0073] For example, when the predicted performance data does not meet the target performance requirement, the to-be-verified formulation can be gradually improved by iteratively optimizing the formulation until the target performance requirement is met. Each optimization step adjusts the formulation parameters based on the comparison between the predicted performance data and the target performance requirement to gradually approach the target performance.

[0074] For example, the recipe to be verified is input into a performance prediction model to obtain predicted performance data, which is then compared with the target performance requirement. If the predicted performance data does not meet the target performance requirement, the difference between the predicted performance data and the target performance requirement can be used to determine the direction of recipe adjustment, and the current recipe can be adjusted to obtain an updated recipe. The updated recipe is then used as the new recipe to be verified and re-input into the performance prediction model, repeating the performance prediction and comparison process. Each iteration adjusts the recipe based on the predicted results, gradually approaching the target performance requirement. When the predicted performance data obtained in a certain iteration meets the target performance requirement, the iteration can be stopped, and the current recipe is determined to be the target recipe.

[0075] In this way, through iterative optimization, adjustments can be made gradually according to the formula to be verified, and the alloy formula that meets the target performance requirements can be determined, avoiding the large number of trial and error experiments in traditional testing methods, reducing the number of experiments and resource consumption, shortening the R&D cycle, and reducing costs.

[0076] In some possible implementations, the recipe to be verified is iteratively optimized based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine a target recipe, including: in response to the predicted performance data satisfying the target performance requirements, based on the recipe parameter range to be optimized, adjusting the recipe to be verified to obtain an optimized updated recipe; using the optimized updated recipe as a new recipe to be verified, and repeatedly executing the step of inputting the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data, until the predicted performance data corresponding to all recipes within the recipe parameter range satisfy the target performance requirements, and an optimized recipe parameter range is obtained; and determining the target recipe based on the optimized recipe parameter range.

[0077] In some possible implementations, obtaining a target alloy formula to be verified includes: Determining a target alloy formulation to be verified from within a range of formulation parameters to be optimized; Iteratively optimize the formulation to be verified to determine the target formulation, including: Iteratively optimizing the formulation to be verified to obtain an optimized formulation parameter range that meets the target performance requirements; The target formulation is determined according to the optimized formulation parameter range.

[0078] Exemplarily, the range of formula parameters to be optimized is the adjustable range of formula parameters such as the content of each component and process parameters in the formula of the target alloy, which can be determined based on the alloy performance requirements and component characteristics, or based on historical experimental data, public literature, patent literature and other relevant data.

[0079] Exemplarily, the range of recipe parameters to be optimized may meet the target performance requirements, and may be used to query the preferred range and / or optimal parameter combination of recipe parameters that further meet the target performance requirements; the range of recipe parameters to be optimized may also be an approximate range that partially does not meet the target performance requirements, and may be used to optimize and determine the preferred range and / or optimal parameter combination of recipe parameters that meet the target performance requirements.

[0080] Illustratively, the optimized formulation parameter range is the formulation parameter range that meets the target performance requirements after iterative optimization, or a preferred range of formulation parameters that further meets the target performance requirements. Compared to the initial range, it is more accurate and more consistent with the target alloy performance requirements.

[0081] For example, an initial formulation to be verified can be selected from a range of formulation parameters to be optimized, combining experience, theoretical guidance, or random methods. This initial formulation to be verified is input into a performance prediction model to determine predicted performance data for the initial formulation to be verified. Based on the discrepancy between the predicted performance data and the target performance requirements, the formulation parameters are adjusted to form an updated formulation. Repeating this process can gradually narrow the range of formulation parameters to be optimized, improving the match between the predicted performance data and the target performance requirements. After multiple rounds of iterative optimization, an optimized formulation parameter range that meets the target performance requirements is obtained. All formulation parameter combinations within this range can ensure that the alloy properties meet the target requirements.

[0082] For example, after determining the optimized formulation parameter range that meets the target performance requirements, R&D personnel can select the final target formulation within the optimized formulation parameter range. When selecting the target formulation, factors such as performance, cost, and process feasibility can be comprehensively considered.

[0083] By providing a range of formulation parameters to be optimized and narrowing the range, the number of iterative optimization and experimental verification steps can be further reduced, saving computing resources and improving alloy R&D efficiency. Furthermore, the optimized formulation parameter range obtained through iterative optimization can also provide a formulation parameter range that meets performance requirements, allowing R&D personnel to flexibly adjust the formulation based on actual needs, improving the formulation's adaptability and practicality.

[0084] In some possible implementations, determining the target formulation according to the optimized formulation parameter range includes: Determining a candidate formulation based on the optimized formulation parameter range; Obtaining measured performance data corresponding to the candidate formula; Based on the measured performance data and the target performance requirements, the candidate formulations are verified to obtain the target formulation.

[0085] For example, the candidate recipe is a specific recipe instance selected from the optimized recipe parameter range. The measured performance data is obtained by performing actual tests on the candidate recipe to obtain performance data corresponding to the candidate recipe.

[0086] For example, several candidate formulations can be selected from the optimized formulation parameter ranges obtained by the model. The candidate formulations are then tested to obtain corresponding measured performance data. By verifying the measured performance data of the candidate formulations against the target performance requirements, it can be determined whether the candidate formulations meet the target performance requirements.

[0087] For example, when the measured performance data of the candidate recipes all meet the target performance requirements, the target recipe can be determined from the candidate recipes, or the target recipe can be re-determined from the optimized recipe parameter range.

[0088] Here, there may be multiple candidate recipes, and the verification results of the candidate recipes may include whether the candidate recipes meet or do not meet the target performance requirements. When multiple candidate recipes all meet the target performance requirements, it can be considered that the optimized recipe parameter range meets the target performance requirements. When at least one of the multiple candidate recipes does not meet the target performance requirements, it can be considered that the optimized recipe parameter range does not meet the target performance requirements.

[0089] In this way, by verifying the candidate formulas obtained after iterative optimization of the model through actual test data, it can be ensured that the target formula can truly meet the target performance requirements, avoiding the errors and uncertainties that may be caused by relying solely on the prediction model, and effectively improving the performance stability and reliability of the target formula in actual production applications.

[0090] In some possible implementations, verifying the candidate formulation based on the measured performance data and the target performance requirements to obtain the target formulation includes: In response to the measured performance data not meeting the target performance requirement, iteratively training the performance prediction model using the candidate recipe and the measured performance data to obtain an updated performance prediction model; The updated performance prediction model is used as the performance prediction model, and the step of inputting the recipe to be verified into the pre-trained performance prediction model to obtain predicted performance data is repeatedly performed to iteratively optimize the recipe to be verified to obtain a new candidate recipe until the test performance data corresponding to the candidate recipe meets the target performance requirements.

[0091] Exemplarily, the updated performance prediction model is obtained by repeatedly training the existing performance prediction model with candidate recipes and their measured performance data. The updated performance prediction model incorporates new data and information on the basis of the original model, and can perform performance prediction more accurately.

[0092] For example, when the measured performance data of the candidate formula does not meet the target performance requirements, it can be considered that the results of the iterative optimization do not fully meet the target performance requirements, or there are errors in the prediction results of the performance prediction model. The candidate formula and the measured performance data corresponding to the candidate formula can be used as training data to iteratively train the performance prediction model again.

[0093] For example, the updated performance prediction model after retraining can be reused for performance prediction of the recipe to be verified, and the process of performance prediction and iterative optimization of the recipe to be verified can be repeated to obtain a new optimized recipe parameter range, and the candidate recipes within the new optimized recipe parameter range can be verified until the measured performance of the new candidate recipe meets the target performance requirements.

[0094] In this way, when the measured data for a candidate formulation doesn't meet the target performance requirements, the performance prediction model can be trained iteratively to improve its accuracy. This organically integrates actual testing with model predictions, creating a closed-loop optimization loop. This significantly improves R&D efficiency and product stability, while shortening R&D cycles and reducing costs.

[0095] This paper uses the research and development of high-strength steel as an example. By establishing an automated data closed loop between the laboratory and the server, it achieves efficient iteration and forward optimization based on machine learning. The server is a key device for machine learning model training, data optimization, and data management.

[0096] In the laboratory, researchers conducted small-scale trial production and physical experiments on different batches of alloy formulas and heat treatment schemes, including melting or rolling of alloy formulas, annealing temperature experiments, dew point experiments in the heating section, etc.; through tensile tests, impact tests, metallographic structure analysis, etc., the mechanical properties data of the materials, such as tensile strength, yield strength, elongation, impact toughness, etc., were obtained, and the microstructure of the trial materials was observed; the formula information, process parameters, test results and other data generated during the experiment were obtained and uploaded to the server in a standardized format.

[0097] Once the lab completes a round of small-batch testing and uploads the data, the server automatically reads the latest experimental data and analyzes it alongside historical experimental data. The performance prediction model on the server then learns the mapping between the alloy's formula parameters, including composition and process parameters, and its performance parameters. The server's model then automatically iterates, optimizes, and outputs new formula parameters. For example, it can fine-tune the Al, Si, and Mg contents, or modify certain heat treatment temperature ranges and holding times. The server utilizes a variety of machine learning and deep learning methods, such as random forests, neural networks, and the XGBoost algorithm, and sends the model-optimized formula back to the lab.

[0098] Once the lab receives the optimized formula from the server, it can conduct a round of validation testing using a small batch of production and performance tests. If the experimental results meet or approach the target performance requirements, including strength, toughness, and formability, the optimization process is considered successful. If there is a significant gap between the experimental results and the target performance requirements, the new experimental results can be uploaded to the server again for a new round of model training and optimization.

[0099] In this way, through multiple rounds of closed-loop research and development including laboratory trial production, data uploading, automatic model optimization, return of new formula, and laboratory verification, not only can the R&D cycle be significantly shortened, but the cost of blind trial and error can also be effectively reduced, and the ultra-high-strength steel formula that meets the needs of automotive anti-collision components can be accurately and quickly located.

[0100] In machine learning, when the test sample input falls within the training set input range, the model prediction process is called interpolation, while when it falls outside this range, it is called extrapolation. By determining interpolation and extrapolation test sets for the optional content range, performance data such as tensile strength, yield strength, and elongation at break were obtained, as shown in Table 1 below. Based on this performance data, the corresponding mean absolute error (MAE) and mean absolute percentage error (MAPE) can be determined. As can be seen, within the optional parameter range, the model's predictive performance significantly improves.

[0101] Table 1 Corresponding results of internalization test and externalization test

[0102] The present disclosure can also be applied and expanded to vehicle manufacturing. After the material composition and process are fully verified in the laboratory, expansion or mass production can be carried out on the steel plant production line. For example, during actual mass production, if there is actual need for temperature, dew point control or composition fine-tuning, the actual data can be transmitted to the server so that the server's model output can be further optimized and updated to ensure the stability of mass production batches. The overall process can also be applied to the development of other alloy systems or steels of different strength grades. By replacing or expanding the data set, a wider range of material research and development and optimization can be achieved.

[0103] As described above, the automated closed-loop system between the laboratory and the server, as proposed in this disclosure, enables efficient iteration from small-batch verification to mass production of ultra-high-strength steel. By using machine learning models to proactively predict and continuously optimize formulation ingredients and heat treatment processes, R&D efficiency is significantly improved while ensuring product stability, providing strong support for meeting increasingly stringent vehicle crash safety requirements.

[0104] The present disclosure also provides an alloy formula, which is obtained by the alloy formula determination method.

[0105] Here, alloy refers to a metallic substance that can be used to make parts, components, machines or other products.

[0106] The alloy formula determination method disclosed in the present invention can obtain an alloy formula that meets the target performance requirements, reduce the number of trial and error times, improve the success rate of research and development, and greatly shorten the research and development cycle.

[0107] In some possible embodiments, the alloy is high-strength steel, and the components in the formula and their mass percentages include: carbon: 0.29~0.44%; silicon: 0.10~1.0%; manganese: 0.30~1.5%; chromium: 0.1~1%; boron: 0.0005~0.005%; aluminum: 0.10~0.60%; niobium: 0.02~0.06%; vanadium: 0.05~0.20%; cobalt: 0.001~0.08%; and the balance is iron.

[0108] Here, carbon (C) is the main element, which can be used to ensure that the steel has high strength and hardness after heat treatment.

[0109] Here, silicon (Si) element can be used for deoxidation and a certain solid solution strengthening effect, which helps to improve the strength and has a certain effect on the toughness of steel.

[0110] Here, manganese (Mn) element has a deoxidation and desulfurization effect, provides significant solid solution strengthening, and can work together with other elements to greatly improve the hardenability of steel and improve the hot working performance of metal materials.

[0111] Here, chromium (Cr) is an auxiliary element that can be used to help improve hardenability and provide a certain wear resistance.

[0112] Here, the boron (B) element can greatly reduce the cost and significantly improve the hardenability of steel.

[0113] Here, the aluminum (Al) element can be used to ensure that a fine and uniform structure is obtained after heat treatment, while improving the strength and toughness of the steel. It also indirectly protects the role of boron and prevents nitrogen (N) and boron (B) from forming boron nitride during heat treatment and losing its function.

[0114] Here, niobium (Nb) elements can be used to refine grains during hot working and provide additional precipitation strengthening during subsequent heat treatment or use, further improving strength and toughness.

[0115] Here, vanadium (V) can be used to provide strong precipitation strengthening and is one of the main elements for improving the strength, hardness and wear resistance of steel. It can work synergistically with niobium to further enhance the effects of grain refinement and precipitation strengthening.

[0116] Here, the cobalt (Co) element mainly enhances the high-temperature performance and tempering stability of steel through solid solution strengthening, high-temperature structural stability and delayed tempering softening, and can also improve strength and some toughness.

[0117] Those skilled in the art will appreciate that in the preparation process of high-strength steel, unavoidable impurities may also exist, that is, in addition to the alloying elements provided in the above mass percentages, the remainder is actually iron (Fe) and unavoidable impurities.

[0118] Preferably, the ingredients in the formula and their mass percentages include: carbon: 0.33~0.40%; silicon: 0.2~0.5%; manganese: 0.5~0.8%; chromium: 0.1~0.5%; boron: 0.001~0.01%; aluminum: 0.10~0.40%; niobium: 0.02~0.05%; vanadium: 0.05~0.20%; cobalt: 0.004~0.06%; and the balance is iron.

[0119] Preferably, the ingredients in the formula and their mass percentages include: carbon: 0.36%; silicon: 0.33%; manganese: 0.71%; chromium: 0.36%; boron: 0.0025%; aluminum: 0.33%; niobium: 0.024%; vanadium: 0.13%; cobalt: 0.005%; and the balance is iron.

[0120] Among them, Figure 3 As shown in FIG, a schematic diagram of the strain-stress curve in a preferred embodiment is shown. Figure 4 , which is a schematic diagram of the metallographic structure corresponding to a preferred embodiment.

[0121] It will be appreciated that quenching and tempering methods are well known to those skilled in the art and are not described in detail herein. A preferred embodiment of the tempering treatment in this disclosure is as follows: the process parameters in the formulation include an autotempering temperature among the heat treatment parameters, the autotempering temperature being determined based on the multiple ingredients in the formulation and their mass percentages, as well as the relationship between the multiple ingredients and the autotempering temperature; wherein the autotempering temperature is greater than or equal to a set temperature. Preferably, the autotempering temperature is greater than or equal to 340 degrees Celsius.

[0122] For example, to address the reduced toughness associated with high carbon content, the model discovered during iterative formulation optimization that higher self-tempering temperatures (M) improve material toughness. In particular, Ms ≥ 340°C exhibits excellent tempering effects, eliminating or reducing internal stresses in the material and ensuring sufficient toughness in ultra-high-strength materials. The self-tempering temperature can be determined based on the composition and mass percentages of the various ingredients in the formulation, as well as the relationship between these ingredients and the self-tempering temperature. For example, the self-tempering temperature (M) can be calculated using the following formula: Ms = 500 - 320 [C] - 50 [Mn] - 30 [Cr] - 5 [Si] + 10 Co + 20 Al.

[0123] The present disclosure will be further described below through specific examples, but the present disclosure is not limited thereby.

[0124] The alloy formula determination method disclosed herein provides an optional content range and a preferred content range for a formula with a UTS>2200 MPa, as shown in Table 2 below. Experimental verification shows that the test results are consistent with expectations.

[0125] Table 2 Composition ranges of alloying elements in high-strength steel

[0126] In Examples 1-3, all component contents and autotempering temperatures are within selectable ranges. In Example 1, all component contents and the autotempering temperature are within the preferred range; in Example 2, all component contents are within the preferred range, but the autotempering temperature is outside the preferred range; and in Example 3, all component contents and the autotempering temperature are outside the preferred range. Comparative Examples 1-3 represent high-strength steel formulations provided by other companies, excluding niobium (Nb) and cobalt (Co). This formulation has been shown to further enhance strength and toughness.

[0127] like Figure 2 The figure shows a parallel coordinate plot corresponding to the optional content range of a high-strength steel formulation, where UTS represents tensile strength, YS represents yield strength, and EL represents elongation. Lines of different colors represent different tensile strengths, with green representing lower tensile strength and red representing higher tensile strength.

[0128] As shown in Table 3 below, for Examples 1-3 and Comparative Examples 1-3 of the present disclosure, the obtained untempered hot stamping flat die parts were tested for tensile strength, elongation, and maximum bend angle according to the GBT 228.1 room temperature tensile test standard and the VDA-238 three-point bend test standard. The VDA angle was calculated according to the calculation method in Appendix D of VDA 238-100: 2020. The tensile strength, yield strength, and elongation of the metal materials were determined using the room temperature tensile test method for metal materials. To reduce measurement errors, the final test results are the average of the three test results.

[0129] Table 3 Test results corresponding to the examples and comparative examples

[0130] It can be seen that through the alloy formula determination method disclosed in the present invention, the tensile strength of the obtained alloy formula can basically reach the 2200 MPa level.

[0131] Based on the same inventive concept, the present disclosure further provides an automobile component, which includes an alloy made by the alloy formula of the present disclosure.

[0132] Based on the same inventive concept, the present disclosure also provides an automobile, which includes the automobile component described in the present disclosure, or an alloy part made by the alloy formula determination method provided by the present disclosure.

[0133] Here, the automobile components described in the present disclosure may be one or more of an engine system, a transmission system, a braking system, a steering system, a driving system, an electrical instrument system, a vehicle body, and accessories.

[0134] Figure 5FIG. 5 is a block diagram of an alloy formula determination device 500 according to an exemplary embodiment. Figure 5 The alloy formula determination device 500 includes an acquisition module 501, an acquisition module 502 and a determination module 503.

[0135] An acquisition module 501 is configured to acquire a target alloy formula to be verified; An acquisition module 502 is configured to input the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data, wherein the performance prediction model is used to determine the predicted performance data based on the recipe to be verified and a mapping relationship between recipe parameters and performance parameters; The determination module 503 is configured to iteratively optimize the recipe to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine the target recipe.

[0136] In some possible implementations, the training of the performance prediction model includes: Obtaining a sample formula of the target alloy and actual performance data corresponding to the sample formula; Determining sample predicted performance data based on the sample formula; The model parameters of the performance prediction model are updated according to the sample predicted performance data and the actual performance data.

[0137] In some possible implementations, updating the model parameters of the performance prediction model according to the sample predicted performance data and the actual performance data includes: determining a prediction deviation based on the sample predicted performance data and the actual performance data; According to the prediction deviation, the model parameters of the performance prediction model are updated.

[0138] In some possible implementations, the determining module 503 is further configured to: In response to the predicted performance data not meeting the target performance requirement, adjusting the recipe to be verified to obtain updated recipe data; The updated recipe data is used as a new recipe to be verified, and the step of inputting the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data is repeated until the predicted performance data meets the target performance requirements, thereby determining the target recipe.

[0139] In some possible implementations, the acquisition module 501 is further configured to: Determining a target alloy formulation to be verified from within a range of formulation parameters to be optimized; The determining module 503 is further configured to: Iteratively optimizing the formulation to be verified to obtain an optimized formulation parameter range that meets the target performance requirements; The target formulation is determined according to the optimized formulation parameter range.

[0140] In some possible implementations, the determining module 503 is further configured to: Determining a candidate formulation based on the optimized formulation parameter range; Obtaining measured performance data corresponding to the candidate formula; Based on the measured performance data and the target performance requirements, the candidate formulations are verified to obtain the target formulation.

[0141] In some possible implementations, the determining module 503 is further configured to: In response to the measured performance data not meeting the target performance requirement, iteratively training the performance prediction model using the candidate recipe and the measured performance data to obtain an updated performance prediction model; The updated performance prediction model is used as the performance prediction model, and the step of inputting the recipe to be verified into the pre-trained performance prediction model to obtain predicted performance data is repeatedly performed to iteratively optimize the recipe to be verified to obtain a new candidate recipe until the test performance data corresponding to the candidate recipe meets the target performance requirements.

[0142] Regarding the device 500 for determining the alloy formula in the above embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method for determining the alloy formula, and will not be elaborated on here.

[0143] Based on the same inventive concept, the present disclosure further provides an electronic device, comprising: processor; a memory for storing processor-executable instructions; The processor is configured to execute the alloy formula determination method disclosed in the present invention.

[0144] Based on the same inventive concept, the present disclosure further provides a computer-readable storage medium having a computer program stored thereon, which implements the method for determining the alloy formula of the present disclosure when executed by a processor.

[0145] Based on the same inventive concept, the present disclosure further provides a computer program product, including a computer program, which implements the alloy formula determination method of the present disclosure when executed by a processor.

[0146] In another exemplary embodiment, a computer program product is further provided. The computer program product includes a computer program executable by a programmable device, and the computer program has a code portion for executing the above-mentioned method for determining an alloy formula when executed by the programmable device.

[0147] Figure 6 6 is a block diagram of an electronic device 600 according to an exemplary embodiment. For example, the electronic device 600 can be provided as a server, a computer, a tablet computer, a mobile terminal, or other electronic devices. Figure 6 The electronic device 600 includes a processing component 622, which further includes one or more processors, and a memory resource represented by a memory 632 for storing instructions, such as applications, that can be executed by the processing component 622. The application stored in the memory 632 may include one or more modules, each corresponding to a set of instructions. In addition, the processing component 622 is configured to execute the instructions to perform the above-mentioned method for determining the alloy formula.

[0148] The electronic device 600 may further include a power supply component 626 configured to perform power management of the electronic device 600, a wired or wireless network interface 650 configured to connect the electronic device 600 to a network, and an input / output interface 658. The electronic device 600 may operate based on an operating system stored in the memory 632, such as Windows Server 200. TM , Mac OS X TM , Unix TM , Linux TM , FreeBSD TM or similar.

[0149] Although terms such as "first", "second" and "third" may be used herein to describe various components, parts, regions, layers or sections, these components, parts, regions, layers or sections are not limited to these terms. On the contrary, these terms are only used to distinguish one component, part, region, layer or section from another component, part, region, layer or section. Therefore, without departing from the teachings of each example, the first component, part, region, layer or section mentioned in the examples described herein may also be referred to as the second component, part, region, layer or section. In addition, the terms "first" and "second" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, the features defined as "first" and "second" can explicitly or implicitly include at least one such feature. In the description herein, the meaning of "multiple" is at least two, for example, two, three, etc., unless otherwise clearly and specifically defined.

[0150] Furthermore, the word "exemplary" is used herein to mean serving as an example, instance, or illustration. Any aspect or design described herein as "exemplary" is not necessarily to be construed as advantageous over other aspects or designs. Rather, the use of the word exemplary is intended to present concepts in a concrete manner. As used herein, the term "or" is intended to mean an inclusive "or" rather than an exclusive "or." That is, unless otherwise specified or clear from the context, "X applies to A or B" is intended to mean any of the natural inclusive permutations. That is, if X applies to A; X applies to B; or X applies to both A and B, then "X applies to A or B" satisfies any of the aforementioned instances. Furthermore, the articles "a" and "an," as used in this application and the appended claims, are generally understood to mean "one or more," unless otherwise specified or clear from the context to refer to the singular form.

[0151] Likewise, although the present disclosure has been shown and described with respect to one or more implementations, equivalent variations and modifications will occur to those skilled in the art upon reading and understanding this specification and the accompanying drawings. The present disclosure includes all such modifications and variations and is limited only by the scope of the claims. With particular regard to the various functions performed by the components described above (e.g., elements, resources, etc.), unless otherwise indicated, terms used to describe such components are intended to correspond to any component (functionally equivalent) that performs the specific function of the described component, even if not structurally equivalent to the disclosed structure. In addition, although particular features of the present disclosure may have been disclosed with respect to only one of several implementations, such features may be combined with one or more other features of other implementations as may be desired and advantageous for any given or particular application. Furthermore, to the extent that the terms "include," "have," "have," "have," or variations thereof are used in the detailed description or claims, such terms are intended to be inclusive in a manner similar to the term "comprising."

[0152] Those skilled in the art will readily appreciate other embodiments of the present disclosure after considering the specification and practicing the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the appended claims.

[0153] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

Claims

1. A method for determining an alloy formula, characterized in that: include: Obtain the target alloy formula to be verified; Inputting the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data, wherein the performance prediction model is used to determine the predicted performance data based on the recipe to be verified and a mapping relationship between recipe parameters and performance parameters; According to the predicted performance data and the target performance requirements corresponding to the target alloy, the formulation to be verified is iteratively optimized to determine the target formulation.

2. The method according to claim 1, characterized in that The training of the performance prediction model includes: Obtaining a sample formula of the target alloy and actual performance data corresponding to the sample formula; Determining sample predicted performance data based on the sample formula; The model parameters of the performance prediction model are updated according to the sample predicted performance data and the actual performance data.

3. The method according to claim 2, characterized in that Updating model parameters of the performance prediction model according to the sample predicted performance data and the actual performance data includes: determining a prediction deviation based on the sample predicted performance data and the actual performance data; According to the prediction deviation, the model parameters of the performance prediction model are updated.

4. The method according to claim 1, wherein Iteratively optimizing the formulation to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine a target formulation includes: In response to the predicted performance data not meeting the target performance requirement, adjusting the recipe to be verified to obtain an updated recipe; The updated formula is used as a new formula to be verified, and the step of inputting the formula to be verified into a pre-trained performance prediction model to obtain predicted performance data is repeated until the predicted performance data of the formula to be verified meets the target performance requirements, and the target formula is determined.

5. The method according to any one of claims 1 to 4, characterized in that Obtain the target alloy formulation to be verified, including: Determining a target alloy formulation to be verified from within a range of formulation parameters to be optimized; Iteratively optimize the formulation to be verified to determine the target formulation, including: Iteratively optimizing the formulation to be verified to obtain an optimized formulation parameter range that meets the target performance requirements; The target formulation is determined according to the optimized formulation parameter range.

6. The method according to claim 5, characterized in that Determining the target formulation according to the optimized formulation parameter range includes: Determining a candidate formulation based on the optimized formulation parameter range; Obtaining measured performance data corresponding to the candidate formula; Based on the measured performance data and the target performance requirements, the candidate formulations are verified to obtain the target formulation.

7. The method according to claim 6, characterized in that Based on the measured performance data and the target performance requirements, the candidate formulation is verified to obtain the target formulation, including: In response to the measured performance data not meeting the target performance requirement, iteratively training the performance prediction model using the candidate recipe and the measured performance data to obtain an updated performance prediction model; The updated performance prediction model is used as the performance prediction model, and the step of inputting the recipe to be verified into the pre-trained performance prediction model to obtain predicted performance data is repeatedly performed to iteratively optimize the recipe to be verified to obtain a new candidate recipe until the test performance data corresponding to the candidate recipe meets the target performance requirements.

8. An alloy formula, characterized in that: The alloy is obtained by the method for determining the alloy formula according to any one of claims 1 to 7.

9. The formulation according to claim 8, characterized in that The alloy is high-strength steel, and the ingredients and their mass percentages in the formula include: Carbon: 0.29~0.44%; Silicon: 0.10~1.0%; Manganese: 0.30~1.5%; Chromium: 0.1~1%; Boron: 0.0005~0.005%; Aluminum: 0.10~0.60%; Niobium: 0.02~0.06%; Vanadium: 0.05~0.20%; Cobalt: 0.001~0.08%; The balance is iron.

10. The formulation according to claim 9, characterized in that The ingredients and their mass percentages in the formula include: Carbon: 0.33~0.40%; Silicon: 0.2~0.5%; Manganese: 0.5~0.8%; Chromium: 0.1~0.5%; Boron: 0.001~0.01%; Aluminum: 0.10~0.40%; Niobium: 0.02~0.05%; Vanadium: 0.05~0.20%; Cobalt: 0.004~0.06%; The balance is iron.

11. The formulation according to claim 9 or 10, characterized in that The ingredients and their mass percentages in the formula include: Carbon: 0.36%; Silicon: 0.33%; Manganese: 0.71%; Chromium: 0.36%; Boron: 0.0025%; Aluminum: 0.33%; Niobium: 0.024%; Vanadium: 0.13%; Cobalt: 0.005%; The balance is iron.

12. The formulation according to claim 9 or 10, characterized in that The process parameters in the formula include the autotempering temperature in the heat treatment parameters, and the autotempering temperature is determined based on multiple components in the formula and their mass percentages, and the relationship between the multiple components and the autotempering temperature; wherein the autotempering temperature is greater than or equal to the set temperature.

13. The formulation according to claim 12, characterized in that The autotempering temperature is greater than or equal to 340 degrees Celsius.

14. A device for determining an alloy formula, characterized in that: include: an acquisition module, configured to acquire a target alloy formula to be verified; an acquisition module configured to input the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data, wherein the performance prediction model is used to determine the predicted performance data based on the recipe to be verified and a mapping relationship between recipe parameters and performance parameters; The determination module is configured to iteratively optimize the formula to be verified based on the predicted performance data and the target performance requirements corresponding to the target alloy to determine the target formula.

15. The device according to claim 14, characterized in that The training of the performance prediction model includes: Obtaining a sample formula of the target alloy and actual performance data corresponding to the sample formula; Determining sample predicted performance data based on the sample formula; The model parameters of the performance prediction model are updated according to the sample predicted performance data and the actual performance data.

16. The device according to claim 14, characterized in that The determining module is further configured to: In response to the predicted performance data not meeting the target performance requirement, adjusting the recipe to be verified to obtain updated recipe data; The updated recipe data is used as a new recipe to be verified, and the step of inputting the recipe to be verified into a pre-trained performance prediction model to obtain predicted performance data is repeated until the predicted performance data meets the target performance requirements, thereby determining the target recipe.

17. An automobile component, characterized in that: The automobile component comprises an alloy made from the alloy formula according to any one of claims 8 to 13.

18. An automobile, characterized in that: The automobile includes the automobile component according to claim 17.

19. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; The processor is configured to execute the method for determining the alloy formula according to any one of claims 1 to 7.

20. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method for determining the alloy formula according to any one of claims 1 to 7 is implemented.

21. A computer program product, characterized in that The invention comprises a computer program, which, when executed by a processor, implements the method for determining the alloy formula according to any one of claims 1 to 7.

Images