Process parameter screening method and device for anti-penetration high-entropy alloy

Abstract

The application provides a process parameter screening method and device for an anti-penetration high-entropy alloy, and relates to the field of armored materials.The method comprises the following steps: obtaining performance indexes based on theoretical and target shooting experiment analysis of an alloy material target plate under normal penetration, and taking heat treatment process parameters of solid solution and aging as characteristic factors to design an orthogonal experiment to obtain a data set; performing standardization processing on characteristic values in the data set, and performing transformation on the characteristics; performing redundancy judgment and selection on time characteristics and temperature characteristics before and after the transformation; establishing a machine learning model, and performing generalization performance evaluation on the model through a root mean square error and a determination coefficient; and establishing a multi-objective optimization method based on a Pareto front and a multi-dimensional joint probability distribution function to realize multi-objective fusion and screening of process parameters.The data set in the application is completely obtained from experiments, which reduces generalization errors generated when a machine learning model is built for a small data set, and can accurately screen required process parameters in a short time.

Description

TECHNICAL FIELD

[0001] The application relates to the field of armor materials, in particular to a process parameter screening method and device for anti-penetration high-entropy alloys. BACKGROUND

[0002] At present, in the field of armor materials, the sensitivity of material performance to process parameters is very high, and the selection space of process parameters is very large, so it is very difficult to screen out process parameters that make alloy materials have high strength and toughness performance in a short time.

[0003] Moreover, limited by computing power, the parameters of the controllable algorithm model are limited, and the data set available in the studied system is small, so the uncertainty of the machine learning model prediction in the small data set will be greatly increased, so whether the uncertainty of the model can be reduced is a problem. SUMMARY

[0004] In view of the above problems, a process parameter screening method and device for anti-penetration high-entropy alloys are provided to automatically and accurately screen process parameters.

[0005] The first aspect of the application provides a process parameter screening method for anti-penetration high-entropy alloys, comprising:

[0006] Based on the theory and target experiment analysis under the target plate penetration of the alloy material, the performance index related to the material penetration resistance is obtained, and the solid solution and aging heat treatment process parameters are designed as characteristic factors for orthogonal experiment to obtain a data set, wherein the performance index includes tensile strength, elongation after fracture and impact work.

[0007] The characteristic values in the data set are standardized, the time characteristics and temperature characteristics are transformed based on the JMAK isothermal kinetics model and the presence of precipitated phase, the time characteristics and temperature characteristics before and after the transformation are judged for redundancy and selected by the mutual information method and the Pearson correlation coefficient method.

[0008] Through cross-validation and random grid search method, machine learning models are established for the process parameters and corresponding performance indexes respectively, and the generalization performance of each model is evaluated by root mean square error RMSE and determination coefficient R2, and the modeling method is determined.

[0009] A multi-objective optimization method based on Pareto front and multi-dimensional joint probability distribution function is established to realize multi-objective fusion and realize screening of the process parameters.

[0010] Optionally, the solid solution and aging heat treatment process parameters are designed as characteristic factors for orthogonal experiment to obtain a data set, comprising:

[0011] For the solid solution temperature, the solid solution time, the aging temperature and the aging time, the initial range and the level number of each characteristic factor are determined, and a plurality of test combinations are designed based on an orthogonal table;

[0012] Based on each test combination, corresponding heat treatment, static tensile test and dynamic impact test are performed on the material, and performance indexes under each test combination are obtained;

[0013] For different performance indexes, range analysis and variance analysis are performed on each characteristic factor to obtain the significance level of each characteristic factor on each performance index.

[0014] Optionally, the time characteristic and the temperature characteristic are transformed based on the JMAK isothermal kinetics model and whether the precipitated phase exists, and the time characteristic and the temperature characteristic before and after the transformation are judged for redundancy and selected by mutual information method and Pearson correlation coefficient method, including:

[0015] Based on the equation transformation of the JMAK isothermal kinetics equation, the time characteristic is logarithmically transformed to realize the mapping relationship between the time characteristic and the performance index;

[0016] According to the geometric characteristics of the Sigmoid function, the temperature characteristic is transformed to judge whether the precipitated strengthening phase exists;

[0017] The correlation of each characteristic with the tensile strength, elongation at break and impact work is calculated by the mutual information method, and the characteristics before and after the transformation are judged for redundancy by the Pearson correlation coefficient method, wherein the characteristics with a correlation coefficient greater than a preset threshold are discarded.

[0018] Optionally, the process parameters and the corresponding performance indexes are respectively established by machine learning models by cross-validation and random grid search method, and the generalization performance of each model is evaluated by RMSE and R2 to determine the modeling method, including:

[0019] The data set is divided by introducing a bootstrap sampling strategy, and the evaluation index is averaged for each machine learning algorithm according to a plurality of repeated sampling times;

[0020] A model ensemble method based on meta-learning is introduced, the three best single models are selected for integration, and the modeling method is determined.

[0021] Optionally, a multi-objective optimization method based on Pareto frontier and multi-dimensional joint probability distribution function is established to realize multi-objective fusion and realize the screening of the process parameters, including:

[0022] The Pareto frontier points of the performance indexes in the data set are obtained by a non-dominated sorting algorithm as the upper and lower limits of the multi-dimensional joint probability distribution function;

[0023] Based on the selected machine learning method, a machine learning model is built under different data sets by using a bootstrap sampling strategy to obtain the performance distribution of the to-be-predicted process parameters, and the mean and variance of each performance distribution are obtained;

[0024] Based on the concept of improvement expectation, a covariance matrix is introduced as a probability density function, the performance target is fused into a function, and the improvement expectation value of each group of to-be-predicted process parameters is calculated, and the screening of the process parameters is realized through the size of the improvement expectation value.

[0025] Optionally, the method further comprises:

[0026] Based on the idea of active learning, the screened process parameters are verified by experiments, new data is added to the data set, the data set is updated, and the iterative optimization of the model is realized.

[0027] The second aspect of the application proposes a process parameter screening device for anti-penetration high-entropy alloy, characterized by comprising:

[0028] The analysis module is used for obtaining performance indicators related to material anti-penetration based on theoretical and target shooting experiment analysis under the condition of positive penetration of the alloy material target plate, and performing orthogonal experiment on the heat treatment process parameters of solid solution and aging as characteristic factors to obtain a data set, wherein the performance indicators include tensile strength, elongation after fracture and impact work.

[0029] The selection module is used for standardizing the characteristic values in the data set, transforming the time characteristics and temperature characteristics based on the JMAK isothermal kinetics model and the presence or absence of precipitated phase, and judging and selecting the time characteristics and temperature characteristics before and after the transformation through the mutual information method and the Pearson correlation coefficient method.

[0030] The evaluation module is used for establishing machine learning models for the process parameters and corresponding performance indicators respectively through cross-validation and random grid search method, and evaluating the generalization performance of each model through root mean square error RMSE and determination coefficient R2 to determine the modeling method.

[0031] The screening module is used for establishing a multi-objective optimization method based on Pareto front and multi-dimensional joint probability distribution function to realize multi-objective fusion and realize screening of the process parameters.

[0032] The third aspect of the application proposes a computer device, comprising a memory, a processor and a computer program stored on the memory and executable on the processor, and when the processor executes the computer program, the method of any one of the above first aspect is realized.

[0033] The fourth aspect of the present application provides a non-transitory computer-readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method of any one of the above first aspects.

[0034] The embodiments of the present application provide at least the following beneficial effects:

[0035] Through the method provided by the present application, the process parameters with excellent comprehensive mechanical properties can be screened for alloy materials in a short time. The data set used by the present application is completely from experiments, without considering some uncontrollable factors such as experimental environment and experimental equipment when searching for data in literature. On the other hand, the generalization error generated when building a machine learning model for a small data set is reduced based on the distribution of model prediction values. Finally, the synergistic optimization of multiple objectives is completed by combining the Pareto frontier and the multi-dimensional probability distribution function, thereby ensuring the comprehensive performance of the material.

[0036] Additional aspects and advantages of the present application will be in part apparent and in part pointed out hereinafter in the description of the application. BRIEF DESCRIPTION OF DRAWINGS

[0037] The above and / or additional aspects and advantages of the present application will become apparent and be readily appreciated from the following description of the embodiments, taken in conjunction with the accompanying drawings.

[0038] Figure 1 FIG. 1 is a flowchart of a process parameter screening method for an anti-penetration high-entropy alloy according to an embodiment of the present application;

[0039] Figure 2 FIG. 2 is a flowchart of obtaining a data set according to an embodiment of the present application;

[0040] Figure 3 FIG. 3 is a structural diagram of obtaining a data set according to an embodiment of the present application;

[0041] Figure 4 FIG. 4 is a flowchart of establishing a search space of process parameters according to an embodiment of the present application;

[0042] Figure 5 FIG. 5 is a distribution diagram of mutual information values between process parameters and performance indicators according to an embodiment of the present application;

[0043] Figure 6 FIG. 6 is a schematic diagram of Pearson correlation between process parameter features according to an embodiment of the present application;

[0044] Figure 7 FIG. 7 is a schematic diagram of Pearson correlation between process parameter features and performance indicators according to an embodiment of the present application;

[0045] Figure 8 This is a line graph showing the test error of each model under cross-validation based on different training set ratios, according to an embodiment of this application.

[0046] Figure 9(a) is a line graph showing the generalization performance of each model in tensile strength prediction under cross-validation with a training set ratio of 0.75, according to an embodiment of this application.

[0047] Figure 9(b) is a scatter plot of tensile strength prediction of the GBDT model under cross-validation according to an embodiment of this application;

[0048] Figure 10 This is a line graph showing the test error of each model predicting elongation after fracture under cross-validation based on different training set ratios, according to an embodiment of this application.

[0049] Figure 11(a) is a line graph showing the generalization performance of each model in predicting elongation after fracture under cross-validation with a training set ratio of 0.75, according to an embodiment of this application.

[0050] Figure 11(b) is a scatter plot of the elongation prediction after fracture of the GBDT model under cross-validation according to an embodiment of this application.

[0051] Figure 12 This is a line graph showing the test error of each model predicting impact energy under different training set ratios under cross-validation, according to an embodiment of this application.

[0052] Figure 13(a) is a line graph showing the generalization performance of each model in predicting impact energy under cross-validation with a training set ratio of 0.75, according to an embodiment of this application.

[0053] Figure 13(b) is a scatter plot of the impact energy prediction of the GBDT model under cross-validation according to an embodiment of this application;

[0054] Figure 14(a) is a bar chart of the root mean square error of tensile strength prediction for each model under the self-sampling strategy according to the embodiments of this application;

[0055] Figure 14(b) is a bar chart of the determination coefficients of tensile strength prediction for each model under the self-sampling strategy according to the embodiments of this application;

[0056] Figure 15(a) is a bar chart of the root mean square error of the elongation prediction after fracture for each model under the self-sampling strategy shown in the embodiments of this application;

[0057] Figure 15(b) is a bar chart of the determination coefficients for elongation prediction after fracture of each model under the self-sampling strategy shown in the embodiments of this application;

[0058] FIG. 16(a) is a column chart of root mean square error of impact energy prediction of each model under a self-sampling strategy according to an embodiment of the present application;

[0059] FIG. 16(b) is a column chart of determination coefficient of impact energy prediction of each model under a self-sampling strategy according to an embodiment of the present application;

[0060] Figure 17 is a model integration flowchart under stack generalization according to an embodiment of the present application;

[0061] FIG. 18(a) is a column chart of root mean square error of tensile strength prediction of Ext, RF, GBDT and STG under a self-sampling strategy according to an embodiment of the present application;

[0062] FIG. 18(b) is a column chart of determination coefficient of tensile strength prediction of Ext, RF, GBDT and STG under a self-sampling strategy according to an embodiment of the present application;

[0063] FIG. 19(a) is a column chart of root mean square error of elongation after break prediction of Ext, RF, GBDT and STG under a self-sampling strategy according to an embodiment of the present application;

[0064] FIG. 19(b) is a column chart of determination coefficient of elongation after break prediction of Ext, RF, GBDT and STG under a self-sampling strategy according to an embodiment of the present application;

[0065] FIG. 20(a) is a column chart of root mean square error of impact energy prediction of Ext, RF, GBDT and STG under a self-sampling strategy according to an embodiment of the present application;

[0066] FIG. 20(b) is a column chart of determination coefficient of impact energy prediction of Ext, RF, GBDT and STG under a self-sampling strategy according to an embodiment of the present application;

[0067] Figure 21 is a flowchart of process parameter screening according to an embodiment of the present application;

[0068] Figure 22 is a flowchart of a non-dominated sorting algorithm for obtaining a Pareto frontier point according to an embodiment of the present application;

[0069] Figure 23 is a flowchart of a multi-objective optimization algorithm according to an embodiment of the present application;

[0070] Figure 24 is a block diagram of a process parameter screening device for an anti-penetration high-entropy alloy according to an embodiment of the present application;

[0071] Figure 25 is a block diagram of an electronic device. DETAILED DESCRIPTION

[0072] Embodiments of the present application are described in detail below with reference to the accompanying drawings, in which the same or like elements or elements having the same or similar functions are denoted by the same reference numerals throughout the drawings. The embodiments described below with reference to the drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0073] Figure 1 is a process parameter screening method of an anti-penetration high-entropy alloy according to an embodiment of the present application, comprising:

[0074] In step 101, based on the theory and target shooting experiment analysis under the target shooting of the alloy material, the performance index related to the material anti-penetration is obtained, and the heat treatment process parameters of solid solution and aging are taken as characteristic factors to design an orthogonal experiment, and a data set is obtained, wherein the performance index includes tensile strength, elongation after fracture and impact work.

[0075] In the embodiments of the present application, the following conclusions are obtained based on the target shooting experiment analysis: the target plate with good toughness and plasticity has smaller penetration depth.

[0076] Optionally, as shown in Figure 2 , step 101 further comprises:

[0077] In step 201, the initial range and the number of levels of each characteristic factor are determined for the solid solution temperature, the solid solution time, the aging temperature and the aging time, and a plurality of test combinations are designed based on an orthogonal table.

[0078] In the embodiments of the present application, as shown in Figure 3 and Figure 4 , for the solid solution temperature T 固溶 , the solid solution time t 固溶 , the aging temperature T 时效 and the aging time t 时效 , if in an embodiment, T 固溶 ∈[820, 860, 5], t 固溶 ∈[30, 60, 10], T 时效 ∈[450, 520, 5] and t 时效 ∈[10, 30, 3], take four parameters as a group, store the parameter combinations, and design 16 experimental combinations.

[0079] It should be noted that the parameter combination storage is divided into the following cases:

[0080] If each group of parameters is combined in order, after storing the correct parameter combination, it is judged whether the traversal is ended, if the traversal is ended, the storage is terminated, otherwise the parameter combination storage is continued;

[0081] If the parameter combination is not combined in order, the parameter combination is discarded, and it is judged whether the traversal is ended, if the traversal is ended, the storage is terminated, otherwise the parameter combination storage is continued.

[0082] In step 202, the material is subjected to corresponding heat treatment, static tensile test and dynamic impact test based on each test combination, and the performance index under each test combination is obtained.

[0083] As shown in the embodiment of the application, Figure 3 the process of heat treatment is in turn solution, water quenching, aging and air cooling.

[0084] In step 203, range analysis and variance analysis are performed on each characteristic factor for different performance indexes, and the significance level of each characteristic factor on each performance index is obtained.

[0085] In a possible embodiment, if the table lookup obtains that the degree of freedom of A factor is 3 and the total error degree of freedom is 19.

[0086] The threshold value of the statistic quantity is set according to the variance value, that is, α = 0.01, F 0.01 = 5.01 and α = 0.05, F 0.05 = 3.13.

[0087] If the statistic quantity F of A factor satisfies F > F 0.01 , the influence of A factor is highly significant;

[0088] If the statistic quantity F of A factor satisfies F 0.05 ≤ F ≤ F 0.01 , the influence of A factor is significant;

[0089] If the statistic quantity F of A factor satisfies F < F 0.05 , the influence of A factor is not significant.

[0090] In step 102, the characteristic values in the data set are subjected to standardization processing, the time characteristic and the temperature characteristic are transformed based on the JMAK isothermal kinetics model and whether the precipitated phase exists, and the time characteristic and the temperature characteristic before and after the transformation are subjected to redundancy judgment and selection through the mutual information method and the Pearson correlation coefficient method.

[0091] In the embodiment of the application, the formula for standardization processing of the characteristics in the data set is as follows:

[0092]

[0093] Wherein, is the scaled value of the i th data of the j th variable, is the unscaled value of the i th data of the j th variable, μ (j) is the mean value of the j th variable, and σ (j)Standard deviation of the jth variable;

[0094] The formula of JMAK isothermal kinetics model is as follows:

[0095]

[0096] Wherein, Y is the volume fraction of new phase, t is the heat treatment time, K and n are model parameters related to temperature;

[0097] In the embodiment of the application, through equivalent transformation of the JMAK isothermal kinetics model, the logarithmic transformation of time t is obtained, and the formula is as follows:

[0098] ln(ln(1-Y))=ln K+n ln t.

[0099] Secondly, whether the precipitated strengthening phase is precipitated at different temperatures is judged by a variant of the sigmoid function, so as to transform the temperature characteristics, and the formula of the variant of the sigmoid function is as follows:

[0100]

[0101] Wherein, T is the heat treatment temperature, and θ is the precipitation temperature of the compound.

[0102] It should be noted that the correlation of each feature with the tensile strength, elongation at break and impact work is calculated by the mutual information method, the redundancy of the features before and after transformation is judged by the Pearson correlation coefficient method, wherein the features with a correlation coefficient greater than a preset threshold are discarded.

[0103] In a possible embodiment, the mutual information values between the process parameters and the performance indicators are as shown in Figure 5 .

[0104] In a possible embodiment, the Pearson correlation between the process parameter features is as shown in Figure 6 .

[0105] In a possible embodiment, the Pearson correlation between the process parameter features and the performance indicators is as shown in Figure 7 .

[0106] In a possible embodiment, the features with a correlation coefficient greater than 0.95 are discarded.

[0107] Step 103, through cross-validation and random grid search method, machine learning models are respectively established for the process parameters and the corresponding performance indicators, and the generalization performance of each model is evaluated through root mean square error RMSE and determination coefficient R2, and the modeling method is determined.

[0108] In the embodiments of the present application, in order to solve the deviation problem caused by the inconsistent size of the training set due to the selection of the K value in K-fold cross-validation, a bootstrap sampling strategy is introduced for data set division. In order to reduce the randomness problem caused by single sampling, the sampling is repeated 100 times, and the mean value of the evaluation index is taken to evaluate each machine learning algorithm. A model ensemble method based on meta-learning is introduced, and three models with good performance are selected for ensemble to reduce the prediction error of the model in the small data set.

[0109] In the embodiments of the present application, seven models of Ext, RF, Svr.r, Ada, GBDT, Lasso and BPNN are introduced.

[0110] In a possible embodiment, the schematic diagram of the generalization performance evaluation process of each model under cross-validation is as shown in Figure 8 , FIG. 9(a), FIG. 9(b), Figure 10 , FIG. 11(a), FIG. 11(b), Figure 12 , FIG. 13(a), FIG. 13(b).

[0111] In a possible embodiment, the performance diagrams of the bootstrap sampling strategy anti-tensile strength prediction model, the post-break elongation prediction model and the impact work prediction model are as shown in FIG. 14(a), FIG. 14(b), FIG. 15(a), FIG. 15(b), FIG. 16(a) and FIG. 16(b).

[0112] In a possible embodiment, the model ensemble method of meta-learning is as shown in Figure 17 .

[0113] Based on the above evaluation process and performance diagram, the Ext, RF and GBDT with good performance are selected for ensemble, and the ensemble model is STG.

[0114] In a possible embodiment, the diagrams of the Ext, RF, GBDT and STG under the bootstrap sampling strategy in the anti-tensile strength prediction, the post-break elongation prediction and the impact work prediction are as shown in FIG. 18(a), FIG. 18(b), FIG. 19(a), FIG. 19(b), FIG. 20(a) and FIG. 20(b).

[0115] In step 104, a multi-objective optimization method based on the Pareto frontier and the multi-dimensional joint probability distribution function is used to realize multi-objective fusion and realize the screening of process parameters.

[0116] In the embodiments of the present application, as shown in Figure 21 , step 104 further includes:

[0117] In step 301, the Pareto frontier points of the performance index in the data set are obtained by using a non-dominated sorting algorithm, as the upper and lower limits of the multi-dimensional joint probability distribution function.

[0118] In one possible embodiment, a flow chart of a non-dominated sorting algorithm for obtaining a Pareto front is shown in Figure 22 .

[0119] At step 302, based on the selected machine learning method, a machine learning model is built under different data sets using a bootstrap sampling strategy to obtain the performance distribution of the to-be-predicted process parameters, and the mean and variance of each performance distribution are obtained.

[0120] At step 303, based on the concept of improvement expectation, a covariance matrix is introduced as a probability density function, the performance target is integrated into a function, and the improvement expectation value of each group of to-be-predicted process parameters is calculated, and the screening of process parameters is realized through the size of the improvement expectation value.

[0121] In the embodiments of the present application, a flow chart of a multi-objective optimization algorithm is shown in Figure 23 , and the steps include:

[0122] Based on the process parameter set X 训练 , the performance index set Y 训练 of the training model, the to-be-predicted process parameter set X 预测 , and the set Y pareto of the original data set Pareto points, a machine learning model is trained: f(X 训练 )=Y 训练 .

[0123] The improvement expectation of the performance corresponding to X 预测 is calculated, and the data distribution T(X i ), E(X i ), and K(X i ) of three performances are obtained through bootstrap sampling training model, and the predicted mean [μ i i , μ i E, μ iK ] of the performance of each group of process parameters X T and the covariance matrix cov i( T, E, K) are calculated.

[0124] The improvement probability P(I)x i of the comprehensive performance of X i is calculated: P(I)x i =P(μ i ,cov pareto ,Y i ), and the improvement size I i is calculated: I T =max(min(μ pareto T -Y i E -Ypareto E ,μ i K -Y pareto K

[0125] By X i The improvement probability and the improvement size of the comprehensive performance of X i The improvement expectation of the comprehensive performance of X: E(I)x i =P(I)x i *I i .

[0126] Thus, the screening of the process parameters is realized.

[0127] In addition, based on the idea of active learning, the screened process parameters are verified by experiments, and new data are added to the data set, the data set is updated, and the iterative optimization of the model is realized, so as to solve the poor generalization ability of the model with small data samples.

[0128] In a possible embodiment, the iterative optimization of the screened process parameters and the mechanical property experimental values thereof are shown in Table 1.

[0129]

[0130] Table 1

[0131] In a possible embodiment, the percentage of improvement of the comprehensive performance after optimization of the process parameters is shown in Table 2.

[0132]

[0133] Table 2

[0134] Wherein, ~ represents that the percentage of improvement is within 1%.

[0135] The application can screen process parameters with excellent comprehensive mechanical properties for alloy materials in a short time, and the data set used by the application is completely from experiments, without considering some uncontrollable factors such as experimental environment and experimental equipment when searching for data in literature. On the other hand, based on the distribution of model prediction values, the generalization error generated when building a machine learning model for a small data set is reduced. Finally, the synergistic optimization of multiple objectives is completed by combining the Pareto frontier and the multi-dimensional probability distribution function, and the comprehensive performance of the material is guaranteed.

[0136] Figure 24 is a block diagram of a process parameter screening device 400 of an anti-penetration high-entropy alloy according to an embodiment of the application, which comprises an analysis module 410, a selection module 420, an evaluation module 430 and a screening module 440.

[0137] ​The analysis module 410 is configured to obtain performance indexes related to material penetration resistance based on theoretical analysis and target penetration experiment analysis of the alloy material target plate under positive penetration, and take heat treatment process parameters of solid solution and aging as characteristic factors to design an orthogonal experiment, and obtain a data set, wherein the performance indexes include tensile strength, elongation after fracture and impact work.

[0138] The selection module 420 is configured to perform standardization processing on characteristic values in the data set, transform time characteristics and temperature characteristics based on the JMAK isothermal kinetics model and whether the precipitated phase exists, and perform redundancy judgment and selection on the time characteristics and the temperature characteristics before and after the transformation through the mutual information method and the Pearson correlation coefficient method.

[0139] The evaluation module 430 is configured to establish machine learning models for the process parameters and corresponding performance indexes respectively through cross-validation and random grid search methods, and perform generalization performance evaluation on each model through root mean square error (RMSE) and a decision coefficient (R2), and determine a modeling method.

[0140] The screening module 440 is configured to establish a multi-objective optimization method based on a Pareto front and a multi-dimensional joint probability distribution function to realize multi-objective fusion, and realize screening of the process parameters.

[0141] As to the apparatus in the above-mentioned embodiments, the specific manner in which each module performs operations has been described in detail in the embodiments related to the method, and will not be described in detail here.

[0142] Figure 25 A schematic block diagram of an example electronic device 500 that can be used to implement embodiments of the present disclosure is shown. The electronic device is intended to represent various forms of digital computers, such as laptops, desktops, tablets, personal digital assistants, servers, blade servers, mainframes, and other appropriate computers. The electronic device can also represent various forms of mobile devices, such as personal digital assistants, cellular telephones, smartphones, wearable devices, and other similar computing devices. The components shown here, their connections and relationships, and their functions, are meant to be examples only, and are not meant to limit implementations of the present disclosure described and / or claimed in this document.

[0143] As Figure 25As shown, the device 500 includes a computing unit 501 that can perform various appropriate actions and processes in accordance with a computer program stored in a read-only memory (ROM) 502 or a computer program loaded from a storage unit 503 into a random access memory (RAM) 503. In the RAM 503, various programs and data required for the operation of the device 500 can also be stored. The computing unit 501, the ROM 502, and the RAM 503 are connected to each other through a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0144] A plurality of components in the device 500 are connected to the I / O interface 505, including an input unit 506 such as a keyboard, a mouse, etc., an output unit 507 such as various types of displays, speakers, etc., a storage unit 508 such as a magnetic disk, an optical disk, etc., and a communication unit 509 such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 509 allows the device 500 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0145] The computing unit 501 can be various general and / or special purpose processing components having processing and computing capabilities. Some examples of the computing unit 501 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various computing units running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The computing unit 501 performs various methods and processes described above, such as the voice instruction response method. For example, in some embodiments, the voice instruction response method can be implemented as a computer software program that is tangibly embodied in a machine-readable medium, such as the storage unit 508. In some embodiments, part or all of the computer program can be loaded and / or installed on the device 500 via the ROM 502 and / or the communication unit 509. When the computer program is loaded into the RAM 503 and executed by the computing unit 501, one or more steps of the voice instruction response method described above can be performed. Alternatively, in other embodiments, the computing unit 501 can be configured to perform the voice instruction response method by any other appropriate means, such as by means of firmware.

[0146] The various embodiments of the systems and techniques described above can be implemented in digital electronic circuitry, integrated circuitry, a field programmable gate array (FPGA), an application specific integrated circuit (ASIC), a system on a chip (SOC), a load programmable logic device (CPLD), computer hardware, firmware, software, and / or combinations thereof. These various embodiments can include implementation in one or more computer programs that are executable and / or interpretable on a programmable system including at least one programmable processor, which can be special or general purpose, coupled to receive data and instructions from, and to transmit data and instructions to, a storage system, at least one input device, and at least one output device.

[0147] Program code for carrying out methods of the present disclosure can be written in any combination of one or more programming languages. The program code can be provided to a processor or controller of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the program code, when executed by the processor or controller, produces the functions / operations specified in the flowcharts and / or the block diagrams. The program code can be executed entirely on a machine, partially on a machine, partially on a machine as a stand-alone software package, or entirely on a remote machine or server.

[0148] In the context of the present disclosure, a machine-readable medium can be a tangible medium that contains or stores a program for use by or in connection with an instruction execution system, apparatus, or device. The machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium can include but is not limited to an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any suitable combination of the foregoing. More specific examples of the machine-readable storage medium will include one or more lines of electrical connections, portable computer disks, hard disk drives, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or Flash memory), optical fibers, portable compact disc read-only memories (CD-ROMs), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0149] To provide for interaction with a user, the systems and techniques described here can be implemented on a computer having a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user and a keyboard and a pointing device (e.g., a mouse or a trackball) by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form, including acoustic, speech, or tactile input.

[0150] The systems and techniques described here can be implemented in a computing system that includes a back end component (e.g., as a data server), or that includes a middleware component (e.g., an application server), or that includes a front end component (e.g., a user computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the systems and techniques described here), or any combination of such back end, middleware, or front end components. The components of the system can be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network (LAN), a wide area network (WAN), the Internet, and a blockchain network.

[0151] The computer system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. A server can be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system, to solve the defects of large management difficulty and weak business scalability in traditional physical host and VPS service ("Virtual Private Server", or simply "VPS"). The server can also be a server of a distributed system, or a server combined with a blockchain.

[0152] It should be understood that various forms of flow shown above can be used, with steps reordered, added, or removed. For example, the steps recited in the present disclosure can be performed in parallel, in series, or in a different order, without limitation herein, as long as the desired results of the technical solutions of the present disclosure can be achieved.

[0153] The above detailed description does not limit the scope of the disclosure. Various modifications, combinations, sub-combinations and alternatives can be made to the detailed description. Any modification, equivalent replacement and improvement etc. made within the spirit and principle of the disclosure shall be included in the scope of the disclosure.

Claims

1. A method for screening process parameters of a penetration-resistant high-entropy alloy, characterized in that, include: Based on theoretical and experimental analysis of the penetration resistance of alloy material target plates, performance indicators related to the penetration resistance of materials are obtained. Orthogonal experiments are designed with solution treatment and aging heat treatment process parameters as characteristic factors to obtain a dataset. The performance indicators include tensile strength, elongation after fracture, and impact energy. The feature values ​​in the dataset are standardized, and the time and temperature features are transformed based on the JMAK isothermal kinetic model and the presence of precipitates. The redundancy of the time and temperature features before and after the transformation is judged and selected using the mutual information method and the Pearson correlation coefficient method. Machine learning models were established for the process parameters and corresponding performance indicators using cross-validation and random grid search methods. The generalization performance of each model was evaluated by the root mean square error (RMSE) and the coefficient of determination (R²) to determine the modeling method. A multi-objective optimization method based on Pareto front and multidimensional joint probability distribution function is established to achieve multi-objective fusion and to screen the process parameters. The JMAK isothermal kinetic model and whether the precipitated phase transforms the time and temperature characteristics are determined by using mutual information and Pearson correlation coefficient methods to assess redundancy and select the appropriate characteristics before and after the transformation, including: Based on the equation transformation of the JMAK isothermal kinetic equation, a logarithmic transformation is performed on the time characteristics to realize the mapping relationship between the time characteristics and performance indicators. Based on the geometric characteristics of the Sigmoid function, the temperature features are transformed to determine whether a strengthening phase precipitates. The correlation between each feature and tensile strength, elongation after fracture, and impact energy is calculated using the mutual information method. The redundancy of the features before and after transformation is judged by the Pearson correlation coefficient method. Features with a correlation coefficient greater than a preset threshold are discarded. The process involves establishing machine learning models for the process parameters and corresponding performance indicators using cross-validation and random grid search methods, and evaluating the generalization performance of each model using RMSE and R2 to determine the modeling method, including: A self-sampling strategy is introduced to divide the dataset, and multiple repeated samplings are performed according to a preset number of times. The average value of the evaluation index is used to evaluate each machine learning algorithm. A model ensemble method based on meta-learning is introduced, which selects the three models with the best single-model performance for ensemble to determine the modeling method.

2. The method according to claim 1, characterized in that, The orthogonal experiment designed using the heat treatment process parameters of solution treatment and aging as characteristic factors yields a dataset, including: For solution temperature, solution time, aging temperature and aging time, determine the initial range and level number of each characteristic factor, and design various experimental combinations based on orthogonal arrays; Based on each test combination, the material is subjected to corresponding heat treatment, static tensile test and dynamic impact test to obtain the performance index under each test combination. For each of the aforementioned performance indicators, range analysis and variance analysis were performed on each characteristic factor to obtain the significance level of each characteristic factor on each performance indicator.

3. The method according to claim 1, characterized in that, The establishment of a multi-objective optimization method based on Pareto front and multidimensional joint probability distribution function to achieve multi-objective fusion and to screen the process parameters includes: The Pareto front points of the performance indicators in the dataset are obtained through a non-dominated sorting algorithm, which serve as the upper and lower bounds of the multidimensional joint probability distribution function. Based on the selected machine learning method, a bootstrap sampling strategy is used to build machine learning models for different datasets to obtain the performance distributions of the process parameters to be predicted, and to obtain the mean and variance of each performance distribution. Based on the concept of improvement expectation, the covariance matrix is ​​introduced as a probability density function to integrate the performance target into a function, and the improvement expectation value of each set of process parameters to be predicted is calculated. The process parameters are then selected by the magnitude of the improvement expectation value.

4. The method according to claim 3, characterized in that, The method further includes: Based on the idea of ​​active learning, the selected process parameters are experimentally verified, new data is added to the dataset, the dataset is updated, and the model is iteratively optimized.

5. A process parameter screening device for high-entropy alloys resistant to penetration, characterized in that, include: The analysis module is used for theoretical and target-hitting experimental analysis based on the positive penetration of alloy material target plates to obtain performance indicators related to the penetration resistance of materials. Orthogonal experiments are designed with solid solution and aging heat treatment process parameters as characteristic factors to obtain a dataset. The performance indicators include tensile strength, elongation after fracture, and impact energy. The selection module is used to standardize the feature values ​​in the dataset, transform the time and temperature features based on the JMAK isothermal kinetic model and the presence of precipitates, and use the mutual information method and Pearson correlation coefficient method to determine the redundancy of the time and temperature features before and after the transformation and select them. The evaluation module is used to establish machine learning models for the process parameters and corresponding performance indicators through cross-validation and random grid search methods, and to evaluate the generalization performance of each model through root mean square error (RMSE) and coefficient of determination (R²) to determine the modeling method. The screening module is used to establish a multi-objective optimization method based on Pareto front and multidimensional joint probability distribution function to achieve multi-objective fusion and screen the process parameters. The JMAK isothermal kinetic model and whether the precipitated phase transforms the time and temperature characteristics are determined by using mutual information and Pearson correlation coefficient methods to assess redundancy and select the appropriate characteristics before and after the transformation, including: Based on the equation transformation of the JMAK isothermal kinetic equation, a logarithmic transformation is performed on the time characteristics to realize the mapping relationship between the time characteristics and performance indicators. Based on the geometric characteristics of the Sigmoid function, the temperature features are transformed to determine whether a strengthening phase precipitates. The correlation between each feature and tensile strength, elongation after fracture, and impact energy is calculated using the mutual information method. The redundancy of the features before and after transformation is judged by the Pearson correlation coefficient method. Features with a correlation coefficient greater than a preset threshold are discarded. The process involves establishing machine learning models for the process parameters and corresponding performance indicators using cross-validation and random grid search methods, and evaluating the generalization performance of each model using RMSE and R2 to determine the modeling method, including: A self-sampling strategy is introduced to divide the dataset, and multiple repeated samplings are performed according to a preset number of times. The average value of the evaluation index is used to evaluate each machine learning algorithm. A model ensemble method based on meta-learning is introduced, which selects the three models with the best single-model performance for ensemble to determine the modeling method.

6. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, it implements the method as described in any one of claims 1-4.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-4.

Images