Diamond wire drawing heat treatment process-performance data model and strength plasticity intelligent optimization method

Abstract

The invention discloses a diamond wire drawing heat treatment process-performance data model and strength plasticity intelligent optimization method, and belongs to the technical field of heat treatment process design. The problem that in the prior art, a traditional drawing heat treatment process method is difficult to achieve collaborative optimization of high strength and high plasticity is solved. The method comprises the following steps: carrying out a drawing recrystallization annealing experiment on a diamond wire, and collecting drawing heat treatment process parameters and performance data in the experiment to obtain a data set; dividing a data set after data processing, constructing a drawing heat treatment process-performance machine learning model through a Gaussian process regression algorithm, and screening the model by adopting evaluation indexes to obtain a final drawing heat treatment process-performance machine learning model; and a reverse design target is set, and the drawing process meeting different performance requirements is obtained. The method effectively avoids the situation that the diamond wire drawing heat treatment process is difficult to achieve optimal balance between strength and plasticity, and can be applied to design of the diamond wire drawing heat treatment process.

Description

Technical Field

[0001] The invention relates to a diamond wire drawing heat treatment process-performance data model and a strong-plasticity intelligent optimization method, and belongs to the technical field of heat treatment process design. Background Art

[0002] Diamond wire is an advanced cutting tool, mainly used for cutting and processing hard and brittle materials such as silicon wafers, sapphire, optical glass and magnetic materials. Diamond wire gives it excellent cutting ability by firmly attaching diamond particles to the surface of ultra-fine high-strength steel wire or other substrates. The core of diamond wire is an extremely fine steel wire with a diameter usually between tens of microns and hundreds of microns. It has extremely high tensile strength and plasticity and can withstand the huge stress generated during high-speed operation. Diamond particles are evenly distributed on the surface of the steel wire through processes such as electroplating, electrodeposition, or resin bonding, so that the steel wire has the ability to cut hard and brittle materials. During the cutting process, the diamond wire is tightened and passes through the material to be cut at an extremely high speed. The high hardness and sharp edges of the diamond particles can effectively destroy the atomic structure of the material surface, gradually cut the material, and form an extremely thin and flat cutting surface. Due to the fineness and high precision of the diamond wire, it can cut very thin sheets and significantly reduce material loss, which is particularly beneficial for cutting expensive or rare materials.

[0003] Diamond wire plays a vital role in the solar photovoltaic industry, and is mainly used to cut silicon rods into silicon wafers. Silicon wafers are the basic material for manufacturing solar cells, and the use of diamond wire can significantly improve cutting efficiency, reduce silicon wafer loss, and thus reduce production costs. In the semiconductor manufacturing process, diamond wire is used to cut high-hardness materials such as silicon wafers and sapphire substrates. Since semiconductor manufacturing requires extremely high precision, diamond wire can provide smooth and high-precision cutting effects to meet the strict requirements of the industry; in LED manufacturing, the cutting of sapphire substrates is one of the key steps, and diamond wire can provide faster cutting speeds and higher cutting quality, thereby improving the production efficiency of LED chips. Compared with the traditional mortar cutting method, diamond wire cutting has many significant advantages. Diamond wire cutting has a fast cutting speed and can greatly improve production efficiency. Diamond wire cutting does not use a large amount of chemicals, which is relatively more environmentally friendly and reduces pollution to the environment. With the continuous progress of material science and manufacturing technology, diamond wire has become an indispensable and important tool in the field of precision processing of hard and brittle materials.

[0004] In the existing technology, with the rise of solar photovoltaic industry and LED industry, the demand for large-scale and efficient cutting technology has promoted the rapid development of diamond wire technology. More researchers have begun to improve the manufacturing process and performance of diamond wire and have made great technological breakthroughs, such as the research and development of ultra-fine steel wire, the uniform distribution of diamond particles, and large-scale production technology. The profit growth brought by technological progress has brought new vitality to the diamond wire market. The research and development of diamond wire technology has entered a new stage, and new materials and processes have been continuously introduced. For example, through the introduction of nanotechnology, the diamond particles on the surface of the diamond wire can be finer and denser, and the cutting process is more delicate and smooth. The steel wire substrate has also been partially replaced by other lighter and stronger composite materials, further improving the cutting efficiency and durability of the diamond wire. At the same time, modern intelligent manufacturing technology has been applied to the production and use of diamond wire, which has significantly improved the production accuracy and consistency. In addition, the application of automatic control technology in cutting equipment has improved the stability and efficiency of the cutting process.

[0005] At present, although the research and development design of diamond wire has made significant progress, it still faces some limitations and challenges in practical applications. There is an inherent trade-off between the cutting efficiency of diamond wire and the life of the wire. Improving cutting efficiency usually means increasing the size and concentration of diamond particles, or increasing the cutting speed, but the above situation may also lead to increased wire wear and shorten the service life of the diamond wire; how to find a balance between high efficiency and long life is an important challenge in the research and development of diamond wire. Another decisive factor affecting the life of diamond wire is the performance of diamond wire. During the cutting process, strength enables diamond wire to better resist fatigue and wear, while plasticity can reduce early failures caused by stress concentration or design defects. Extending the service life can not only reduce production costs, but also reduce the maintenance and replacement frequency of equipment, and improve the operating efficiency of the production line; diamond wire faces some technical limitations in the coordinated optimization design of strength and plasticity, and has the necessity of continuous optimization. The base material of diamond wire is usually ultra-fine steel wire, which needs to have both high strength and high plasticity to withstand the huge stress and friction during high-speed cutting. However, the performance of materials always has its limits. Improving strength usually leads to a decrease in plasticity, and vice versa. Therefore, it is very difficult to find a material that achieves the best balance between strength and plasticity, and the trade-off in material properties limits the performance of diamond wire under extreme conditions.

[0006] The production process of diamond wire is to achieve a close combination of drawing and heat treatment in the automated production line. Among them, the drawing process parameters have a direct impact on the heat treatment process parameters. Specifically, the drawing temperature determines the heat treatment temperature, while the drawing speed and the production line length determine the heat treatment time. During the drawing process, since the material undergoes huge deformation, it is calculated that its axial strain is usually more than 9. Therefore, it is necessary to eliminate stress through annealing to facilitate subsequent further processing; the drawing process widely used by enterprises corresponds to austenitizing annealing. The drawing temperature is usually more than 1000℃ and the drawing time is 10 to 20 seconds. Under this process, after annealing, the material presents an equiaxed, non-deformed organizational structure. Although the test results show that it has a high strength of about 1400Mpa, the plasticity is insufficient, about 45%.

[0007] Machine learning is a branch of artificial intelligence that involves the use of statistical techniques and algorithms to enable computers to automatically learn and make decisions from data without explicit programming. Machine learning models identify patterns and relationships from large amounts of data and then use these patterns to predict or determine future behavior or outcomes. Machine learning can automatically process and analyze large amounts of data, greatly improving efficiency and reducing the need for human intervention. Through in-depth learning of data, it can often provide more accurate predictions and decisions than traditional methods. Machine learning models can also be updated and adjusted as data changes to maintain their effectiveness, but it is still difficult to effectively apply them in the materials field.

[0008] In summary, a diamond wire drawing heat treatment process is needed that enables the diamond wire to have both high strength and high plasticity - a performance data model and a high plasticity intelligent optimization method. Summary of the invention

[0009] A brief overview of the present invention is provided below in order to provide a basic understanding of certain aspects of the present invention. It should be understood that this overview is not an exhaustive overview of the present invention. It is not intended to identify key or important parts of the present invention, nor is it intended to limit the scope of the present invention. Its purpose is merely to present certain concepts in a simplified form as a prelude to a more detailed description discussed later.

[0010] In view of this, in order to solve the problem that the traditional drawing heat treatment process method in the prior art is difficult to achieve the coordinated optimization of high strength and high plasticity, the present invention provides a diamond wire drawing heat treatment process-performance data model and a high plasticity intelligent optimization method.

[0011] The technical solution is as follows: A diamond wire drawing heat treatment process-performance data model and strong plasticity intelligent optimization method, comprising the following steps:

[0012] S1. Conduct a drawing and recrystallization annealing experiment on the diamond wire, collect the drawing heat treatment process parameters and performance data in the experiment, and obtain a data set;

[0013] S2. Processing the data set to obtain a processed data set;

[0014] S3. Divide the processed data set, construct a drawing heat treatment process-performance machine learning model through Gaussian process regression algorithm, and use evaluation indicators to screen the model to obtain the final drawing heat treatment process-performance machine learning model;

[0015] S4. According to the final drawing heat treatment process-performance machine learning model, the reverse design goal is set to obtain a drawing process that meets different performance requirements.

[0016] Furthermore, in S1, the drawing heat treatment process parameters include drawing temperature, drawing time and cooling method, and the performance data of the drawing heat treatment process include tensile strength, yield strength, cross-sectional shrinkage rate and hardness.

[0017] Furthermore, the step S2 specifically includes the following steps:

[0018] S21. Preprocess the data set, including cleaning the data and processing missing values, outliers and duplicate values;

[0019] S22. Perform consistency check and clean inconsistent data based on the reasonable value range and correlation of each variable in the data set;

[0020] S23. Standardize all data in the data set by multivariate Gaussian distribution to obtain a processed data set;

[0021] The multivariate Gaussian distribution GP is expressed as:

[0022] f(x)~GP(m(x),k(x,x'))

[0023] Among them, the function f(x) represents the relationship between the input x, i.e., the drawing heat treatment process parameters, and the output y, i.e., the performance data of the drawing heat treatment process, m(x) is the mean function, k(x, x') is the kernel function, x is the first data point in the data set, and x' is the second data point in the data set.

[0024] Furthermore, the step S3 specifically includes the following steps:

[0025] S31. Use visualization methods to plot and analyze data points in a data set in data space;

[0026] S32. Divide the data set into two parts, wherein 85% of the data constitutes a training set for building a drawing heat treatment process-performance machine learning model, and 15% of the data constitutes a test set for testing the drawing heat treatment process-performance machine learning model;

[0027] S33. Based on the data set, select the Gaussian process regression algorithm, perform parameter design, and build a drawing heat treatment process-performance machine learning model;

[0028] In S33, the Gaussian process regression algorithm is specifically: any set of input data points {x, x, ..., x} on the function f(x) and the output function values ​​{y, y, ..., y} obey a multivariate Gaussian distribution;

[0029] S34. The regression effect of machine learning models of different algorithms is evaluated using correlation coefficient, mean square error and absolute relative error distribution evaluation indicators. After testing the test set, the machine learning model that meets the engineering error requirements is screened out, that is, the machine learning model with an error of less than 5% is used as the final drawing heat treatment process-performance machine learning model.

[0030] Furthermore, the step S4 specifically includes the following steps:

[0031] S41. Based on the final drawing heat treatment process-performance machine learning model, use the frog leaping algorithm to modify the algorithm parameters for reverse optimization, and the algorithm parameters include the maximum number of iterations, frog group size, and frog leaping step length;

[0032] S42. Set the target performance value and output it as the final result, use the adjusted parameters to initialize the frog group, and obtain the drawing process that meets different performance requirements.

[0033] The beneficial effects of the present invention are as follows: the present invention proposes a drawing process parameter design strategy, that is, a rapid recrystallization annealing process is used to replace the traditional austenitizing annealing, so as to retain the deformed structure and realize deformation strengthening, and the optimal process parameters are designed intelligently through a data model. The use of a rapid recrystallization annealing process instead of the traditional austenitizing annealing can improve the strength of the material while still maintaining its plasticity, thereby realizing a coordinated optimization design of strength and plasticity;

[0034] refer to Figure 2 and Figure 3In the process of constructing the database, the present invention conducts actual production tests on a series of drawing recrystallization annealing temperatures and drawing times. The drawing process parameters for tensile strength exceeding 1400Mpa are drawing at 520-620℃ for 30s, drawing at 520-660℃ for 20s, and drawing at 520-720℃ for 10s. The drawing process parameters for cross-sectional shrinkage exceeding 45% are drawing at 520-860℃ for 30s, drawing at 520-860℃ for 20s, drawing at 580-660℃ for 10s, and drawing at 740-860℃ for 10s. Compared with the performance after austenitizing annealing, the low temperature range and short time range of most recrystallization annealing can meet the performance requirements. The test results show that there is an opportunity to achieve the goal through the design of drawing process parameters, and the temperature is controlled below 1000℃, the time is controlled at about 10s, and the drawing temperature corresponding to the recrystallization annealing is much lower than that of the austenitizing annealing, which greatly reduces the energy consumption.

[0035] The process drawing time described in the present invention is shorter, and the length of the production line is correspondingly shortened, thereby saving production resources and reducing production costs; the present invention utilizes computer simulation and artificial intelligence technology, especially machine learning technology, to construct a data model, and uses the data model established by the Gaussian process regression algorithm, as well as specially designed model parameters, and uses the frog leaping algorithm to perform intelligent reverse drawing process parameter optimization, thereby realizing the intelligent optimization of the diamond wire drawing production process. The constructed data model intelligent optimization design method has good portability, and can quickly design heat treatment process parameters that meet the target performance requirements. Multi-objective optimization can not only achieve dual performance goals, but also can be expanded to three, four or even five goals, and the goals can be randomly combined, fully demonstrating the potential and advantages of multi-objective optimization, and providing strong technical support for R&D personnel to accurately design the drawing temperature and time, and obtain the relationship between the drawing process parameters and the mechanical properties of the diamond wire; the present invention can be applied to automated production line equipment to realize intelligent drawing heat treatment. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0037] Figure 1 It is a schematic diagram of the process of heat treatment process-performance data model and strength-plasticity intelligent optimization method for diamond wire drawing;

[0038] Figure 2 It is a broken line diagram of the section shrinkage rate;

[0039] Figure 3 It is a broken line diagram of tensile strength. DETAILED DESCRIPTION

[0040] In order to make the technical solutions and advantages of the embodiments of the present invention more clearly understood, the exemplary embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than an exhaustive list of all the embodiments. It should be noted that the embodiments of the present invention and the features in the embodiments can be combined with each other without conflict.

[0041] Example 1: Reference Figure 1-Figure 3 The present embodiment is described in detail, a diamond wire drawing heat treatment process-performance data model and a strength-plasticity intelligent optimization method, which specifically includes the following steps:

[0042] S1. Conduct a drawing and recrystallization annealing experiment on the diamond wire, collect the drawing heat treatment process parameters and performance data in the experiment, and obtain a data set;

[0043] S2. Processing the data set to obtain a processed data set;

[0044] S3. Divide the processed data set, construct a drawing heat treatment process-performance machine learning model through Gaussian process regression algorithm, and use evaluation indicators to screen the model to obtain the final drawing heat treatment process-performance machine learning model;

[0045] S4. According to the final drawing heat treatment process-performance machine learning model, the reverse design goal is set to obtain a drawing process that meets different performance requirements.

[0046] Furthermore, in S1, the drawing heat treatment process parameters include drawing temperature, drawing time and cooling method, etc., and the performance data of the drawing heat treatment process include tensile strength, yield strength, cross-sectional shrinkage rate and hardness, etc.

[0047] Specifically, in this embodiment, the diameter of the diamond wire is 0.46 mm. A 0.46 mm diamond wire that has been drawn four times using a high carbon steel busbar is subjected to a series of drawing recrystallization annealing experiments at different drawing temperatures and drawing times. After drawing, the structure is observed and the performance is tested to complete data collection.

[0048] The high carbon steel busbar used has the following chemical components in mass percentage in addition to Fe: 92% C, 17% Si, 32% Mn, 11% P, 7% S;

[0049] The high carbon steel busbar is drawn to obtain a wire rod, the wire rod is roughly drawn, and the diameter of the diamond wire after drawing is 2.30 mm, and austenitizing heat treatment is performed at the same time, and the diamond wire is intermediately drawn, and the diameter of the diamond wire after drawing is 0.73 mm, and austenitizing heat treatment is performed at the same time, and the diamond wire is finely drawn, and the diameter of the diamond wire after drawing is 0.46 mm;

[0050] A series of drawing recrystallization annealing experiments with drawing temperature and drawing time were conducted on 0.46 mm diamond wire. The drawing temperature range was 520-860 ° C, the temperature gradient was 20 ° C, the drawing time was 10s, 20s, and 30s, respectively. Air cooling was used to cool to room temperature, and then the microstructure was observed and the performance was tested.

[0051] In the data set of this embodiment, the drawing heat treatment process parameters include the drawing temperature and the drawing time, and the performance data of the drawing heat treatment process include the room temperature tensile strength and the cross-sectional shrinkage rate.

[0052] Furthermore, the step S2 specifically includes the following steps:

[0053] S21. Preprocess the data set, including cleaning the data and processing missing values, outliers and duplicate values;

[0054] S22. Perform consistency check and clean inconsistent data based on the reasonable value range and correlation of each variable in the data set;

[0055] S23. Standardize all data in the data set by multivariate Gaussian distribution to obtain a processed data set, thereby eliminating errors caused by differences in data magnitude;

[0056] The multivariate Gaussian distribution GP is expressed as:

[0057] f(x)~GP(m(x),k(x,x'))

[0058] Among them, the function f(x) represents the relationship between the input x, i.e., the drawing heat treatment process parameters, and the output y, i.e., the performance data of the drawing heat treatment process (mechanical properties in this embodiment), m(x) is the mean function, which is usually set to 0, k(x, x') is the kernel function, which describes the similarity between x and x', x is the first data point in the data set, and x' is the second data point in the data set.

[0059] Furthermore, the step S3 specifically includes the following steps:

[0060] S31. Use visualization methods such as scatter plots, line graphs, heat maps or 3D charts to plot and analyze the data points in the data set in the data space, and the trend of experimental data changes is consistent with the basic principles of materials science;

[0061] S32. Divide the data set into two parts, wherein 85% of the data constitutes a training set for building a drawing heat treatment process-performance machine learning model, and 15% of the data constitutes a test set for testing the drawing heat treatment process-performance machine learning model;

[0062] S33. Based on the data set, the Gaussian Progress Regress (GPR) algorithm is selected to build a drawing heat treatment process-performance machine learning model, and the regression effect of the model is improved through parameter design, the error is reduced and the prediction accuracy is guaranteed;

[0063] In S33, the Gaussian process regression algorithm is specifically as follows: any set of input data points {x, x, …, x} on the function f(x) and the output function values ​​{y, y, …, y} all obey a multivariate Gaussian distribution;

[0064] S34. The regression effects of machine learning models of different algorithms were evaluated using evaluation indicators such as correlation coefficient (R), mean squared error (MSE) and absolute relative error distribution (ARE). The machine learning models that met the engineering error requirements were screened out through test set testing, that is, the machine learning models with an error of less than 5% were used as the drawing heat treatment process-performance machine learning models;

[0065] Specifically, a Gaussian process regression model, i.e., a drawing heat treatment process-performance machine learning model, was created using the training set to obtain a drawing process-tensile strength model and a drawing process-section shrinkage model, respectively;

[0066] When performing special parameter design, using the Crossval function to cross-validate the two models is helpful to evaluate the generalization ability of the model and avoid overfitting. Cross-validation automatically divides the training set into k subsets of equal size, and uses each subset as a secondary test set and the remaining set as a secondary training set. After training the fixed model, the mean square error is calculated. The above operation is performed on each subset in turn. Finally, the error is averaged through the average formula and used as the loss of the drawing process-tensile strength model and the drawing process-section shrinkage model, that is, the mean square error result of the drawing heat treatment process-performance machine learning model;

[0067] The average formula is expressed as:

[0068]

[0069] Among them, k is the number of subsets, L i is the error for each subset, L CV is the model loss;

[0070] Use the Standardize function to standardize the training set, calculate its mean μ and variance σ', and use the standardization formula to process each data point to adjust the data to the standard range;

[0071] Data x' within the standard range i It is expressed as:

[0072]

[0073] Among them, x i are data points belonging to the training set;

[0074] Specify the Basis Function of the drawing heat treatment process-performance machine learning model as constant, which means that the model assumes that the mean function is a constant and does not change with the input data. Specify the Kernel Function of the drawing heat treatment process-performance machine learning model as exponential, which is used to define the similarity between data points.

[0075] The kernel function k(x,x') is expressed as:

[0076]

[0077] Among them, ||x-x'|| is the Euclidean distance between two data points, σ ​​is the scale parameter of the kernel function, and exp represents the exponential function;

[0078] Refer to Table 1, the correlation coefficient, mean square error, and maximum absolute relative error (MARE) in the absolute relative error distribution obtained from the training of the drawing heat treatment process-performance machine learning model;

[0079] Table 1 Drawing heat treatment process-performance machine learning model training results

[0080]

[0081] The test set was used for test verification, and the overall error was 3.6%, which met the engineering error requirements.

[0082] Furthermore, the step S4 specifically includes the following steps:

[0083] S41. Based on the drawing heat treatment process-performance machine learning model, the frog leaping algorithm is used to modify the algorithm parameters for reverse optimization. The algorithm parameters include the maximum number of iterations, frog group size and frog leaping step length. Different data sets and data models have different corresponding optimal parameters.

[0084] S42. Set the target performance value (single or multiple) and output it as the final result, use the adjusted parameters to initialize the frog group, and obtain the drawing process that meets different performance requirements.

[0085] Specifically, using the constructed drawing heat treatment process-performance machine learning model, the reverse design targets are the tensile strength target and the cross-sectional shrinkage rate target. According to the frog leaping optimization algorithm, the balance between strength and plasticity is achieved based on the multi-performance target design. The frog leaping algorithm parameters are selected through gradient testing, the maximum number of iterations maxIter is 1000, the frog population size populationSize is 50, and the frog leap step frogLeap is 0.01. The target performance values ​​are set to tensile strength greater than 1400MPa and cross-sectional shrinkage greater than 45%. Reverse optimization is performed. Referring to Table 2, the drawing process parameters that meet the conditions are obtained.

[0086] Table 2 Drawing process parameters that meet the conditions

[0087]

[0088] Example 2: Reference Figure 1-Figure 3 This embodiment is described in detail, a diamond wire drawing heat treatment process-performance data model and strong plasticity intelligent optimization method. The difference between this embodiment and embodiment 1 is that the target of the reverse design in step S4 is a single target of tensile strength;

[0089] Using the constructed drawing process-tensile strength model, the drawing recrystallization process parameters corresponding to different tensile strength requirements are reversely optimized according to the frog leaping algorithm. The frog leaping algorithm parameters are selected through gradient testing, the maximum number of iterations maxIter is 1000, the frog population size populationSize is 30, the frog leap step frogLeap is 0.05, and the target performance value is set to a tensile strength greater than 1500MPa. Referring to Table 3, 5 process parameters are reversely optimized;

[0090] Table 3 Process parameters for reverse optimization of target tensile strength 1500 MPa

[0091]

[0092] Example 3: Reference Figure 1-Figure 3 This embodiment is described in detail, a diamond wire drawing heat treatment process-performance data model and strong plasticity intelligent optimization method. The difference between this embodiment and embodiment 1 is that the target of the reverse design in step S4 is a single target of cross-sectional shrinkage rate;

[0093] Using the constructed drawing process-section shrinkage model, the drawing recrystallization process parameters corresponding to different section shrinkage requirements are reversely optimized according to the frog leaping algorithm. The frog leaping algorithm parameters are selected through gradient testing. The maximum number of iterations maxIter is 1000, the frog population size populationSize is 30, and the frog leap step frogLeap is 0.05. The target performance value is set to a section shrinkage greater than 55%. Refer to Table 4 for the five process parameters obtained by reverse optimization.

[0094] Table 4 Process parameters for reverse optimization of target section shrinkage of 55%

[0095]

[0096] Although the present invention has been described according to a limited number of embodiments, it will be apparent to those skilled in the art, with the benefit of the above description, that other embodiments may be envisioned within the scope of the invention thus described. In addition, it should be noted that the language used in this specification is selected primarily for readability and teaching purposes, rather than for explaining or defining the subject matter of the present invention. Therefore, many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the appended claims. The disclosure of the present invention is illustrative, not restrictive, with respect to the scope of the present invention, which is defined by the appended claims.

Claims

1. A diamond wire drawing heat treatment process-performance data model and strong plasticity intelligent optimization method, characterized in that: The following steps are involved: S1. Conduct a drawing and recrystallization annealing experiment on the diamond wire, collect the drawing heat treatment process parameters and performance data in the experiment, and obtain a data set; S2. Processing the data set to obtain a processed data set; S3. Divide the processed data set, construct a drawing heat treatment process-performance machine learning model through Gaussian process regression algorithm, and use evaluation indicators to screen the model to obtain the final drawing heat treatment process-performance machine learning model; S4. According to the final drawing heat treatment process-performance machine learning model, the reverse design goal is set to obtain a drawing process that meets different performance requirements.

2. According to claim 1, a diamond wire drawing heat treatment process-performance data model and strength-plasticity intelligent optimization method is characterized in that: In S1, the drawing heat treatment process parameters include drawing temperature, drawing time and cooling method, and the performance data of the drawing heat treatment process include tensile strength, yield strength, cross-sectional shrinkage and hardness.

3. The diamond wire drawing heat treatment process-performance data model and strength-plasticity intelligent optimization method according to claim 2, characterized in that: The S2 specifically includes the following steps: S21. Preprocess the data set, including cleaning the data and processing missing values, outliers and duplicate values; S22. Perform consistency check and clean inconsistent data based on the reasonable value range and correlation of each variable in the data set; S23. Standardize all data in the data set by multivariate Gaussian distribution to obtain a processed data set; The multivariate Gaussian distribution GP is expressed as: f(x)~GP(m(x),k(x,x')) Among them, the function f(x) represents the relationship between the input x, i.e., the drawing heat treatment process parameters, and the output y, i.e., the performance data of the drawing heat treatment process, m(x) is the mean function, k(x, x') is the kernel function, x is the first data point in the data set, and x' is the second data point in the data set.

4. The diamond wire drawing heat treatment process-performance data model and strength-plasticity intelligent optimization method according to claim 3, characterized in that: The S3 specifically includes the following steps: S31. Use visualization methods to plot and analyze data points in a data set in data space; S32. Divide the data set into two parts, wherein 85% of the data constitutes a training set for building a drawing heat treatment process-performance machine learning model, and 15% of the data constitutes a test set for testing the drawing heat treatment process-performance machine learning model; S33. Based on the data set, select the Gaussian process regression algorithm, perform parameter design, and build a drawing heat treatment process-performance machine learning model; In S33, the Gaussian process regression algorithm is specifically: any set of input data points {x, x, ..., x} on the function f(x) and the output function values ​​{y, y, ..., y} obey a multivariate Gaussian distribution; S34. The regression effect of machine learning models of different algorithms is evaluated using correlation coefficient, mean square error and absolute relative error distribution evaluation indicators. After testing the test set, the machine learning model that meets the engineering error requirements is screened out, that is, the machine learning model with an error of less than 5% is used as the final drawing heat treatment process-performance machine learning model.

5. The diamond wire drawing heat treatment process-performance data model and strength-plasticity intelligent optimization method according to claim 4, characterized in that: The S4 specifically includes the following steps: S41. Based on the final drawing heat treatment process-performance machine learning model, use the frog leaping algorithm to modify the algorithm parameters for reverse optimization, and the algorithm parameters include the maximum number of iterations, frog group size, and frog leaping step length; S42. Set the target performance value and output it as the final result, use the adjusted parameters to initialize the frog group, and obtain the drawing process that meets different performance requirements.

Images