Model training method and device, yield strength prediction method and device and electronic equipment

Abstract

The invention discloses a model training method and device, a yield strength prediction method and device and electronic equipment, and the method comprises the steps: obtaining a plurality of training samples which comprise a first environment temperature, a first stress, and a first resistivity of a first metal sample at a target sampling moment; the target sampling time is the sampling time of the first metal sample under the first environment temperature and the first stress, and at least one of the first environment temperature, the first stress and the target sampling time of any two training samples is different; the training sample is adopted to train a neural network model to obtain a target model, the target model is used for predicting the yield strength of the to-be-tested metal sample, the target model is adopted to predict the yield strength of the to-be-tested metal sample, nondestructive evaluation can be carried out on the metal material, and universality is achieved.

Description

Technical Field

[0001] This application belongs to the technical field of metal detection, and particularly relates to a model training method, a yield strength prediction method, a device and an electronic device. Background Art

[0002] Under extreme service environments, the monitoring of the performance degradation of metal materials is of crucial significance. Especially under complex environmental conditions such as high temperature and irradiation, the mechanical properties of metal materials often change significantly, affecting the safety and reliability of structures.

[0003] Traditional mechanical property detection methods usually rely on destructive experiments. For example, the mechanical properties of metal materials are evaluated by measuring the yield strength, where the yield strength refers to the yield limit when the metal material exhibits a yield phenomenon, and the yield strength detection of the metal is obtained through a tensile test.

[0004] The above method will cause irreversible damage to the metal material and cannot be applied to the mechanical property evaluation of in-service metal materials, lacking universality. Summary of the Invention

[0005] Embodiments of this application provide a model training method, a yield strength prediction method, a device and an electronic device. The yield strength of a metal sample to be measured is predicted using a target model. During the evaluation of the mechanical properties of the metal material, it is not necessary to damage the metal material, enabling non-destructive evaluation of the metal material and having universality.

[0006] In a first aspect, embodiments of this application provide a model training method, the method comprising:

[0007] Obtain a plurality of training samples, the training samples including a first environmental temperature, a first stress, a first resistivity of a first metal sample at a target sampling moment, and a reference yield strength of the first metal sample at the target sampling moment, where the target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress, and at least one of the first environmental temperature, the first stress, and the target sampling moment is different between any two of the training samples;

[0008] Use the training samples to train a neural network model to obtain a target model, the target model being used to predict the yield strength of a metal sample to be measured.

[0009] In a second aspect, embodiments of this application provide a yield strength prediction method, comprising:

[0010] Obtain the resistivity of a metal sample to be measured at a second environmental temperature, a second stress, and a first sampling moment;

[0011] Input the resistivity of the metal sample to be measured into the target model to obtain the target yield strength of the metal sample, where the target model is obtained based on the model training method described in the first aspect.

[0012] In a third aspect, an embodiment of the present application provides a model training device, which includes:

[0013] A first acquisition module, configured to acquire a plurality of training samples, where the training samples include a first environmental temperature, a first stress, a first resistivity of a first metal sample at a target sampling moment, and a reference yield strength of the first metal sample at the target sampling moment, the target sampling moment being the sampling moment of the first metal sample under the first environmental temperature and the first stress, and at least one of the first environmental temperature, the first stress, and the target sampling moment is different for any two of the training samples;

[0014] A training module, configured to use the training samples to train a neural network model to obtain a target model, where the target model is used to predict the yield strength of a metal sample to be measured.

[0015] In a fourth aspect, an embodiment of the present application provides a yield strength prediction device, which includes:

[0016] A second acquisition module, configured to acquire the resistivity of a metal sample to be measured at a second environmental temperature, a second stress, and a first sampling moment;

[0017] A prediction module, configured to input the resistivity of the metal sample to be measured into the target model to obtain the target yield strength of the metal sample, where the target model is obtained based on the model training method described in the first aspect.

[0018] In a fourth aspect, an embodiment of the present application provides an electronic device, including: a processor and a memory storing computer program instructions;

[0019] When the processor executes the computer program instructions, the model training method described in the first aspect or the yield strength prediction method described in the second aspect is implemented.

[0020] In a fifth aspect, an embodiment of the present application provides a computer-readable storage medium, on which computer program instructions are stored, and when the computer program instructions are executed by a processor, the model training method described in the first aspect or the yield strength prediction method described in the second aspect is implemented.

[0021] In a sixth aspect, an embodiment of the present application provides a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device is caused to execute the model training method as described in the first aspect or the yield strength prediction method as described in the second aspect.

[0022] In the model training method, yield strength prediction method, device, and electronic device according to the embodiments of the present application, by obtaining a plurality of training samples, where the training samples include the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield strength of the first metal sample at the target sampling moment, the target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress, and at least one of the first environmental temperature, the first stress, and the target sampling moment of any two training samples is different. The neural network model is trained using the training samples to obtain a target model, and the target model is used to predict the yield strength of the metal sample to be measured. Subsequently, during the evaluation of the mechanical properties of the metal material, it is not necessary to damage the metal material, and non-destructive evaluation of the metal material can be performed, which has universality. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] To more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required to be used in the embodiments of the present application. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings without creative efforts.

[0024] Figure 1 is a flowchart of the model training method provided by an embodiment of the present application;

[0025] Figure 2a is a schematic diagram of the area ratio of the precipitated phase provided by an embodiment of the present application;

[0026] Figure 2b is a schematic diagram of the average area of the precipitated phase provided by an embodiment of the present application;

[0027] Figure 2c is a schematic diagram of the average equivalent circle diameter of the grains provided by an embodiment of the present application;

[0028] Figure 2d is a schematic diagram of the dislocation density provided by an embodiment of the present application;

[0029] Figure 3 is a schematic diagram of the neural network model provided by an embodiment of the present application;

[0030] Figure 4 is a flowchart of the yield strength prediction method provided by an embodiment of the present application;

[0031] Figure 5Another schematic flowchart of the yield strength prediction method provided by the embodiments of the present application;

[0032] Figure 6 A schematic diagram of the creep situation of the sample provided by the embodiments of the present application;

[0033] Figure 7 A schematic diagram of resistivity comparison provided by the embodiments of the present application;

[0034] Figure 8 A schematic diagram of yield strength comparison provided by the embodiments of the present application;

[0035] Figure 9 A schematic diagram of the comparison between the experimental value and the predicted value of the yield strength provided by the embodiments of the present application;

[0036] Figure 10 A schematic diagram of the structure of the model training device provided by the embodiments of the present application;

[0037] Figure 11 A schematic diagram of the structure of the yield strength prediction device provided by the embodiments of the present application;

[0038] Figure 12 A schematic diagram of the structure of the electronic device provided by the embodiments of the present application. Detailed implementation manners

[0039] The features and exemplary embodiments of various aspects of the present application will be described in detail below. To make the objectives, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without some of these specific details. The following description of the embodiments is only intended to provide a better understanding of the present application by showing examples of the present application.

[0040] It should be noted that, in this document, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variation thereof is intended to cover a non-exclusive inclusion, such that a process, method, article or device comprising a series of elements includes not only those elements, but also other elements not expressly listed, or elements inherent to such process, method, article or device. Without further limitation, elements defined by the statement "comprising..." do not preclude the existence of additional identical elements in the process, method, article or device comprising the said elements.

[0041] To solve the problems of the prior art, the embodiments of the present application provide a model training method, a yield strength prediction method, an apparatus, and an electronic device. First, the model training method provided by the embodiments of the present application will be introduced below.

[0042] Figure 1 FIG. 1 shows a schematic flow chart of the model training method provided by the embodiments of the present application. As Figure 1 shown, the model training method provided by the embodiments of the present application is applied to an electronic device and includes the following steps 101-102, where:

[0043] Step 101, obtain a plurality of training samples, where the training samples include the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield strength of the first metal sample at the target sampling moment. The target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress. At least one of the first environmental temperature, the first stress, and the target sampling moment of any two of the training samples is different.

[0044] In this embodiment, a plurality of training samples are obtained. The training samples are used for model training and include the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield strength of the first metal sample at the target sampling moment. Among them, the yield strength refers to the stress value at which a metal material begins to produce irreversible plastic deformation when subjected to an external force, and the yield strength is an important index of the mechanical properties of materials.

[0045] The above target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress. At least one of the first environmental temperature, the first stress, and the target sampling moment of any two training samples is different. For example, the environmental temperatures of two training samples are the same, the stresses are the same, and the sampling moments are different.

[0046] Step 102, use the training samples to train a neural network model to obtain a target model, where the target model is used to predict the yield strength of a metal sample to be measured.

[0047] In this embodiment, the neural network model is trained using the training samples. The neural network model is composed of multiple fully connected layers. The target model is obtained through training, and the target model is used to predict the yield strength of the metal to be measured.

[0048] In this embodiment, a plurality of training samples are obtained. The training samples include the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield strength of the first metal sample at the target sampling moment. The target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress. At least one of the first environmental temperature, the first stress, and the target sampling moment of any two training samples is different. The neural network model is trained using the training samples to obtain a target model. The target model is used to predict the yield strength of the metal sample to be measured. Subsequently, during the evaluation of the mechanical properties of the metal material, it is not necessary to damage the metal material, and the metal material can be evaluated nondestructively, which has universality.

[0049] In an embodiment of the present application, in step 101, the obtaining of a plurality of training samples includes:

[0050] Obtain a plurality of first microstructure information corresponding to the first metal sample. The first microstructure information includes the first average equivalent circle diameter of grains, the first average area of precipitates, the first area ratio of precipitates, and the first dislocation density obtained from the first metal sample under the first environmental temperature, the first stress, and the sampling moment. At least one of the first environmental temperature, the first stress, and the sampling moment of any two pieces of the first microstructure information is different;

[0051] Noise addition processing is respectively performed on the first average equivalent circle diameter of grains, the first average area of precipitates, the first area ratio of precipitates, and the first dislocation density in each piece of the first microstructure information to obtain the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density; the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density are used as the second microstructure information;

[0052] According to the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density, calculate the resistivity corresponding to each piece of the second microstructure information of the first metal sample;

[0053] According to the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density, calculate the yield strength corresponding to each piece of the second microstructure information of the first metal sample;

[0054] Multiple training samples are obtained based on the resistivity corresponding to each of the second microstructure information of the first metal sample, the corresponding yield strength, the experimental resistivity corresponding to each of the first microstructure information of the first metal sample, and the corresponding experimental yield strength, where the experimental resistivity and the experimental yield strength are obtained through experimental measurement.

[0055] In the above step, multiple first microstructure information corresponding to the first metal sample is obtained. The first microstructure information includes: the first average equivalent circle diameter of grains, the first average area of precipitates, the proportion of the precipitate area, and the first dislocation density obtained from the first metal sample at the first environmental temperature, the first stress, and the sampling time. At least one of the first environmental temperature, the first stress, and the sampling time of any two pieces of first microstructure information is different. For example, the environmental temperature is the same, the stress is the same, but the sampling time is different, or the environmental temperature is different, the stress is the same, and the sampling time is the same.

[0056] Optionally, the electron backscatter diffraction technique (EBSD) is used to obtain the average equivalent circle diameter of grains. EBSD is a microstructure analysis tool that can provide various microstructure information including grain size. The scanning electron microscope (SEM) is used to obtain the average area of precipitates and the proportion of the precipitate area. The X-ray diffraction (XRD) is used to obtain the dislocation density. XRD is a non-destructive testing method.

[0057] Furthermore, according to the evolution law of the above multi-scale microstructure information, oscillating noise is added to generate a large amount of microstructure information that conforms to this law. Specifically, noise addition processing is performed on the first average equivalent circle diameter of grains, the first average area of precipitates, the proportion of the precipitate area, and the first dislocation density in each piece of first microstructure information to obtain the second average equivalent circle diameter of grains, the second average area of precipitates, the proportion of the second precipitate area, and the second dislocation density. The second average equivalent circle diameter of grains, the second average area of precipitates, the proportion of the second precipitate area, and the second dislocation density are used as the second microstructure information.

[0058] Based on the second average equivalent circle diameter of grains, the second average area of precipitates, the proportion of the second precipitate area, and the second dislocation density, the resistivity corresponding to each second microstructure information in the first metal sample is calculated; based on the second average equivalent circle diameter of grains, the second average area of precipitates, the proportion of the second precipitate area, and the second dislocation density, the yield strength corresponding to each second microstructure information of the first metal sample is calculated.

[0059] Obtain the experimental resistivity and the corresponding experimental yield strength corresponding to each first microstructure information of the first metal sample. The experimental resistivity and the experimental yield strength are obtained through experimental measurements. Optionally, the experimental resistivity is measured by the probe method, and the experimental yield strength is measured by a room temperature tensile test.

[0060] The training samples can be divided into first training samples and second training samples. The experimental resistivity corresponding to each first microstructure information of the first metal sample is used as the first resistivity, the environmental temperature corresponding to the experimental resistivity is used as the first environmental temperature, the stress corresponding to the experimental resistivity is used as the first stress, the sampling moment corresponding to the experimental resistivity is used as the target sampling moment, and the experimental yield strength corresponding to the first environmental microstructure information is used as the reference yield strength. According to the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield resistance corresponding to the first microstructure information, a first training sample is constructed.

[0061] Second training sample: The resistivity corresponding to each second microstructure information of the first metal sample is used as the first resistivity, the environmental temperature corresponding to the resistivity is used as the first environmental temperature, the stress corresponding to the resistivity is used as the first stress, the sampling moment corresponding to the resistivity is used as the target sampling moment, and the yield strength corresponding to the first environmental microstructure information is used as the reference yield strength. According to the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield resistance corresponding to the second microstructure information, a second training sample is constructed.

[0062] The first resistivity and the reference yield resistance in the first training sample are obtained through experiments, while the first resistivity and the reference yield resistance in the second training sample are calculated from data with added noise. Compared with the second training sample, the data accuracy of the first training sample is higher than that of the second training sample. Larger data noise may mask the true features. By adding noise to the training samples, the model can learn the essential features of the data. When the model encounters real data containing noise in actual applications, it can better identify and process it, improving its adaptability in different situations. The microstructure largely determines the physical and mechanical properties of metal materials. Adding noise based on microstructure information can enrich the training data.

[0063] Optionally, the first grain average equivalent circular diameter, the first precipitated phase average area, the first precipitated phase area ratio, and the first dislocation density in each of the first microstructure information are respectively subjected to noise addition processing to obtain the second grain average equivalent circular diameter, the second precipitated phase average area, the second precipitated phase area ratio, and the second dislocation density; the second grain average equivalent circular diameter, the second precipitated phase average area, the second precipitated phase area ratio, and the second dislocation density are used as the second microstructure information, including:

[0064] See also Figure 2a , for the first precipitated phase area ratio in multiple first microscopic information, the precipitated phase is composed of M on the grain boundary 22 C6 and M inside the grain 22 C6 composition, Figure 2a M 22 The experimental data of C6 is used as an example to illustrate that the M 22 The experimental data of C6 has 5 basic data points. The addition of oscillation noise according to the evolution law mentioned above is actually an interpolation process of the above 5 basic data points to obtain the M 22 The first curve corresponding to the experimental data of C6, the data points on the first curve are called first data points, multiple first data points are selected from the first curve, noise is added on the basis of the multiple first data points, and the second data points corresponding to the multiple first data points are obtained, that is, Figure 2a M 22 C6 generation data; Figure 2a The M inside the grain 22 The experimental data of C6 is used as an example to illustrate that the M 22 The generated data of C6 also has 5 basic data points. These 5 basic data points are interpolated to obtain the M value inside the grain. 22 The first curve corresponding to the experimental data of C6, multiple first data points are selected from the first curve, and noise is added on the basis of the multiple first data points to obtain the second data points corresponding to the multiple first data points, that is, Figure 2a The M inside the grain 22 The generated data of C6 obtains the area ratio of the second precipitated phase according to the coordinate information of the above second data point.

[0065] See also Figure 2b , for the average area of the first precipitated phase in the plurality of first microscopic information, the average area of the first precipitated phase is determined by M on the grain boundary. 22 C6 and M inside the grain 22 C6 composition, Figure 2b M 22 The experimental data of C6 is used as an example to illustrate that the M 22The experimental data of C6 has 5 basic data points. Interpolation is performed on the 5 basic data points to obtain M on the grain boundary 22 The first curve corresponding to the experimental data of C6. The data points on the first curve are called the first data points. Multiple first data points are selected from the first curve, and noise addition processing is performed on the basis of the multiple first data points to obtain the second data points corresponding to the respective multiple first data points, that is Figure 2b M on the grain boundary in 22 The generated data of C6; taking Figure 2b M in the grain interior in 22 The experimental data of C6 as an example to illustrate, M in the grain interior 22 The generated data of C6 also has 5 basic data points. Interpolation is performed on the 5 basic data points to obtain M in the grain interior 22 The first curve corresponding to the experimental data of C6. Multiple first data points are selected from the first curve, and noise addition processing is performed on the basis of the multiple first data points to obtain the second data points corresponding to the respective multiple first data points, that is Figure 2b M in the grain interior in 22 The generated data of C6. The average area of the second precipitate phase is obtained according to the coordinate information of the above-mentioned second data points.

[0066] See Figure 2c For the average equivalent circle diameter of the first grain in multiple first microscopic information Figure 2c The experimental data in has 5 basic data points. Interpolation is performed on the 5 basic data points to obtain the first curve corresponding to the experimental data. The data points on the first curve are called the first data points. Multiple first data points are selected from the first curve, and noise addition processing is performed on the basis of the multiple first data points to obtain the second data points corresponding to the respective multiple first data points, that is Figure 2c The generated data in. The average equivalent circle diameter of the second grain is obtained according to the coordinate information of the above-mentioned second data points.

[0067] See Figure 2d For the first dislocation density in multiple first microscopic information Figure 2d The experimental data in has 5 basic data points. Interpolation is performed on the 5 basic data points to obtain the first curve corresponding to the experimental data. The data points on the first curve are called the first data points. Multiple first data points are selected from the first curve, and noise addition processing is performed on the basis of the multiple first data points to obtain the second data points corresponding to the respective multiple first data points, that is Figure 2d The generated data in. The second dislocation density is obtained according to the coordinate information of the above-mentioned second data points.

[0068] Take the proportion of the area of the second precipitation phase, the average area of the second precipitation phase, the average equivalent circle diameter of the second grain, and the second dislocation density at the same moment obtained by the above method as the second microstructure information. It should be noted that the number of data points is not limited to the above number.

[0069] In an embodiment of the present application, calculating the resistivity corresponding to each of the second microstructure information of the first metal sample according to the average equivalent circle diameter of the second grain, the average area of the second precipitation phase, the proportion of the area of the second precipitation phase, and the second dislocation density includes:

[0070] Calculate the contribution values of dislocations, grain boundaries, and precipitation phases to the resistivity respectively according to the second dislocation density, the average equivalent circle diameter of the second grain, and the proportion of the area of the second precipitation phase in the second microstructure information;

[0071] Add the intrinsic resistivity of the first metal sample, the contribution values of the dislocations, the grain boundaries, and the precipitation phases to the resistivity to obtain the resistivity corresponding to each second microstructure information of the first metal sample.

[0072] In the above steps, the contribution value of dislocations to the resistivity is calculated according to the second dislocation density in the second microstructure information, the contribution value of grain boundaries to the resistivity is calculated according to the average equivalent circle diameter of the second grain, and the contribution value of precipitation phases to the resistivity is calculated according to the proportion of the area of the second precipitation phase.

[0073] Specifically, obtain the intrinsic resistivity of the first metal sample, add the intrinsic resistivity of the first metal sample, the contribution value of dislocations to the resistivity, the contribution value of grain boundaries to the resistivity, and the contribution value of precipitation phases to the resistivity to obtain the resistivity corresponding to each second microstructure information of the first metal sample. The resistivity is specifically expressed as:

[0074] ρ total = ρ0 + ρ D + ρ GB + ρ P (1)

[0075] Where, P total is the resistivity corresponding to each second microstructure information of the first metal sample, p O is the intrinsic resistivity, p D is the contribution value of dislocations to the resistivity, p GB is the contribution value of grain boundaries to the resistivity, p P is the contribution value of precipitation phases to the resistivity.

[0076] The intrinsic resistivity refers to the resistivity of the first metal sample itself due to lattice vibration (phonon scattering) in the absence of impurities and defects. For example, for austenitic alloys at room temperature, the intrinsic resistivity ρ0=0.311 μΩm.

[0077] By calculating the microstructure information with added noise and obtaining the resistivity, the training data can be enriched and the problem of too few training samples can be solved.

[0078] In one embodiment of the present application, the contribution of the dislocation to the resistivity is calculated in the following manner:

[0079] Taking a square root of the second dislocation density to obtain a first value; multiplying the shear modulus, the Burgers vector and the first value to obtain a second value; calculating the square of the second value to obtain a third value; and multiplying the third value by a preset first constant to obtain a contribution value of the second dislocation density to the resistivity;

[0080] and / or;

[0081] The contribution of the grain boundary to the resistivity is calculated as follows:

[0082] Dividing a preset second constant by the second grain average equivalent circular diameter to obtain a contribution value of the second grain average equivalent circular diameter to the resistivity;

[0083] and / or;

[0084] The contribution of the precipitation relative resistivity is calculated as follows:

[0085]

[0086] Among them, ρ P is the contribution value of the precipitate phase to the resistivity, P is a preset third constant, f is the area ratio of the second precipitate phase, r is the average radius of the precipitate phase, and r is calculated from the average area of the second precipitate phase.

[0087] In this embodiment, the square root of the second dislocation density is taken to obtain a first value, the shear modulus, the Burgers vector and the first value are multiplied to obtain a second value, the square of the second value is calculated to obtain a third value, and the third value is multiplied by a preset first constant to obtain a contribution value of the dislocation to the resistivity. The contribution value of the dislocation to the resistivity is expressed as:

[0088]

[0089] Among them, p D is the contribution of dislocation to resistivity, K is the preset first constant, G is the shear modulus, b is the Burgers vector, and Λ is the second dislocation density.

[0090] Among them, in the austenitic alloy, G = 81 GPa and b = 0.25 nm.

[0091] and / or;

[0092] Dividing the preset second constant by the average equivalent circle diameter of the second grains to obtain the contribution value of the grain boundary to the resistivity, and the contribution value of the grain boundary to the resistivity is expressed as:

[0093]

[0094] Among them, p GB is the contribution value of the grain boundary to the resistivity, A is the preset second constant, and d is the average equivalent circle diameter of the second grains.

[0095] and / or;

[0096] Substitute the preset third constant, the proportion of the area of the second precipitation phase, and the average radius of the precipitation phase into Equation (2) to calculate the contribution value of the precipitation phase to the resistivity. Among them, r is the average radius of the precipitation phase, which is calculated according to the average area of the second precipitation phase.

[0097] In an embodiment of the present application, calculating the yield strength corresponding to each of the second microstructure information of the first metal sample according to the average equivalent circle diameter of the second grains, the average area of the second precipitation phase, the proportion of the area of the second precipitation phase, and the second dislocation density includes:

[0098] Respectively calculate the contribution values of dislocations, grain boundaries, and precipitation phases to the yield strength according to the second dislocation density, the average equivalent circle diameter of the second grains, the average area of the second precipitation phase, and the proportion of the area of the second precipitation phase in the second microstructure information;

[0099] Substitute the contribution values of the dislocations, the grain boundaries, and the precipitation phases to the yield strength into the formula to calculate the yield strength corresponding to each of the second microstructure information of the first metal sample. The formula is:

[0100]

[0101] Among them, σ y is the yield strength of the second microstructure information, M is the Taylor factor, τ A is the contribution value of the dislocations to the yield strength, τ gb is the contribution value of the grain boundaries to the yield strength, τ ph is the contribution value of the precipitation phases to the yield strength.

[0102] In this embodiment, the contribution value of dislocation to yield strength is calculated based on the second dislocation density in the second microstructure information, the contribution value of grain boundary to yield strength is calculated based on the second grain average equivalent circle diameter in the second microstructure information, and the contribution value of precipitation to yield strength is calculated based on the second precipitate phase average area and the second precipitate phase area ratio in the second microstructure information.

[0103] Furthermore, the contribution value of dislocation to yield strength, the contribution value of grain boundary to yield strength, and the contribution value of precipitation to yield strength corresponding to each second microstructure information corresponding to the first metal sample are substituted into formula (5) to obtain the yield strength of each second microstructure information corresponding to the first metal sample.

[0104] By calculating the microstructure information with added noise and obtaining the yield strength, the training data can be enriched and the problem of too few training samples can be solved.

[0105] In one embodiment of the present application, the contribution of the dislocation to the yield strength is calculated in the following manner:

[0106] Taking a square root of the second dislocation density to obtain a first value; multiplying a preset fourth constant, a shear modulus, a Burgers vector and the first value to obtain a contribution value of the dislocation to the yield strength;

[0107] and / or;

[0108] The contribution of the grain boundaries to the yield strength is calculated as follows:

[0109] τ gb =k hp d -0.5 (6)

[0111] Among them, τ gb is the contribution of the grain boundary to the yield strength, k hp is a fifth preset constant, d is an average equivalent circular diameter of the second grains;

[0112] and / or;

[0113] The contribution of the precipitation phase to the yield strength is calculated as follows:

[0114] τ ph =P BND τ ph,BND +P INT τ ph,INT (7)

[0115] Among them, τ ph is the contribution of the precipitation to the yield strength, P BNDis the ratio of the precipitates on the grain boundary to the total precipitates, τ ph,BND is the contribution value of the precipitates on the grain boundary to the yield strength, p INT is the ratio of the precipitates within the grains to the total precipitates, τ ph,INT is the contribution value of the precipitates within the grains to the yield strength. The contribution value of the precipitates on the grain boundary to the yield strength and the contribution value of the precipitates within the grains to the yield strength are determined by the average area of the second precipitate phase and the proportion of the area of the second precipitate phase.

[0116] In this embodiment, taking the square root of the second dislocation density gives a fourth value. Multiplying the preset fourth constant, the shear modulus, the Burgers vector, and the fourth value gives the contribution value of the dislocations to the yield strength. The contribution value of the dislocations to the yield strength is expressed as:

[0117]

[0118] where τ A is the contribution value of the dislocations to the yield strength, α is the preset fourth constant, G is the shear modulus, b is the Burgers vector, and Λ is the second dislocation density.

[0119] and / or;

[0120] Substituting the fifth preset constant and the average equivalent circle diameter of the second grains into Equation (6), the contribution value of the grain boundary to the yield strength is calculated, where the fifth preset constant is the Hall-Petch constant.

[0121] and / or;

[0122] Substituting the ratio of the precipitates on the grain boundary to the total precipitates, the contribution value of the precipitates on the grain boundary to the yield strength, the ratio of the precipitates within the grains to the total precipitates, and the contribution value of the precipitates within the grains to the yield strength into Equation (7), the contribution value of the precipitates to the yield strength is calculated. Where p BND +P INT = 1.

[0123] Specifically, the contribution value of the precipitates on the grain boundary to the yield strength is calculated by the following method:

[0124]

[0125] where τ ph , BND is the contribution value of the precipitates on the grain boundary to the yield strength, L is the average particle spacing of the precipitate phase, X is the particle radius of the precipitate phase, G is the shear modulus, and b is the Burgers vector.

[0126] Specifically, the contribution value of the precipitates within the grains to the yield strength is calculated by the following method:

[0127]

[0128] Among them, τ ph , INT is the contribution value of the precipitates in the grains to the yield strength, L is the average particle spacing of the precipitate phase, X is the particle radius of the precipitate phase, G is the shear modulus, and b is the Burgers vector.

[0129] Among them, the average particle spacing of the precipitate phase is calculated as follows:

[0130]

[0131] Among them, L i is the average particle spacing of the precipitate phase. The subscript i of L = BND, INT, m is the number of precipitate types, f is the proportion of the precipitate phase area, r is the average radius of the precipitate phase, and r is calculated from the average area of the second precipitate phase.

[0132] Among them, the particle radius of the precipitate phase is calculated as follows:

[0133]

[0134] Among them, X i is the average particle spacing of the precipitate phase. The subscript i of X = BND, INT, m is the number of precipitate types, f is the proportion of the precipitate phase area, r is the average radius of the precipitate phase, and r is calculated from the average area of the second precipitate phase.

[0135] In an embodiment of the present application, training the neural network model with the training sample to obtain the target model includes:

[0136] During the process of training the neural network with the training sample, calculating the prediction error of the neural network model according to the yield strength predicted by the neural network and the reference yield strength;

[0137] In the case where the prediction error is greater than or equal to the preset error, taking the neural network model as the target model.

[0138] In this embodiment, during the process of training the neural network with the training sample, calculating the prediction error of the neural network model, comparing the prediction error with the preset error, and evaluating the training accuracy of the neural network model.

[0139] Specifically, during the training process, calculating the prediction error of the neural network model according to the yield strength predicted by the neural network model and the reference yield strength, and the prediction error is expressed as:

[0140]

[0141] where MAPE is the prediction error, N is the number of training samples, C i and E i represent the predicted yield strength and the reference yield resistance respectively.

[0142] When the prediction error is less than the preset error, continue to optimize the parameters of the neural network model, and automatically adjust the parameters and structure of the model according to the characteristics of the input data. When new data is added, the neural network can re-learn and adapt, so as to continuously improve the prediction accuracy. When the prediction error is greater than or equal to the preset error, the preset error can be set according to actual needs. For example, it can be set to 90%. If it is greater than or equal to 90%, it means that the prediction accuracy of the model is relatively high, and the training is completed. The neural network model is used as the target model to predict the reference yield strength.

[0143] See Figure 3 , the neural network model is composed of a shared fully connected layer, a first neural network, and a second neural network. The first neural network and the second neural network are both composed of two layers of fully connected layers, and each layer of fully connected layer can be composed of 16 neurons. When constructing the training samples, they can be divided into the first training samples and the second training samples.

[0144] The resistivity and yield strength in the first training samples are obtained through experiments and can be considered as high-precision data. The resistivity and yield strength in the second training samples are calculated through formulas and have relatively lower precision compared to the data obtained through experiments, and can be considered as low-precision data. Input the resistivity in the second training samples into the shared fully connected layer, extract features through the shared fully connected layer, and input the extracted features into the first neural network for training. Input the resistivity in the first training samples into the shared fully connected layer, extract features through the shared fully connected layer, and input the extracted features into the trained first neural network to obtain the first predicted yield strength value σ L The first predicted yield strength value σ L is concatenated with the features extracted by the shared fully connected layer, and the concatenated data is input into the second neural network to obtain the second predicted yield strength value σ H , and combined with the result h of the residual connection to obtain the predicted yield strength, and output the predicted yield strength.

[0145] By training the neural network model, a target model is obtained. Using the target model to predict the yield strength of metal samples can perform non-destructive evaluation of metal materials.

[0146] Figure 4 shows a schematic flow chart of the yield strength prediction method provided by the embodiment of the present application. As Figure 4As shown, the model training method provided by the embodiment of the present application is applied to an electronic device and includes the following steps 401-402, where:

[0147] Step 401, obtain the resistivity of the metal sample to be measured at the second environmental temperature, the second stress, and the first sampling moment.

[0148] In this embodiment, to obtain the resistivity of the metal sample to be measured at the second environmental temperature, the second stress, and the first sampling moment, the resistivity of the metal sample to be measured can be obtained by the probe method, which is a commonly used non-destructive testing method. This method places multiple probes on the surface of the material and measures the voltage and current between the probes, thereby calculating the resistivity of the metal sample to be measured, such as the four-probe method.

[0149] Step 402, input the resistivity of the metal sample to be measured into the target model to obtain the target yield strength of the metal sample, where the target model is obtained based on the above model training method.

[0150] In this embodiment, input the resistivity of the metal sample to be measured at the second environmental temperature, the second stress, and the first sampling moment into the target model, and output the target yield strength of the metal sample. The target yield strength can be predicted through the target model without damaging the metal material, which is suitable for in-service metals, that is, metals that are being used, and can perform non-destructive evaluation of metal materials.

[0151] Figure 5 Another schematic flow diagram of the yield strength prediction method provided by the embodiment of the present application is shown. The yield strength prediction method provided by the embodiment of the present application is illustrated by the following examples.

[0152] Step 501, perform quantitative characterization of the microstructure of multiple metal samples to obtain multiple first microstructure information of each metal sample.

[0153] In this embodiment, samples in different damage states are obtained through the damage interruption test, and quantitative characterization of the microstructure of multiple metal samples is performed to obtain multiple first microstructure information of each metal sample. The first microstructure information includes the first average equivalent circle diameter of grains, the first average area of precipitates, the first area ratio of precipitates, and the first dislocation density obtained by the first metal sample at the first environmental temperature, the first stress, and the sampling moment. At least one of the first environmental temperature, the first stress, and the sampling moment of any two pieces of the first microstructure information is different.

[0154] Taking the 800H alloy as an example, taking the high-temperature and high-stress creep damage of the 800H alloy as an example, for example, select the creep conditions of 750°C and 95 MPa for the interruption test to obtain metal samples in different creep stages, such as Figure 6 As shownFigure 6 The creep conditions of the samples at 750 °C, 95 MPa, and different sampling times are shown. Referring to Table 1, the data of the sample numbered O is the control group. For example, the sample numbered A is the 800H alloy at 750 °C, 95 MPa, and the sampling time is 10 h. At this time, the strain of the 800H alloy is 0.364, and the state is the decelerated creep stage. As time changes, the strain also increases.

[0155] Table 1

[0156]

[0157] Perform multi-scale microstructural quantitative characterization on the samples in different creep stages. Specifically, use electron backscatter diffraction (EBSD) technology to obtain the average equivalent circle diameter of grains. EBSD is a microstructure analysis tool that can provide various microstructural information including grain size. Use a scanning electron microscope (SEM) to obtain the average area of the precipitates and the proportion of the precipitate area. Use X-ray diffraction (XRD) to obtain the dislocation density.

[0158] Step 502, generate corresponding second microstructural information based on each piece of the first microstructural information of the metal sample.

[0159] As Figures 2a - 2d shown, Figure 2a is a schematic diagram of the proportion of the precipitate area, Figure 2b is a schematic diagram of the average area of the precipitates, Figure 2c is a schematic diagram of the average equivalent circle diameter of grains, Figure 2d is a schematic diagram of the dislocation density; among them, the precipitates formed by the 800H alloy during creep are mainly divided into two types, namely M 23 C6 on the grain boundaries and M 23 C6 inside the grains, Figures 2a - 2d The experimental data in 23 are obtained by the above EBSD, SEM, and XRD technical means. The generation data of M 23 C6 on the grain boundaries is obtained by adding noise to the experimental data (that is, adding noise to the first average equivalent circle diameter of grains, the first average area of the precipitates, the first proportion of the precipitate area, and the first dislocation density in each piece of the first microstructural information respectively to obtain the second average equivalent circle diameter of grains, the second average area of the precipitates, the second proportion of the precipitate area, and the second dislocation density). By adding oscillating noise, a large amount of microstructural information conforming to this law is generated.

[0160] For each metal sample, the resistivity was measured by the four-probe method, and the yield strength was measured by a room-temperature tensile test, obtaining the experimental resistivity and the corresponding experimental yield strength corresponding to each of the first microstructure information of the first metal sample (i.e., the experimental resistivity and the corresponding experimental yield strength corresponding to each of the first microstructure information of the first metal sample as described above).

[0161] Step 503: Establish a resistivity calculation model and a yield strength calculation model based on microstructure characteristics.

[0162] In this embodiment, combining the influence mechanisms of different microstructures on resistivity and yield strength, a resistivity calculation model and a yield strength calculation model based on microstructure characteristics are established, and the calculation models include multiple formulas (i.e., Formulas (1)-(12) as described above).

[0163] Step 504: Generate corresponding second microstructure information based on each of the first microstructure information of the metal sample, and use the resistivity calculation model to calculate the resistivity corresponding to the second microstructure information.

[0164] In this embodiment, substituting each of the second microstructure information into the resistivity calculation model to obtain the resistivity corresponding to the second microstructure information (i.e., calculating the resistivity corresponding to each of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of the grains, the average area of the second precipitated phases, the area ratio of the second precipitated phases, and the second dislocation density as described above).

[0165] In addition, substituting each of the first microstructure information into the resistivity calculation model to obtain the resistivity corresponding to the first microstructure information. Comparing the resistivity corresponding to the first microstructure information, the experimental resistivity, and the resistivity corresponding to the second microstructure information, see Figure 7 , Figure 7 where the model prediction value in is the resistivity obtained by substituting the first microstructure information into the resistivity calculation model, the experimental value refers to the experimental resistivity corresponding to the first microstructure information, and the calculated value refers to substituting each of the second microstructure information into the resistivity calculation model to obtain the resistivity corresponding to the second microstructure information.

[0166] Optionally, in the calculation model, k hp , α are set as constants to be optimized. Combining the quantitative information of the microstructure, the method of minimizing the sum of squared residuals is used to fit and optimize the constants of the calculation model, and the optimized constants are respectively: k hp = 0.27 MPa·m0.5, M = 2.87, α = 0.6.

[0167] Optionally, combining the quantitative information of the microstructure and the actually measured resistivity value, the constants of the calculation model are fitted and optimized by the method of minimizing the sum of squared residuals. The optimized constants are: K = 1.92e-22, A = 1.59e-11 μΩm 2 , P = 7.90e-11.

[0168] Step 505: Generate corresponding second microstructure information based on each first microstructure information of the metal sample, and use the yield strength calculation model to calculate the yield strength corresponding to the second microstructure information.

[0169] In this embodiment, each of the second microstructure information is substituted into the yield strength calculation model to obtain the yield strength corresponding to the second microstructure information (that is, the yield strength corresponding to each of the second microstructure information of the first metal sample is calculated according to the second average equivalent circle diameter of the grains, the second average area of the precipitates, the second area ratio of the precipitates, and the second dislocation density as described above).

[0170] In addition, each of the first microstructure information is substituted into the yield strength calculation model to obtain the yield strength corresponding to the first microstructure information. Compare the yield strength corresponding to the first microstructure information, the experimental yield strength, and the yield strength corresponding to the second microstructure information. See Figure 8 , Figure 8 In, the model prediction value is the yield strength obtained by substituting the first microstructure information into the yield strength calculation model, the experimental value refers to the experimental yield strength corresponding to the first microstructure information, and the calculated value refers to the yield strength corresponding to each of the second microstructure information obtained by substituting each of the second microstructure information into the yield strength calculation model.

[0171] Step 506: Take the resistivity corresponding to the second microstructure information and the corresponding yield strength as low-precision data, and take the experimental resistivity corresponding to the first microstructure information and the corresponding experimental yield strength as high-precision data.

[0172] In this embodiment, the resistivity corresponding to the second microstructure information and the corresponding yield strength are taken as low-precision data (that is, the first training sample as described above), and the experimental resistivity corresponding to the first microstructure information and the corresponding experimental yield strength are taken as high-precision data (that is, the second training sample as described above).

[0173] Step 507: Train the neural network model based on the low-precision data and the high-precision data to obtain the target model.

[0174] In this embodiment, the experimental data serves as high-precision data, and the generated data serves as low-precision data. Together, they form a data set, with 60% used for model training and 40% used for testing. The neural network model is trained using the low-precision data and the high-precision data. The neural network model consists of multiple fully connected layers, and through training, a target model is obtained. The target model is used to predict the yield strength of the metal to be measured.

[0175] The neural network model consists of a shared fully connected layer, a high-precision neural network, and a low-precision neural network. Both the high-precision neural network and the low-precision neural network consist of two fully connected layers, and each fully connected layer can be composed of 16 neurons.

[0176] The resistivity and yield strength in the high-precision data are obtained through experiments, while the resistivity and yield strength in the low-precision data are calculated through formulas, and the data accuracy is relatively lower compared to the data obtained through experiments. The resistivity in the low-precision data is input into the shared fully connected layer, and features are extracted through the shared fully connected layer. The extracted features are input into the first neural network for training. The resistivity in the high-precision data is input into the shared fully connected layer, and features are extracted through the shared fully connected layer. The extracted features are input into the trained low-precision neural network to obtain the first predicted yield strength value σ L , the first predicted yield strength value σ L is concatenated with the features extracted by the shared fully connected layer, and the concatenated data is input into the high-precision neural network to obtain the second predicted yield strength value σ H , and is combined with the result h of the residual connection to obtain the predicted yield strength, and the predicted yield strength is output.

[0177] During the training process, the prediction error of the neural network model is calculated based on the yield strength predicted by the neural network model and the reference yield strength. The calculation formula is shown in Equation (13).

[0178] See Figure 9 , through the comparison between the yield strength prediction value obtained by the neural network model and the experimental value of the yield strength, it can be shown that the prediction ability of the model is good. Among them, the prediction results of the high-precision data test set and the high-precision data training set are both near the perfect prediction line of y = x, and more than 90% of the low-precision data falls within the 90% prediction accuracy band, and the MAPE is 4.4%. The trained neural network model is used as the target model.

[0179] Step 508, input the resistivity obtained by non-destructive testing into the target model to obtain the predicted yield strength.

[0180] In this embodiment, the resistivity obtained from non-destructive testing, such as the resistivity measured by the probe method, is input into the target model (that is, in the process of training the neural network with the training samples, according to the yield strength predicted by the neural network and the reference yield strength, the prediction error of the neural network model is calculated; in the case where the prediction error is greater than or equal to the preset error, the neural network model is used as the target model).

[0181] In the prior art, model training is basically based on experimental data, and different models are used to fine-tune model parameters. For fields where a large amount of experimental data cannot be obtained, there is an obvious overfitting phenomenon and the model generalization ability is poor. Based on the evolution trend of the quantitative information of the microstructure during the material damage process, this application generates a large amount of quantitative microstructure data that conforms to this trend. Through the established physical property and mechanical property calculation models based on microstructure characteristics, the physical properties and mechanical properties corresponding to specific microstructures are calculated. Finally, using the multi-precision neural network method, data with different precisions are integrated for training and testing, which can solve the problems of too little experimental data and large data noise. Moreover, using the target model to predict the yield strength of metal samples, subsequent in the process of evaluating the mechanical properties of metal materials, it is not necessary to damage the metal materials, and the metal materials can be non-destructively evaluated, which has universality.

[0182] Figure 10 The structural diagram of the model training device provided by the embodiment of the present application is shown. As Figure 10 shown, the model training device 1000 includes:

[0183] The first acquisition module 1001 is used to acquire a plurality of training samples, where the training samples include the first environmental temperature, the first stress, the first resistivity of the first metal sample at the target sampling moment, and the reference yield strength of the first metal sample at the target sampling moment. The target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress, and at least one of the first environmental temperature, the first stress, and the target sampling moment of any two training samples is different;

[0184] The training module 1002 is used to train the neural network model with the training samples to obtain a target model, and the target model is used to predict the yield strength of the metal sample to be measured.

[0185] In an embodiment of the present application, the first acquisition module 1001 includes: a first acquisition sub-module, a noise processing sub-module, a calculation sub-module, and a construction sub-module;

[0186] The first acquisition sub-module is used to acquire a plurality of first microstructure information corresponding to the first metal sample. The first microstructure information includes the first average equivalent circle diameter of grains, the first average area of precipitation phases, the proportion of the area of precipitation phases, and the first dislocation density obtained from the first metal sample at a first environmental temperature, a first stress, and a sampling time. At least one of the first environmental temperature, the first stress, and the sampling time of any two pieces of the first microstructure information is different;

[0187] The noise processing sub-module is used to perform noise addition processing on the first average equivalent circle diameter of grains, the first average area of precipitation phases, the proportion of the area of precipitation phases, and the first dislocation density in each piece of the first microstructure information respectively, to obtain a second average equivalent circle diameter of grains, a second average area of precipitation phases, a proportion of the area of precipitation phases, and a second dislocation density. The second average equivalent circle diameter of grains, the second average area of precipitation phases, the proportion of the area of precipitation phases, and the second dislocation density are used as the second microstructure information;

[0188] The calculation sub-module is used to calculate the resistivity corresponding to each piece of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of grains, the second average area of precipitation phases, the proportion of the area of precipitation phases, and the second dislocation density; calculate the yield strength corresponding to each piece of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of grains, the second average area of precipitation phases, the proportion of the area of precipitation phases, and the second dislocation density;

[0189] The construction sub-module is used to obtain a plurality of training samples according to the resistivity corresponding to each piece of the second microstructure information of the first metal sample, the corresponding yield strength, the experimental resistivity corresponding to each piece of the first microstructure information of the first metal sample, and the corresponding experimental yield strength. The experimental resistivity and the experimental yield strength are obtained through experimental measurement.

[0190] In an embodiment of the present application, the calculation sub-module includes: a first calculation sub-unit;

[0191] The first calculation sub-unit is used to calculate the contribution values of dislocations, grain boundaries, and precipitation phases to the resistivity respectively according to the second dislocation density, the second average equivalent circle diameter of grains, and the proportion of the area of precipitation phases in the second microstructure information; add the intrinsic resistivity of the first metal sample, the contributions of the dislocations, the grain boundaries, and the precipitation phases to the resistivity to obtain the resistivity corresponding to each piece of the second microstructure information of the first metal sample.

[0192] In an embodiment of the present application, the calculation sub-module further includes: a second calculation sub-unit;

[0193] A second calculation subunit, configured to calculate the contribution values of dislocations, grain boundaries, and precipitation phases to the yield strength respectively according to the second dislocation density, the second average equivalent circle diameter of grains, the second average area of precipitation phases, and the proportion of the area of the second precipitation phase in the second microstructure information;

[0194] Substitute the contribution values of the dislocations, the grain boundaries, and the precipitation phases to the yield strength into a formula to calculate the yield strength of each second microstructure information corresponding to the first metal sample, where the formula is:

[0195]

[0196] where, σ y is the yield strength of the second microstructure information, M is the Taylor factor, τ A is the contribution value of the dislocations to the yield strength, τ gb is the contribution value of the grain boundaries to the yield strength, τ ph is the contribution value of the precipitation phases to the yield strength.

[0197] In an embodiment of the present application, the first calculation subunit is specifically configured to take the square root of the second dislocation density to obtain a first value; multiply the shear modulus, the Burgers vector, and the first value to obtain a second value; calculate the square of the second value to obtain a third value; multiply the third value by a preset first constant to obtain the contribution value of the dislocations to the resistivity; and / or;

[0198] The first calculation subunit is specifically configured to calculate the contribution value of the grain boundaries to the resistivity in the following manner:

[0199] Divide a preset second constant by the second average equivalent circle diameter of grains to obtain the contribution value of the grain boundaries to the resistivity;

[0200] and / or;

[0201] The first calculation subunit is specifically configured to calculate the contribution value of the precipitation phases to the resistivity in the following manner:

[0202]

[0203] where, ρ P is the contribution value of the precipitation phases to the resistivity, P is a preset third constant, f is the proportion of the area of the second precipitation phase, r is the average radius of the precipitation phases, and r is calculated from the second average area of the precipitation phases.

[0204] In an embodiment of the present application, the second calculation subunit is specifically configured to take the square root of the second dislocation density to obtain a fourth value; multiply a preset fourth constant, a shear modulus, a Burgers vector, and the fourth value to obtain the contribution value of the dislocation to the yield strength;

[0205] and / or;

[0206] The second calculation subunit is specifically configured to calculate the contribution value of the grain boundary to the yield strength in the following manner:

[0207] τ gb =k hp d -0.5

[0208] where τ gb is the contribution value of the grain boundary to the yield strength, k hp is a fifth preset constant, and d is the average equivalent circle diameter of the second grain;

[0209] and / or;

[0210] The second calculation subunit is specifically configured to calculate the contribution value of the precipitate to the yield strength in the following manner:

[0211] τ ph =p BND τ ph,BND +P INT τ ph,INT

[0212] where τ ph is the contribution value of the precipitate to the yield strength, P BND is the ratio of the precipitate on the grain boundary to the total precipitate, τ ph,BND is the contribution value of the precipitate on the grain boundary to the yield strength, P INT is the ratio of the precipitate in the grain to the total precipitate, τ ph,INT is the contribution value of the precipitate in the grain to the yield strength, and the contribution value of the precipitate on the grain boundary to the yield strength and the contribution value of the precipitate in the grain to the yield strength are determined by the average area of the second precipitate phase and the proportion of the second precipitate phase area.

[0213] In an embodiment of the present application, the training module is specifically configured to calculate the prediction error of the neural network model according to the yield strength predicted by the neural network and the reference yield strength during the process of training the neural network with the training sample; in the case where the prediction error is greater than or equal to a preset error, use the neural network model as the target model.

[0214] The model training device provided by the embodiments of the present application can implement each process implemented by the foregoing embodiments of the model training method and achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0215] Figure 11 The structural diagram of the yield strength prediction device provided by the embodiments of the present application is shown. As Figure 11 shown, the yield strength prediction device 1100 includes:

[0216] A second acquisition module 1101, configured to acquire the resistivity of a metal sample to be measured at a second ambient temperature, a second stress, and a first sampling moment;

[0217] A prediction module 1102, configured to input the resistivity of the metal sample to be measured into a target model to obtain the target yield strength of the metal sample, where the target model is obtained based on the model training method according to any one of claims 1 to 6.

[0218] The yield strength prediction device provided by the embodiments of the present application can implement each process implemented by the foregoing embodiments of the yield strength prediction method and achieve the same technical effects. To avoid repetition, it will not be elaborated here.

[0219] Figure 12 The schematic hardware structure diagram of the electronic device provided by the embodiments of the present application is shown.

[0220] The electronic device may include a processor 1201 and a memory 1202 storing computer program instructions.

[0221] Specifically, the foregoing processor 1201 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application.

[0222] The memory 1202 may include a mass storage for data or instructions. By way of example and not limitation, the memory 1202 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. In a suitable case, the memory 1202 may include a removable or non-removable (or fixed) medium. In a suitable case, the memory 1202 may be inside or outside the integrated gateway disaster recovery device. In a specific embodiment, the memory 1202 is a non-volatile solid state memory.

[0223] The memory may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage media device, an optical storage media device, a flash memory device, an electrical, optical, or other physical / tangible memory storage device. Thus, generally, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., memory devices) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to the first aspect or the second aspect of the present disclosure.

[0224] The processor 1201 realizes any one of the information auditing methods in the above embodiments by reading and executing the computer program instructions stored in the memory 1202.

[0225] In one example, the electronic device may further include a communication interface 1203 and a bus 1210. Among them, as Figure 12 shown, the processor 1201, the memory 1202, and the communication interface 1203 are connected through the bus 1210 and complete communication with each other.

[0226] The communication interface 1203 is mainly used to realize the communication between each module, device, unit, and / or device in the embodiments of the present application.

[0227] The bus 1210 includes hardware, software, or both, and couples the components of the information auditing method or the verification device to each other. By way of example and not limitation, the bus may include an Accelerated Graphics Port (AGP) or other graphics bus, an Enhanced Industry Standard Architecture (EISA) bus, a Front Side Bus (FSB), a HyperTransport (HT) interconnect, an Industry Standard Architecture (ISA) bus, an InfiniBand interconnect, a Low Pin Count (LPC) bus, a memory bus, a MicroChannel Architecture (MCA) bus, a Peripheral Component Interconnect (PCI) bus, a PCI-Express (PCI-X) bus, a Serial Advanced Technology Attachment (SATA) bus, a Video Electronics Standards Association Local (VLB) bus, or other suitable buses or a combination of two or more of these. In a suitable case, the bus 1210 may include one or more buses. Although the embodiments of the present application describe and illustrate specific buses, the present application contemplates any suitable bus or interconnect.

[0228] In addition, in combination with the model training method in the above embodiments, the embodiments of the present application can be implemented by providing a computer storage medium. Computer program instructions are stored on the computer storage medium; when the computer program instructions are executed by a processor, any one of the model training methods in the above embodiments is realized.

[0229] In addition, embodiments of the present application may be implemented by providing a computer program product. When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device implements any one of the model training methods in the above embodiments.

[0230] It should be clear that the present application is not limited to the specific configurations and processes described above and shown in the figures. For the sake of brevity, detailed descriptions of known methods are omitted here. In the above embodiments, several specific steps are described as examples. However, the method process of the present application is not limited to the described specific steps. Those skilled in the art can make various changes, modifications, and additions, or change the order between steps after understanding the spirit of the present application.

[0231] The functional blocks shown in the above block diagrams can be implemented as hardware, software, firmware, or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application-specific integrated circuit (ASIC), appropriate firmware, a plug-in, a functional card, and so on. When implemented in software, the elements of the present application are programs or code segments used to perform the required tasks. The program or code segment can be stored in a machine-readable medium or transmitted via a data signal carried in a carrier wave on a transmission medium or a communication link. A "machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical discs, hard disks, fiber optic media, radio frequency (RF) links, and so on. The code segment can be downloaded via a computer network such as the Internet, an intranet, and so on.

[0232] It should also be noted that the exemplary embodiments mentioned in the present application describe some methods or systems based on a series of steps or devices. However, the present application is not limited to the order of the above steps, that is, the steps can be executed in the order mentioned in the embodiments, can be different from the order in the embodiments, or several steps can be executed simultaneously.

[0233] As described above with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems) and computer program products according to embodiments of the present disclosure. It should be understood that each block in the flowchart and / or block diagram, and the combination of blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general purpose computer, a special purpose computer, or other programmable data processing device to produce a machine, such that the instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the functions / actions specified in one or more blocks of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general purpose processor, a special purpose processor, a special application processor or a field programmable logic circuit. It should also be understood that each block in the block diagram and / or flowchart, and the combination of blocks in the block diagram and / or flowchart, can also be implemented by dedicated hardware that performs the specified functions or actions, or can be implemented by a combination of dedicated hardware and computer instructions.

[0234] As described above, the above is only the specific implementation manner of the present application. Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present application.

Claims

1. A model training method, characterized in that, The method includes: Obtaining a plurality of training samples, where the training samples include a first environmental temperature, a first stress, a first resistivity of a first metal sample at a target sampling moment, and a reference yield strength of the first metal sample at the target sampling moment. The target sampling moment is the sampling moment of the first metal sample under the first environmental temperature and the first stress. At least one of the first environmental temperature, the first stress, and the target sampling moment is different for any two of the training samples; Using the training samples to train a neural network model to obtain a target model, where the target model is used to predict the yield strength of a metal sample to be measured.

2. The model training method according to claim 1, wherein The obtaining of the plurality of training samples includes: Obtaining a plurality of first microstructure information corresponding to the first metal sample, where the first microstructure information includes a first average equivalent circle diameter of grains, a first average area of precipitates, a first area ratio of precipitates, and a first dislocation density obtained from the first metal sample under the first environmental temperature, the first stress, and the sampling moment. At least one of the first environmental temperature, the first stress, and the sampling moment is different for any two of the first microstructure information; Performing noise addition processing on the first average equivalent circle diameter of grains, the first average area of precipitates, the first area ratio of precipitates, and the first dislocation density in each of the first microstructure information to obtain a second average equivalent circle diameter of grains, a second average area of precipitates, a second area ratio of precipitates, and a second dislocation density; the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density are used as second microstructure information; Calculating the resistivity corresponding to each of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density; Calculating the yield strength corresponding to each of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density; Obtaining a plurality of training samples according to the resistivity corresponding to each of the second microstructure information of the first metal sample, the corresponding yield strength, the experimental resistivity corresponding to each of the first microstructure information of the first metal sample, and the corresponding experimental yield strength, where the experimental resistivity and the experimental yield strength are obtained through experimental measurement; 3. The model training method according to claim 2, wherein The calculating of the resistivity corresponding to each of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of grains, the second average area of precipitates, the second area ratio of precipitates, and the second dislocation density includes: Respectively calculating the contribution values of dislocations, grain boundaries, and precipitates to the resistivity according to the second dislocation density, the second average equivalent circle diameter of grains, and the second area ratio of precipitates in the second microstructure information; Add the contribution values of the intrinsic resistivity of the first metal sample, the dislocations, the grain boundaries, and the precipitates to the resistivity respectively to obtain the resistivity of each second microstructure information corresponding to the first metal sample.

4. The model training method according to claim 2, wherein Calculating the yield strength corresponding to each of the second microstructure information of the first metal sample according to the second average equivalent circle diameter of the grains, the average area of the second precipitates, the area ratio of the second precipitates, and the second dislocation density includes: Calculate the contribution values of the dislocations, grain boundaries, and precipitates to the yield strength respectively according to the second dislocation density, the second average equivalent circle diameter of the grains, the average area of the second precipitates, and the area ratio of the second precipitates in the second microstructure information; Substitute the contribution values of the dislocations, grain boundaries, and precipitates to the yield strength into the formula to calculate the yield strength of each second microstructure information corresponding to the first metal sample. The formula is: Among them, σ y is the yield strength of the second microstructure information, M is the Taylor factor, τ A is the contribution value of the dislocation to the yield strength, τ gb is the contribution value of the grain boundary to the yield strength, τ ph is the contribution value of the precipitate to the yield strength.

5. The model training method according to claim 3, wherein The contribution value of the dislocations to the resistivity is calculated by the following method: Take the square root of the second dislocation density to get a first value; multiply the shear modulus, Burgers vector, and the first value to get a second value; calculate the square of the second value to get a third value; multiply the third value by a preset first constant to get the contribution value of the dislocations to the resistivity; and / or; The contribution value of the grain boundaries to the resistivity is calculated by the following method: Divide a preset second constant by the second average equivalent circle diameter of the grains to get the contribution value of the grain boundaries to the resistivity; and / or; The contribution value of the precipitates to the resistivity is calculated by the following method: where ρ P is the contribution value of the precipitation phase to the relative resistivity, P is a preset third constant, f is the area fraction of the second precipitation phase, r is the average radius of the precipitation phase, and r is calculated from the average area of the second precipitation phase.

6. The model training method according to claim 4, wherein The contribution value of the dislocations to the yield strength is calculated by the following method: Take the square root of the second dislocation density to get a fourth value; multiply a preset fourth constant, the shear modulus, the Burgers vector, and the fourth value to get the contribution value of the dislocations to the yield strength; and / or; The contribution value of the grain boundaries to the yield strength is calculated by the following method: τ gb = k hp d -0.5 Among them, τ gb is the contribution value of the grain boundary to the yield strength, k hp is the fifth preset constant, and d is the average equivalent circle diameter of the second grain; and / or; The contribution value of the precipitates to the yield strength is calculated by the following method: τ ph = P BND τ ph,BND + P INT τ ph,INT Among them, τ ph is the contribution value of the precipitation phase to the yield strength, P BND is the ratio of the precipitates on the grain boundaries to the total precipitates, τ ph,BND is the contribution value of the precipitates on the grain boundaries to the yield strength, P INT is the ratio of the precipitates within the grains to the total precipitates, τ ph,INT is the contribution value of the precipitates within the grains to the yield strength. The contribution value of the precipitates on the grain boundaries to the yield strength and the contribution value of the precipitates within the grains to the yield strength are determined by the average area of the second precipitation phase and the proportion of the area of the second precipitation phase.

7. The model training method according to any one of claims 1 to 6, characterized in that, Training the neural network model with the training samples to obtain the target model includes: During the process of training the neural network with the training samples, calculate the prediction error of the neural network model according to the yield strength predicted by the neural network and the reference yield strength; In the case where the prediction error is greater than or equal to the preset error, use the neural network model as the target model.

8. A yield strength prediction method, characterized in that, The method includes: Obtain the resistivity of the metal sample to be measured at the second environmental temperature, the second stress, and the first sampling moment; Input the resistivity of the metal sample to be measured into the target model to obtain the target yield strength of the metal sample. The target model is obtained based on the model training method according to any one of claims 1 to 7.

9. A model training device, characterized in that, The device includes: A first acquisition module, configured to acquire a plurality of training samples, where the training samples include a first environmental temperature, a first stress, a first resistivity of a first metal sample at a target sampling moment, and a reference yield strength of the first metal sample at the target sampling moment, the target sampling moment being the sampling moment of the first metal sample under the first environmental temperature and the first stress, and at least one of the first environmental temperature, the first stress, and the target sampling moment of any two of the training samples is different; A training module, configured to use the training samples to train a neural network model to obtain a target model, where the target model is used to predict the yield strength of a metal sample to be measured.

10. A yield strength prediction device, characterized in that, The device includes: A second acquisition module, configured to acquire the resistivity of a metal sample to be measured at a second environmental temperature, a second stress, and a first sampling moment; A prediction module, configured to input the resistivity of the metal sample to be measured into the target model to obtain the target yield strength of the metal sample, where the target model is obtained based on the model training method according to any one of claims 1 to 7.

11. An electronic device, characterized in that, including: A processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the model training method according to any one of claims 1 to 7 or the yield strength prediction method according to claim 8 is implemented.

Images