A method for testing the resistivity of molten metal for identifying its liquid structure

Abstract

The present invention discloses a metal melt resistivity testing method for metal liquid structure identification, which relates to the field of resistance measurement technology. The method comprises the following steps: obtaining a conversion function between conventional melt partitions and characteristic melt partitions; forming a neural network model for the characteristic melt partitions; forming a hidden layer consisting of a convection simulation model and a temperature correlation simulation model; training the convection simulation model; training the temperature correlation simulation model; forming a superposition algorithm mechanism for the convection simulation model and the temperature correlation simulation model; measuring the temperature of the characteristic melt partition, and obtaining the temperature of the melt partition through the superposition algorithm mechanism; establishing a temperature-resistivity relationship model, obtaining the resistivity of the melt partition based on the temperature-resistivity relationship model, and averaging the resistance of the melt partitions to obtain the resistivity of the metal melt. By forming the neural network model, the overall resistivity of the metal melt is comprehensively obtained based on the relationship between temperature and resistivity.

Description

Technical Field

[0001] The present invention relates to the technical field of resistance measurement, and in particular to a method for testing the resistivity of a molten metal for identifying the structure of the molten metal. Background Art

[0002] The main methods for studying liquid metal structure include direct testing, theoretical calculations, and physical property testing. Direct testing mainly includes melt X-ray diffraction, electron and neutron diffraction, and synchrotron radiation technology. Some structure-sensitive physical properties of liquid metal are closely related to the melt structure, and changes in macroscopic physical properties are the outward manifestation of changes in the melt's microstructure. Structure-sensitive physical properties mainly include melt viscosity, density, resistivity, and thermoelectric potential testing. Resistivity, as an electrical property of alloy melts, can reflect the alloy melt's microstructure, the potential energy of interactions between electrons and atoms, and the electronic state.

[0003] During measurement, due to the complex fluid movement, not only does temperature loss occur, but fluid movement also occurs, and the internal solution is blocked by the external solution, making it difficult to accurately measure the temperature. The resistivity of the metal melt is closely related to the temperature, which leads to inaccurate resistivity measurements. Summary of the Invention

[0004] In order to solve the above technical problems, a metal melt resistivity testing method for metal liquid structure identification is provided. This technical solution solves the problems raised in the above background technology.

[0005] In order to achieve the above objects, the technical solution adopted by the present invention is:

[0006] A method for testing the resistivity of a molten metal for identifying its liquid structure, comprising:

[0007] Establish a three-dimensional coordinate system for the environment where the metal melt is located;

[0008] Divide the metal melt into blocks to obtain at least one melt partition, use the melt partition at the front of the upper left corner as a characteristic melt partition, use the remaining melt partitions as regular melt partitions, and obtain a conversion function between the regular melt partition and the characteristic melt partition;

[0009] A neural network model is formed for the characteristic melt partition. The neural network model consists of an input layer, a hidden layer, and an output layer. The input layer inputs the surface temperature of the characteristic melt partition and the coordinates of the temperature prediction position, and the output layer outputs the temperature of the temperature prediction position.

[0010] The hidden layer consists of a convection simulation model and a temperature correlation simulation model;

[0011] The convection simulation model is trained until the prediction error of the convection motion of the molten metal by the flow simulation model is less than a preset distance;

[0012] The temperature correlation simulation model is trained until the prediction error of the temperature correlation simulation model for the heat conduction of the metal melt is less than a preset temperature;

[0013] A superposition algorithm mechanism is formed for the convection simulation model and the temperature correlation simulation model, and the superposition algorithm mechanism converts the surface temperature of the characteristic melt partition and the coordinates of the temperature prediction position into the temperature of the temperature prediction position;

[0014] The temperature of the characteristic melt partition is measured and the temperature of the melt partition is obtained through the superposition algorithm mechanism;

[0015] A temperature-resistivity relationship model is established. Based on the temperature-resistivity relationship model, the resistivity of the melt partition is obtained. The resistance of the melt partition is averaged to obtain the resistivity of the metal melt.

[0016] Preferably, the step of dividing the metal melt into blocks to obtain at least one melt partition comprises the following steps:

[0017] The test obtains a preset volume, and satisfies that the fluctuation range of the resistivity within the preset volume is less than the allowable error of the test;

[0018] The metal melt is evenly divided into blocks to obtain at least one melt partition, and the volume of the melt partition is equal to the preset volume.

[0019] Preferably, the step of obtaining the conversion function between conventional melt partitions and characteristic melt partitions comprises the following steps:

[0020] The melt partition located at the center of the metal melt is used as the target melt partition;

[0021] Obtain the center coordinates of the characteristic melt partition as the characteristic coordinates, and obtain the center coordinates of the conventional melt partition as the conventional coordinates;

[0022] Calculate the distance between the characteristic coordinate and the target melt partition to obtain the characteristic distance, calculate the distance between the conventional coordinate and the target melt partition to obtain the conventional distance, and divide the conventional distance by the characteristic distance to obtain the conversion coefficient;

[0023] The conventional coordinates and the conversion coefficients are paired and fitted to obtain a conversion function, where the conventional coordinates are the independent variables and the conversion coefficients are the dependent variables.

[0024] Preferably, the training of the convection simulation model comprises the following steps:

[0025] Obtaining a first displacement range of molecular motion in the metal melt, dividing the first displacement range of molecular motion into equal intervals to obtain at least one first displacement point;

[0026] In a non-viscous environment, a second displacement range of a single molecule motion of the metal melt is obtained, and the second displacement range of the single molecule motion is divided into equal intervals to obtain at least one second displacement point, wherein the number of the second displacement points is the same as that of the first displacement points, the second displacement points and the first displacement points are both vectors, and the size of the second displacement points relative to the first displacement points is their modulus.

[0027] The first displacement points are numbered from small to large, and the second displacement points are numbered from small to large. The viscosity resistance value is obtained by subtracting the value of the second displacement point with the same number from the first displacement point.

[0028] The second displacement point is paired with the viscous barrier value and fitted to obtain the viscous barrier function, which is a vector;

[0029] In a non-viscous environment, the displacement of a single molecule of the metal melt is equal to the ambient temperature of the second displacement point;

[0030] Pair and fit the ambient temperature and motion displacement to obtain the thermal motion function, which is a vector;

[0031] A convection simulation model is formed, and the convection simulation model includes aP(Q(k))+bQ(k), wherein a is a first training coefficient, k is temperature, b is a second training coefficient, Q(k) is a thermal motion function, and P(Q(k)) is a viscous barrier function;

[0032] Obtaining a first data set trained by a user, where the first data set consists of temperature and convective displacement caused by the temperature;

[0033] The temperature in the first data set is input into the convection simulation model. When the difference between the displacement output by the convection simulation model and the convection displacement caused by the temperature in the first data set is not less than a preset distance, a and b are adjusted until the difference between the displacement output by the convection simulation model and the convection displacement caused by the temperature in the first data set is less than a preset distance. The preset distance is set based on the allowable error of the test.

[0034] Preferably, the training of the temperature correlation simulation model comprises the following steps:

[0035] Obtaining the temperature variation range of the characteristic melt partition, dividing the temperature variation range into equal intervals, and obtaining at least one test point;

[0036] Obtaining a distance range from the melt partition to the characteristic melt partition, dividing the distance range into equal intervals, and obtaining at least one sample point;

[0037] Randomly combine test points and sample points to obtain at least one test group;

[0038] Under the conditions of the test group, obtain the temperature value of the melt partition;

[0039] The test points, sample points and the temperature values ​​at the melt partition are paired and fitted to obtain a temperature correlation function, wherein the test points and sample points are independent variables and the temperature value at the melt partition is the dependent variable;

[0040] A temperature correlation simulation model is formed. The temperature correlation simulation model is cL(x, y), where c is the training balance coefficient, L(x, y) is the temperature correlation function, x is the surface temperature of the characteristic melt partition, and y is the distance from the temperature prediction position to the characteristic melt partition;

[0041] Acquire a second data set, the second data set consisting of a surface temperature of a characteristic melt partition, a distance from a temperature prediction position to the characteristic melt partition, and a sample temperature at the temperature prediction position;

[0042] The surface temperature of the characteristic melt partition and the distance from the temperature prediction position to the characteristic melt partition in the second data set are input into the temperature correlation simulation model. When the difference between the test temperature output by the temperature correlation simulation model and the corresponding sample temperature is not less than the preset temperature, c is adjusted until the difference between the test temperature output by the temperature correlation simulation model and the corresponding sample temperature is less than the preset temperature. The preset temperature is the allowable error of the test.

[0043] Preferably, the forming of a superposition algorithm mechanism for the convection simulation model and the temperature correlation simulation model comprises the following steps:

[0044] Inputting the surface temperature of the characteristic melt partition and the distance from the temperature prediction position to the characteristic melt partition into the temperature correlation simulation model to obtain the first temperature of the temperature prediction position;

[0045] From the equation w + aP (Q (cL (x, w))) + bQ (cL (x, w)) = y, solve for the compensation distance w;

[0046] The characteristic melt partition surface temperature and compensation distance are input into the temperature correlation simulation model to obtain the second temperature at the coordinate position;

[0047] The coordinates of the temperature prediction position are input into the conversion function to obtain the actual coefficient, and the average of the first temperature and the second temperature is multiplied by the actual coefficient to obtain the temperature of the temperature prediction position.

[0048] Preferably, the establishment of the relationship model between temperature and resistivity includes the following steps:

[0049] Obtaining a sample block, where the volume of the sample block is equal to a preset volume;

[0050] heating the sample block, and obtaining the resistivity of the sample block under the condition that the temperature of the sample block is equal to the value at the test point;

[0051] The values ​​at the identified points are paired with the resistivity of the sample block and fitted to obtain the resistivity prediction function.

[0052] Preferably, obtaining the resistivity of the melt partition based on the relationship model between temperature and resistivity comprises the following steps:

[0053] Substituting the temperature of the melt partition into the resistivity prediction function, the resistivity of the melt partition is obtained.

[0054] Compared with the prior art, the present invention has the following beneficial effects:

[0055] The present invention provides a metal melt resistivity testing method for metal liquid structure identification. By forming a neural network model, not only the temperature diffusion is predicted in the neural network model, but also the effect of fluid viscosity on fluid convection is predicted. Thus, based on the temperature measurement of characteristic melt partitions, the temperature at different positions in the metal melt can be more accurately predicted. Furthermore, the overall resistivity of the metal melt can be comprehensively obtained through the relationship between temperature and resistivity. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 Schematic diagram of the process of the metal melt resistivity test method for metal liquid structure identification of the present invention;

[0057] Figure 2 A schematic diagram of the process of dividing a metal melt into blocks to obtain at least one melt partition according to the present invention;

[0058] Figure 3 Schematic diagram of the process of obtaining the conversion function of conventional melt partition and characteristic melt partition of the present invention;

[0059] Figure 4 A schematic diagram of the process of training a convection simulation model according to the present invention;

[0060] Figure 5 A schematic diagram of the process of training a temperature correlation simulation model according to the present invention;

[0061] Figure 6 A schematic diagram of a flow chart of a superposition algorithm mechanism for forming a convection simulation model and a temperature correlation simulation model according to the present invention;

[0062] Figure 7 Schematic diagram of the process of establishing a temperature-resistivity relationship model according to the present invention. DETAILED DESCRIPTION

[0063] The following description is intended to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are merely examples, and those skilled in the art may conceive of other obvious variations.

[0064] Reference Figure 1 As shown, a metal melt resistivity testing method for identifying metal liquid structure includes:

[0065] Establish a three-dimensional coordinate system for the environment where the metal melt is located;

[0066] Divide the metal melt into blocks to obtain at least one melt partition, use the melt partition at the front of the upper left corner as a characteristic melt partition, use the remaining melt partitions as regular melt partitions, and obtain a conversion function between the regular melt partition and the characteristic melt partition;

[0067] A neural network model is formed for the characteristic melt partition. The neural network model consists of an input layer, a hidden layer, and an output layer. The input layer inputs the surface temperature of the characteristic melt partition and the coordinates of the temperature prediction position, and the output layer outputs the temperature of the temperature prediction position.

[0068] The hidden layer consists of a convection simulation model and a temperature correlation simulation model;

[0069] The convection simulation model is trained until the prediction error of the convection motion of the molten metal by the flow simulation model is less than a preset distance;

[0070] The temperature correlation simulation model is trained until the prediction error of the temperature correlation simulation model for the heat conduction of the metal melt is less than a preset temperature;

[0071] A superposition algorithm mechanism is formed for the convection simulation model and the temperature correlation simulation model, and the superposition algorithm mechanism converts the surface temperature of the characteristic melt partition and the coordinates of the temperature prediction position into the temperature of the temperature prediction position;

[0072] The temperature of the characteristic melt partition is measured and the temperature of the melt partition is obtained through the superposition algorithm mechanism;

[0073] A temperature-resistivity relationship model is established. Based on the temperature-resistivity relationship model, the resistivity of the melt partition is obtained. The resistance of the melt partition is averaged to obtain the resistivity of the metal melt.

[0074] When measuring, the metal melt is the medium for conducting the temperature of the metal melt. Therefore, the temperature at different positions in the metal melt is different, and the internal metal melt is difficult to measure directly due to obstruction. Since the metal melt is fluid, it is difficult to expose the internal metal melt by peeling. In the peeling process, convection will occur inside the metal melt, which will also cause the internal temperature to be exchanged with the outside world, thereby also causing changes in the overall temperature. These changes are caused by traditional measurements. Therefore, these problems need to be avoided. In this solution, the mentioned problems are solved in a targeted manner.

[0075] Reference Figure 2 As shown, dividing the metal melt into blocks to obtain at least one melt partition includes the following steps:

[0076] The test obtains a preset volume, and satisfies that the fluctuation range of the resistivity within the preset volume is less than the allowable error of the test;

[0077] The metal melt is evenly divided into blocks to obtain at least one melt partition, and the volume of the melt partition is equal to the preset volume.

[0078] Melt partitions are areas in the metal melt where different temperature predictions can be made. Therefore, by making different predictions of the melt partitions and based on the superposition effect of different conditions, the temperatures at different locations of the metal melt can be predicted more accurately.

[0079] Reference Figure 3 As shown, obtaining the conversion function between conventional melt partition and characteristic melt partition includes the following steps:

[0080] The melt partition located at the center of the metal melt is used as the target melt partition;

[0081] Obtain the center coordinates of the characteristic melt partition as the characteristic coordinates, and obtain the center coordinates of the conventional melt partition as the conventional coordinates;

[0082] Calculate the distance between the characteristic coordinate and the target melt partition to obtain the characteristic distance, calculate the distance between the conventional coordinate and the target melt partition to obtain the conventional distance, and divide the conventional distance by the characteristic distance to obtain the conversion coefficient;

[0083] The conventional coordinates and the conversion coefficients are paired and fitted to obtain a conversion function, where the conventional coordinates are the independent variables and the conversion coefficients are the dependent variables.

[0084] The conversion coefficient is mainly used to predict the temperature loss, which mainly occurs between the metal melt and the environment. Therefore, the loss is related to the distance to the center of the metal melt. The closer, the less the loss. Therefore, a conversion function is established.

[0085] Reference Figure 4 As shown in Figure 2, training the convection simulation model includes the following steps:

[0086] Obtaining a first displacement range of molecular motion in the metal melt, dividing the first displacement range of molecular motion into equal intervals to obtain at least one first displacement point;

[0087] In a non-viscous environment, a second displacement range of a single molecule motion of the metal melt is obtained, and the second displacement range of the single molecule motion is divided into equal intervals to obtain at least one second displacement point, wherein the number of the second displacement points is the same as that of the first displacement points, the second displacement points and the first displacement points are both vectors, and the size of the second displacement points relative to the first displacement points is their modulus.

[0088] The first displacement points are numbered from small to large, and the second displacement points are numbered from small to large. The viscosity resistance value is obtained by subtracting the value of the second displacement point with the same number from the first displacement point.

[0089] The second displacement point is paired with the viscous barrier value and fitted to obtain the viscous barrier function, which is a vector;

[0090] In a non-viscous environment, the displacement of a single molecule of the metal melt is equal to the ambient temperature of the second displacement point;

[0091] Pair and fit the ambient temperature and motion displacement to obtain the thermal motion function, which is a vector;

[0092] A convection simulation model is formed, and the convection simulation model includes aP(Q(k))+bQ(k), wherein a is a first training coefficient, k is temperature, b is a second training coefficient, Q(k) is a thermal motion function, and P(Q(k)) is a viscous barrier function;

[0093] Obtaining a first data set trained by a user, where the first data set consists of temperature and convective displacement caused by the temperature;

[0094] The temperature in the first data set is input into the convection simulation model. When the difference between the displacement output by the convection simulation model and the convection displacement caused by the temperature in the first data set is not less than a preset distance, a and b are adjusted until the difference between the displacement output by the convection simulation model and the convection displacement caused by the temperature in the first data set is less than a preset distance. The preset distance is set based on the allowable error of the test.

[0095] The output of the convection simulation model is a displacement vector, which can be represented by coordinates and can therefore also be viewed as a coordinate. In the convection process, the main factor affecting the thermal diffusion of molecules is the temperature. However, since the metal melt itself may have viscosity, this viscosity will hinder the movement of molecules. Therefore, it is necessary to comprehensively consider these two situations. Therefore, the convection simulation model includes aP(Q(k))+bQ(k). At the same time, since the initial establishment of the model may not produce sufficient simulation results, data training is required, and a and b are adjusted based on the results of the training. a and b can be ranged, and within the range, a sufficient number of combinations of a and b are taken to select a and b with the best simulation effect, thereby obtaining the final convection simulation model. The temperature-related simulation model is also processed similarly.

[0096] Reference Figure 5 As shown in Figure 2, training the temperature-related simulation model includes the following steps:

[0097] Obtaining the temperature variation range of the characteristic melt partition, dividing the temperature variation range into equal intervals, and obtaining at least one test point;

[0098] Obtaining a distance range from the melt partition to the characteristic melt partition, dividing the distance range into equal intervals, and obtaining at least one sample point;

[0099] Randomly combine test points and sample points to obtain at least one test group;

[0100] Under the conditions of the test group, obtain the temperature value of the melt partition;

[0101] The test points, sample points and the temperature values ​​at the melt partition are paired and fitted to obtain a temperature correlation function, wherein the test points and sample points are independent variables and the temperature value at the melt partition is the dependent variable;

[0102] A temperature correlation simulation model is formed. The temperature correlation simulation model is cL(x, y), where c is the training balance coefficient, L(x, y) is the temperature correlation function, x is the surface temperature of the characteristic melt partition, and y is the distance from the temperature prediction position to the characteristic melt partition;

[0103] Acquire a second data set, the second data set consisting of a surface temperature of a characteristic melt partition, a distance from a temperature prediction position to the characteristic melt partition, and a sample temperature at the temperature prediction position;

[0104] The surface temperature of the characteristic melt partition and the distance from the temperature prediction position to the characteristic melt partition in the second data set are input into the temperature correlation simulation model. When the difference between the test temperature output by the temperature correlation simulation model and the corresponding sample temperature is not less than the preset temperature, c is adjusted until the difference between the test temperature output by the temperature correlation simulation model and the corresponding sample temperature is less than the preset temperature. The preset temperature is the allowable error of the test.

[0105] In the same metal melt, the temperatures of different partitions have a heat conduction relationship with each other. Therefore, in the absence of external influences, there is a certain connection between them. Therefore, through this connection, a relationship function between the temperature of the characteristic melt partition and the remaining melt partitions is established, so that the temperature of the remaining melt partitions can be predicted through more accurate measurement of the characteristic melt partition. Here, since the characteristic melt partition is almost exposed to the outside and its volume is very small, its measurement error is very small and can be accurately measured.

[0106] Reference Figure 6 As shown, the superposition algorithm mechanism for the convection simulation model and the temperature correlation simulation model includes the following steps:

[0107] Inputting the surface temperature of the characteristic melt partition and the distance from the temperature prediction position to the characteristic melt partition into the temperature correlation simulation model to obtain the first temperature of the temperature prediction position;

[0108] From the equation w + aP (Q (cL (x, w))) + bQ (cL (x, w)) = y, solve for the compensation distance w;

[0109] The characteristic melt partition surface temperature and compensation distance are input into the temperature correlation simulation model to obtain the second temperature at the coordinate position;

[0110] The coordinates of the temperature prediction position are input into the conversion function to obtain the actual coefficient, and the average of the first temperature and the second temperature is multiplied by the actual coefficient to obtain the temperature of the temperature prediction position.

[0111] The temperature at the coordinates of the temperature prediction position mainly comes from two aspects. One is directly caused by heat exchange, namely temperature loss to the environment and internal temperature conduction, which can be obtained by the conversion function and the temperature correlation simulation model. The other is the temperature of the molecules moved to this position by convection. For this, it is necessary to determine the original distance from the molecules convectioned to this position to the characteristic melt partition, that is, the compensation distance. Therefore, the compensation distance is obtained by solving the equation, and the temperature of the molecules convectioned to this position can be approximated using the second temperature. The temperature of the temperature prediction position can then be obtained by superposition. Thus, the prediction can be completed, and the neural network model can predict the temperature of the specified position through the established functions and mechanisms.

[0112] Reference Figure 7 As shown in Figure 2, establishing a temperature-resistivity relationship model includes the following steps:

[0113] Obtaining a sample block, where the volume of the sample block is equal to a preset volume;

[0114] heating the sample block, and obtaining the resistivity of the sample block under the condition that the temperature of the sample block is equal to the value at the test point;

[0115] The values ​​at the identified points are paired with the resistivity of the sample block and fitted to obtain the resistivity prediction function.

[0116] Since the resistivity of molten metal is mainly determined by temperature when the material is determined, and the material of the characteristic melt partition is similar to that of the other melt partitions, the temperature of the characteristic melt partition includes the material factor, so we only need to consider the influence of temperature on resistivity. Therefore, a relationship model between temperature and resistivity is established to predict the resistivity of the melt partition.

[0117] Based on the relationship model between temperature and resistivity, obtaining the resistivity of the melt partition includes the following steps:

[0118] Substituting the temperature of the melt partition into the resistivity prediction function, the resistivity of the melt partition is obtained.

[0119] Furthermore, the present solution also proposes a storage medium on which a computer-readable program is stored. When the computer-readable program is called, the above-mentioned metal melt resistivity testing method for metal liquid structure identification is executed.

[0120] It is understandable that the storage medium may be a magnetic medium, such as a floppy disk, a hard disk, or a magnetic tape; an optical medium, such as a DVD; or a semiconductor medium, such as a solid state disk (SSD).

[0121] In summary, the advantages of the present invention are: by forming a neural network model, not only the temperature diffusion is predicted in the neural network model, but also the effect of fluid viscosity on fluid convection is predicted, and then the temperature of different positions in the metal melt can be predicted more accurately based on the temperature measurement of the characteristic melt partition, and then the overall resistivity of the metal melt can be comprehensively obtained through the relationship between temperature and resistivity.

[0122] The above shows and describes the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions merely illustrate the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A method for testing the resistivity of a molten metal for identifying the structure of a molten metal, characterized in that: include: Establish a three-dimensional coordinate system for the environment where the metal melt is located; Divide the metal melt into blocks to obtain at least one melt partition, use the melt partition at the front of the upper left corner as a characteristic melt partition, use the remaining melt partitions as regular melt partitions, and obtain a conversion function between the regular melt partition and the characteristic melt partition; A neural network model is formed for the characteristic melt partition. The neural network model consists of an input layer, a hidden layer, and an output layer. The input layer inputs the surface temperature of the characteristic melt partition and the coordinates of the temperature prediction position, and the output layer outputs the temperature of the temperature prediction position. The hidden layer consists of a convection simulation model and a temperature correlation simulation model; A convection simulation model is formed, and the convection simulation model includes aP(Q(k))+bQ(k); wherein a is a first training coefficient, k is temperature, b is a second training coefficient, Q(k) is a thermal motion function, and P(Q(k)) is a viscous barrier function; Obtaining a first data set trained by a user, where the first data set consists of temperature and convective displacement caused by the temperature; A temperature correlation simulation model is formed, where cL(x, y) is the training equilibrium coefficient, L(x, y) is the temperature correlation function, x is the surface temperature of the characteristic melt partition, and y is the distance from the temperature prediction position to the characteristic melt partition; Acquire a second data set, the second data set consisting of a surface temperature of a characteristic melt partition, a distance from a temperature prediction position to the characteristic melt partition, and a sample temperature at the temperature prediction position; A superposition algorithm mechanism is formed for the convection simulation model and the temperature correlation simulation model, and the superposition algorithm mechanism converts the surface temperature of the characteristic melt partition and the coordinates of the temperature prediction position into the temperature of the temperature prediction position; The temperature of the characteristic melt partition is measured and the temperature of the melt partition is obtained through the superposition algorithm mechanism; A temperature-resistivity relationship model is established. Based on the temperature-resistivity relationship model, the resistivity of the melt partition is obtained. The resistance of the melt partition is averaged to obtain the resistivity of the metal melt.

2. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 1, wherein: The step of dividing the metal melt into blocks to obtain at least one melt partition comprises the following steps: The test obtains a preset volume, and satisfies that the fluctuation range of the resistivity within the preset volume is less than the allowable error of the test; The metal melt is evenly divided into blocks to obtain at least one melt partition, and the volume of the melt partition is equal to the preset volume.

3. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 2, wherein: The conversion function of obtaining conventional melt partitions and characteristic melt partitions comprises the following steps: The melt partition located at the center of the metal melt is used as the target melt partition; Obtain the center coordinates of the characteristic melt partition as the characteristic coordinates, and obtain the center coordinates of the conventional melt partition as the conventional coordinates; Calculate the distance between the characteristic coordinate and the target melt partition to obtain the characteristic distance, calculate the distance between the conventional coordinate and the target melt partition to obtain the conventional distance, and divide the conventional distance by the characteristic distance to obtain the conversion coefficient; The conventional coordinates and the conversion coefficients are paired and fitted to obtain a conversion function, where the conventional coordinates are the independent variables and the conversion coefficients are the dependent variables.

4. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 3, wherein: Training the convection simulation model includes the following steps: Obtaining a first displacement range of molecular motion in the metal melt, dividing the first displacement range of molecular motion into equal intervals to obtain at least one first displacement point; In a non-viscous environment, a second displacement range of a single molecule motion of the metal melt is obtained, and the second displacement range of the single molecule motion is divided into equal intervals to obtain at least one second displacement point, wherein the number of the second displacement points is the same as that of the first displacement points, the second displacement points and the first displacement points are both vectors, and the size of the second displacement points relative to the first displacement points is their modulus. The first displacement points are numbered from small to large, and the second displacement points are numbered from small to large. The viscosity resistance value is obtained by subtracting the value of the second displacement point with the same number from the first displacement point. The second displacement point is paired with the viscous barrier value and fitted to obtain the viscous barrier function, which is a vector; In a non-viscous environment, the displacement of a single molecule of the metal melt is equal to the ambient temperature of the second displacement point; Pair and fit the ambient temperature and motion displacement to obtain the thermal motion function, which is a vector; The temperature in the first data set is input into the convection simulation model. When the difference between the displacement output by the convection simulation model and the convection displacement caused by the temperature in the first data set is not less than a preset distance, a and b are adjusted until the difference between the displacement output by the convection simulation model and the convection displacement caused by the temperature in the first data set is less than a preset distance. The preset distance is set based on the allowable error of the test.

5. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 4, wherein: Training the temperature-dependent simulation model includes the following steps: Obtaining the temperature variation range of the characteristic melt partition, dividing the temperature variation range into equal intervals, and obtaining at least one test point; Obtaining a distance range from the melt partition to the characteristic melt partition, dividing the distance range into equal intervals, and obtaining at least one sample point; Randomly combine test points and sample points to obtain at least one test group; Under the conditions of the test group, obtain the temperature value of the melt partition; The test points, sample points and the temperature values ​​at the melt partition are paired and fitted to obtain a temperature correlation function, wherein the test points and sample points are independent variables and the temperature value at the melt partition is the dependent variable; The surface temperature of the characteristic melt partition and the distance from the temperature prediction position to the characteristic melt partition in the second data set are input into the temperature correlation simulation model. When the difference between the test temperature output by the temperature correlation simulation model and the corresponding sample temperature is not less than the preset temperature, c is adjusted until the difference between the test temperature output by the temperature correlation simulation model and the corresponding sample temperature is less than the preset temperature. The preset temperature is the allowable error of the test.

6. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 5, characterized in that: The superposition algorithm mechanism for the convection simulation model and the temperature correlation simulation model comprises the following steps: Inputting the surface temperature of the characteristic melt partition and the distance from the temperature prediction position to the characteristic melt partition into the temperature correlation simulation model to obtain the first temperature of the temperature prediction position; From the equation w + aP(Q(cL(x, w))) + bQ(cL(x, w)) = y, solve for the compensation distance w; The characteristic melt partition surface temperature and compensation distance are input into the temperature correlation simulation model to obtain the second temperature at the coordinate position; The coordinates of the temperature prediction position are input into the conversion function to obtain the actual coefficient, and the average of the first temperature and the second temperature is multiplied by the actual coefficient to obtain the temperature of the temperature prediction position.

7. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 6, wherein: The establishment of the relationship model between temperature and resistivity comprises the following steps: Obtaining a sample block, where the volume of the sample block is equal to a preset volume; heating the sample block, and obtaining the resistivity of the sample block under the condition that the temperature of the sample block is equal to the value at the test point; The values ​​at the identified points are paired with the resistivity of the sample block and fitted to obtain the resistivity prediction function.

8. The method for testing the resistivity of a molten metal for identifying the structure of a molten metal according to claim 7, wherein: The method of obtaining the resistivity of the melt partition based on the relationship model between temperature and resistivity comprises the following steps: Substituting the temperature of the melt partition into the resistivity prediction function, the resistivity of the melt partition is obtained.

Images