Method for determining the end of the reduction period in a copper smelting anode furnace

Abstract

The application relates to a method for judging the end point of a reduction period of anode furnace copper smelting. The method comprises the following steps: acquiring surface scanning image data and cross-section image data of copper samples at different stages of the reduction period; comparing surface data of a to-be-detected reduction period copper sample with a reduction period copper sample feature database to determine whether the to-be-detected reduction period copper sample is at the end point of the reduction period; if the to-be-detected reduction period copper sample is not at the end point of the reduction period, performing end point prediction of the reduction period on the to-be-detected reduction period copper sample through a reduction period end point GRNN prediction model to obtain a prediction result. The scheme provided by the application can extract information from surface and cross-section images of the reduction period copper sample, judge whether the copper sample is at the end point of the reduction period through a cosine similarity algorithm, and predict the end point of the reduction period of the copper sample through a GRNN prediction model, so that the accuracy of judging the end point of the reduction period of the anode furnace copper smelting is effectively improved.

Description

Technical Field

[0001] This application relates to the field of copper smelting technology, and in particular to a method for determining the end point of the reduction period in copper smelting using an anode furnace. Background Technology

[0002] With the rapid development of modern industry, the demand for high-purity copper is gradually increasing. Anode furnace pyrometallurgical copper refining, as a heat treatment technology, is based on the principle of using high-temperature heat treatment processes to remove impurities and harmful elements such as residual sulfur from the copper anode through reactions such as oxidation, reduction, and volatilization, thereby obtaining high-purity copper.

[0003] In the pyrometallurgical copper refining process using an anode furnace, endpoint determination is a crucial step. Accurately identifying the furnace's endpoint ensures production efficiency and product quality, while also preventing waste and losses during production. Due to the unique characteristics of the pyrometallurgical copper refining process, three main methods are commonly used to determine the endpoint of the reduction period: visual judgment, temperature step method, and current-voltage change method. Visual judgment is an intuitive method where operators observe visual indicators such as the flame color, sparks, and the surface condition of the molten pool to determine the endpoint. At the endpoint of the reduction period, the molten pool surface typically exhibits a brighter color, and the flame becomes more stable. This method is simple and easy to implement, but it requires extensive experience and observational skills from the operator and is significantly influenced by subjective factors, posing a considerable risk of misjudgment. Furthermore, the accuracy of visual judgment is also affected by factors such as ambient light.

[0004] The temperature step method determines the end of the reduction period by monitoring temperature changes within the anode furnace. During reduction, the furnace temperature is continuously recorded. Typically, the temperature gradually rises during reduction, and a significant step change occurs as the temperature approaches the end of the reduction period. Observing this step change indicates that the reduction period has ended. However, the step changes observed by both the current-voltage variation method and the temperature step method are affected by other factors, such as the non-uniformity of temperature distribution within the furnace and changes in the reduction reaction rate, leading to uncertainty in the judgment results. Furthermore, both methods rely on the accuracy and stability of the monitoring equipment. Errors or malfunctions in the monitoring equipment can result in inaccurate judgments. In practical applications, operators need to make adjustments and judgments based on experience and observation, making the determination of the reduction period end dependent on human intervention, increasing subjectivity and operational complexity. Summary of the Invention

[0005] To address or partially address the problems existing in related technologies, this application provides a method for determining the endpoint of the reduction period in copper smelting in an anode furnace. This method can extract features from the surface and cross-sectional images of copper samples during the reduction period, construct a feature database of copper samples during the reduction period, compare the image data of the sample to be tested with the image data in the database, determine whether the copper sample is at the endpoint of the reduction period, and predict the time when the copper sample reaches the endpoint of the reduction period by using a GRNN prediction model for the endpoint of the reduction period. This method can effectively reduce the interference of environmental factors, improve the accuracy of determining the endpoint of the reduction period in copper smelting in an anode furnace, and requires no manual intervention, reducing the influence of human factors and improving the stability of the determination.

[0006] The first aspect of this application provides a method for determining the end point of the reduction period in copper smelting using an anode furnace, comprising the following steps:

[0007] Surface scanning image data and cross-sectional image data of copper samples at different stages of the reduction period were obtained. The surface scanning image data consisted of the mean gray-level difference, entropy, and contrast of the surface image, while the cross-sectional image data consisted of the mean gray-level difference, entropy, and contrast of the cross-sectional image.

[0008] The surface scan image data and cross-sectional image data of the copper sample to be tested during the reduction period are compared with the characteristic database of copper samples during the reduction period to determine whether the copper sample to be tested during the reduction period is at the end of the reduction period. The characteristic database of copper samples during the reduction period includes: images of copper samples at various stages of the reduction period, gray-scale difference mean, entropy, and contrast data of surface scan images and cross-sectional images of copper samples at different stages of the reduction period, and morphological characteristics and defect distribution of copper samples at different stages of the reduction period.

[0009] If the copper sample to be tested is not at the end of the reduction period, the reduction period end prediction model is used to predict the reduction period end of the copper sample to be tested and the prediction result is obtained.

[0010] The GRNN prediction model is trained using data from a database of copper samples in the reduction period. Image data of copper samples in the reduction period is selected from the database as the training set to train and optimize the GRNN prediction model. Image data of copper samples in the reduction period is selected from the database as the test set to validate the GRNN prediction model.

[0011] Optionally, the method for obtaining image information of the surface and cross-section of copper samples at different stages of the reduction period is to use a microscopic imaging device to obtain sufficient image information of the surface and cross-section of copper samples at different stages of the reduction period.

[0012] The copper sample image was processed using the image processing software ImageJ to convert it into a three-dimensional image, resulting in a scanned three-dimensional image and a cross-sectional three-dimensional morphology data image of the copper sample during the reduction period.

[0013] The three-dimensional images were analyzed to obtain the mean grayscale difference, entropy, and contrast of the surface and cross-section images of copper samples at different stages of the reduction period, and to determine the morphological characteristics and defect distribution of the surface and cross-section of copper samples at different stages of the reduction period.

[0014] Optionally, the image analysis and processing process of the copper sample surface and cross section at different stages of the reduction period is as follows: perform gray-level difference matrix operation on the obtained three-dimensional images of the copper sample surface and cross section during the reduction period, compare the images according to a certain pixel distance, and calculate the gray-level difference between each pair of pixels.

[0015] Then, using each grayscale difference value as an index, the frequency of occurrence of the corresponding pixel pair in the image is counted to construct a grayscale difference matrix;

[0016] Based on the gray-level difference matrix, the gray-level differences between pixels in the image are calculated to extract texture features. The mean of the gray-level difference, entropy, and contrast are extracted to describe the texture features of the image.

[0017] Feature extraction of surface and cross-sectional images of copper samples during the reduction period can effectively reduce interference from environmental factors by using gray-level difference matrix calculation.

[0018] Optionally, the process of constructing the characteristic database of copper samples during the reduction period is as follows: using the mean difference and contrast of the grayscale values ​​of all copper samples' images, two linear regression equations y(1) and y(2) are constructed, representing the linear regression equations before and after the reduction endpoint, respectively. Then, entropy is used to verify the linear regression equations, and the linear equations are:

[0019] y(1)=a*meam+(-b)-con (2)

[0020] y(2)=c*meam+(-d)-con (3)

[0021] In the formula, y(1) is the linear regression equation before the restoration endpoint, a is the linear regression coefficient of y(1), mean represents the mean of the image to be judged, b is the linear regression constant of y(1), con represents the contrast of the image to be judged, y(2) is the linear regression equation after the restoration endpoint, c is the linear regression coefficient of y(2), and d is the linear regression constant of y(2).

[0022] The mean gray-level difference and contrast obtained by performing gray-level difference matrix operations on all acquired copper sample images during the reduction period are substituted into two linear regression equations. The residuals of the actual values ​​and the fitted values ​​calculated by the linear equations are compared. The smaller the residual, the closer it is to the reduction endpoint. The stage of the copper sample is determined based on the residual. The images of the copper samples during the reduction period are classified according to their reduction stage. All classified images and image data of different stages of the reduction period are stored in the database.

[0023] Linear processing based on eigenvalues ​​is performed to determine image information for each stage of the reduction period, and a copper reduction period image database is constructed, providing data support for determining whether a copper sample is at the end of the reduction period.

[0024] Optionally, the process for determining whether the copper sample to be tested is at the end of the reduction period involves performing a gray-level difference matrix operation on the image of the copper sample to be tested to obtain the mean gray-level difference, entropy, and contrast of the image. The similarity is measured by calculating the cosine similarity between the features of the copper sample image to be tested and the features of the copper sample image in the database during the reduction period. When the cosine similarity exceeds the set similarity threshold, the two feature vectors are considered similar, thus determining the reduction period stage of the copper sample to be tested.

[0025] By using a cosine similarity algorithm to compare the image of the sample to be tested with the image information in the database, it is possible to determine whether the copper sample is at the end of the reduction period, which effectively improves the accuracy of determining the end of the reduction period in copper smelting in the anode furnace.

[0026] Optionally, the GRNN prediction model for the reduction period endpoint can be used to predict the reduction period endpoint of the copper sample to be tested. Specifically, this includes:

[0027] The feature information of the copper sample image that does not meet the reduction period endpoint is used as the input layer feature value of the GRNN neural network for verification and prediction. Based on the prediction result output by the GRNN neural network prediction model, the prediction result is the time required for the sample to reach the reduction period endpoint.

[0028] By using the GRNN prediction model to predict the reduction period endpoint for copper samples that have not reached the endpoint, the accuracy of judging the reduction period endpoint of copper smelting in anode furnace is effectively improved. This eliminates the need for manual intervention, reduces the impact of human factors, and enhances the stability of the prediction.

[0029] The technical solution provided in this application can include the following beneficial effects: This application extracts features from the surface and cross-sectional images of copper samples during the reduction period, constructs a feature database of copper samples during the reduction period, compares the image data of the sample to be tested with the image data in the database, determines whether the copper sample is at the end of the reduction period, and uses the GRNN prediction model for the end of the reduction period to predict the copper sample that has not reached the end of the reduction period, thereby determining the time when the copper sample reaches the end of the reduction period. This can effectively reduce the interference of environmental factors, improve the accuracy of the determination of the end of the reduction period in anode furnace copper smelting, and eliminate the need for manual intervention, thereby reducing the influence of human factors and improving the stability of the determination.

[0030] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and do not limit this application. Attached Figure Description

[0031] The above and other objects, features and advantages of this application will become more apparent from the more detailed description of exemplary embodiments thereof in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments thereof.

[0032] Figure 1 This is a schematic diagram of the method for determining the end point of the reduction period in copper smelting in an anode furnace, as shown in the embodiments of this application.

[0033] Figure 2 This is a schematic diagram of a preferred implementation process for determining the end point of the reduction period in copper smelting using an anode furnace, as shown in the embodiments of this application. Detailed Implementation

[0034] Embodiments of this application will now be described in more detail with reference to the accompanying drawings. While embodiments of this application are shown in the drawings, it should be understood that this application may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to make this application more thorough and complete, and to fully convey the scope of this application to those skilled in the art.

[0035] It should be understood that although the terms "first," "second," "third," etc., may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish information of the same type from one another. For example, without departing from the scope of this application, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of this application, "multiple" means two or more, unless otherwise explicitly specified.

[0036] In the description of this application, it should be understood that the terms "length", "width", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this application.

[0037] Unless otherwise expressly specified and limited, the terms "installation," "connection," "linking," and "fixing," etc., should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral part; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components or the interaction between two components. Those skilled in the art can understand the specific meaning of the above terms in this application according to the specific circumstances.

[0038] The technical solutions of the embodiments of this application are described in detail below with reference to the accompanying drawings.

[0039] Figure 1 This is a schematic flowchart illustrating the method for determining the end point of the reduction period in copper smelting using an anode furnace, as shown in an embodiment of this application. See also... Figure 1 A method for determining the end point of the reduction period in copper smelting using an anode furnace, comprising the following steps:

[0040] S101. Obtain surface scanning image data and cross-sectional image data of copper samples at different stages of the reduction period.

[0041] Optionally, the surface scan image data includes the mean gray-level difference, entropy, and contrast of the surface image, and the cross-sectional image data includes the mean gray-level difference, entropy, and contrast of the cross-sectional image.

[0042] Optionally, sufficient surface and cross-sectional image information of the copper sample at different stages of the reduction period can be obtained using a microscopic imaging device.

[0043] The copper sample image was processed using the ImageJ image processing software to convert it into a three-dimensional image, resulting in a scanned three-dimensional image and a cross-sectional three-dimensional morphology data image of the copper sample during the reduction period.

[0044] The three-dimensional images were analyzed to obtain the mean grayscale difference, entropy, and contrast of the surface and cross-section images of copper samples at different stages of the reduction period, and to determine the morphological characteristics and defect distribution of the surface and cross-section of copper samples at different stages of the reduction period.

[0045] Optionally, the obtained three-dimensional images of the surface and cross-section of the copper sample during the reduction period are subjected to gray-level difference matrix operation, and the images are compared pixel by pixel according to a certain pixel distance to calculate the gray-level difference between each pair of pixels.

[0046] Using each grayscale difference value as an index, the frequency of occurrence of the corresponding pixel pair in the image is counted to construct a grayscale difference matrix.

[0047] Based on the gray-level difference matrix, the gray-level differences between pixels in the image are calculated to extract texture features. The mean of the gray-level difference, entropy, and contrast are extracted to describe the texture features of the image.

[0048] S102. Compare the surface scan image data and cross-sectional image data of the copper sample to be tested during the reduction period with the characteristic database of copper samples during the reduction period to determine whether the copper sample to be tested during the reduction period is at the end of the reduction period.

[0049] Optionally, the copper sample feature database during the reduction period includes: images of copper samples at various stages of the reduction period, grayscale difference mean, entropy, and contrast data of surface scan images and cross-sectional images of copper samples at different stages of the reduction period, and morphological characteristics and defect distribution of copper samples at different stages of the reduction period.

[0050] Optionally, the process of constructing the feature database of copper samples during the reduction period is as follows: Using the mean grayscale difference and contrast of all copper samples' images, two linear regression equations y(1) and y(2) are constructed, representing the linear regression equations before and after the reduction endpoint, respectively. Entropy is then used to validate the linear regression equations, and the linear equations are:

[0051] y(1)=a*meam+(-b)-con (2)

[0052] y(2)=c*meam+(-d)-con (3)

[0053] In the formula, y(1) is the linear regression equation before the restoration endpoint, a is the linear regression coefficient of y(1), mean represents the mean of the image to be judged, b is the linear regression constant of y(1), con represents the contrast of the image to be judged, y(2) is the linear regression equation after the restoration endpoint, c is the linear regression coefficient of y(2), and d is the linear regression constant of y(2).

[0054] The mean gray-level difference and contrast obtained by performing gray-level difference matrix operations on all acquired copper sample images during the reduction period are substituted into these two linear regression equations. The residuals between the actual values ​​and the fitted values ​​calculated by the linear equations are compared. The smaller the residual, the closer it is to the reduction endpoint.

[0055] The stage of the copper sample is determined based on the residual. The images of the copper sample in the reduction period are classified according to the stage of the reduction period. All classified images of different stages of the reduction period and image data are stored in the database.

[0056] Optionally, the process for determining whether the copper sample to be tested is at the end of the reduction period includes: performing gray-level difference matrix operations on the image of the copper sample to be tested to obtain the mean gray-level difference, entropy, and contrast of the image.

[0057] Similarity is measured by calculating the cosine similarity between the image features of the copper sample to be tested and the image features of copper samples in the reduction period in the database. When the cosine similarity exceeds the set similarity threshold, the two feature vectors are considered similar, and the reduction period stage of the copper sample to be tested is determined.

[0058] S103. If the copper sample to be tested is not at the end of the reduction period, the end of the reduction period is predicted by the GRNN prediction model for the copper sample to be tested, and the prediction result is obtained.

[0059] Optionally, the GRNN prediction model is trained using data from the feature database of copper samples in the reduction period. The GRNN prediction model is then trained and optimized using image data of copper samples in the reduction period selected from the database as the training set. Finally, the GRNN prediction model is validated using image data of copper samples in the reduction period selected from the database as the test set.

[0060] Optionally, the GRNN prediction model for the reduction period endpoint specifically includes predicting the reduction period endpoint of the copper sample to be tested, including:

[0061] The feature information of the copper sample images that do not meet the reduction period endpoint is used as the input layer feature value of the GRNN neural network for verification and prediction.

[0062] Based on the prediction results output by the GRNN neural network prediction model, the prediction result is the time required for the sample to be tested to reach the end of the reduction period.

[0063] This application can extract features from the surface and cross-sectional images of copper samples during the reduction period, construct a feature database of copper samples during the reduction period, compare the image data of the sample to be tested with the image data in the database, determine whether the copper sample is at the end of the reduction period, and predict the time when the copper sample reaches the end of the reduction period by using the GRNN prediction model for the end of the reduction period. This can effectively reduce the interference of environmental factors, improve the accuracy of the determination of the end of the reduction period in anode furnace copper smelting, and eliminate the need for manual intervention, thereby reducing the influence of human factors and improving the stability of the determination.

[0064] Figure 2 This is a schematic diagram illustrating the implementation process of the method for determining the end point of the reduction period in copper smelting using an anode furnace, as shown in an embodiment of this application. See also... Figure 2 The implementation process of a method for determining the end point of the reduction period in copper smelting using an anode furnace is as follows:

[0065] Scanning images and cross-sectional morphology data of different regions of copper samples at different reduction stages were acquired. Sufficient surface and cross-sectional images of the copper samples at different stages of the reduction process were obtained using a microscopic imaging device. The microscopic imaging device must be a high-resolution color camera; a color camera acquires a large amount of image information, which is beneficial for image preprocessing and feature extraction. The copper sample images were processed using ImageJ image processing software to convert them into three-dimensional images, obtaining scanning images and cross-sectional morphology data of the anode copper. Analysis of the three-dimensional images revealed the surface and cross-sectional morphological characteristics and defect distribution of the anode copper plate.

[0066] Gray-level difference matrix operations were performed on the surface and cross-sectional images of copper samples during the reduction period. Linear processing of the image feature values ​​was then applied to preliminarily determine the stage of the copper sample's reduction period and construct a feature database for copper samples during the reduction period. Based on the scanned images and cross-sectional morphology data of different regions of the copper sample during the reduction period captured by a microscope camera, gray-level difference matrix operations were performed on the obtained three-dimensional images of the surface and cross-section of the copper sample during the reduction period. Pixel comparisons were performed at certain pixel distances, and the gray-level difference between each pair of pixels was calculated. Then, each gray-level difference value was used as an index to count the occurrence frequency of corresponding pixel pairs in the image, constructing a gray-level difference matrix. The formula for converting the image to a grayscale image is as follows:

[0067] ΔGLDM(i,j)=∑|g(i,j)-g(i+k,j+l)| (1)

[0068] In the formula, g(i,j) represents the gray value of pixel (i,j) in the image, k and l are the offsets of adjacent pixels, and GLDM(i,j) represents the sum of the absolute values ​​of the gray value differences between all pixel pairs (i,j) and (i+k,j+l) that meet the conditions in the image.

[0069] Based on the gray-level difference matrix, texture features are extracted by statistically analyzing the gray-level differences between pixels in the image. Features such as the mean, variance, entropy, contrast, and correlation of the gray-level differences are calculated to describe the image's texture. The average gray-level difference is calculated as the average of all elements in the gray-level difference matrix, representing the overall degree of gray-level difference in the image. The gray-level difference entropy is calculated as the entropy value of all elements in the gray-level difference matrix, representing the texture complexity of the image. The contrast is calculated as the contrast of all elements in the gray-level difference matrix, representing the magnitude of gray-level differences in the image. The correlation is calculated as the correlation of all elements in the gray-level difference matrix, representing the smoothness of gray-level differences in the image.

[0070] Based on the mean grayscale difference and contrast of all obtained sample images, two linear regression equations y(1) and y(2) are constructed. y(1) and y(2) are the linear regression equations before and after the restoration endpoint, respectively. Entropy is then used to verify the linear regression equations, and their linear equations are as follows:

[0071] y(1)=a*meam+(-b)-con (2)

[0072] y(2)=c*meam+(-d)-con (3)

[0073] In the formula, y(1) is the linear regression equation before the restoration endpoint, a is the linear regression coefficient of y(1), mean represents the mean of the image to be judged, b is the linear regression constant of y(1), con represents the contrast of the image to be judged, y(2) is the linear regression equation after the restoration endpoint, c is the linear regression coefficient of y(2), and d is the linear regression constant of y(2).

[0074] The mean gray-level difference and contrast obtained by performing gray-level difference matrix operations on all acquired sample images are successively substituted into the two linear regression equations. The residuals between the actual values ​​and the fitted values ​​calculated by the linear equations are compared. The smaller the residual, the closer it is to the reduction endpoint. The stage of the copper sample in the reduction period is determined based on the residual, and finally the image feature information of the copper sample in each stage of the reduction period is obtained.

[0075] The process of constructing the feature database of copper samples during the reduction period is as follows: gray-level difference matrix operation and linear operation are performed on all images at different stages of the reduction period, and these processed images and image texture features are stored in the database.

[0076] The process for determining whether a copper sample is at the end of the reduction period involves performing a gray-level difference matrix operation on the image of the sample to be tested. The similarity is measured by calculating the cosine similarity between the features of the copper sample image and the features of copper sample images and the end-of-reduction period image in the database. If the cosine similarity exceeds a set similarity threshold, the two features are considered similar. The formula for calculating cosine similarity is:

[0077] similarity=(A·B) / (||A||*||B||) (4)

[0078] In the formula, A and B represent two feature vectors, ||A|| and ||B|| represent the magnitude of the vectors, and the cosine similarity ranges from [-1, 1], with the value closer to 1 indicating a higher similarity.

[0079] First, calculate the cosine similarity between the image features of the copper sample to be tested and the images and features of copper samples at the end of the reduction period in the database. Determine whether the copper sample in the reduction period is at the end of the reduction period. If it is not at the end of the reduction period, calculate the cosine similarity between the images and features of copper samples at other stages in the database to determine the reduction period stage of the copper sample to be tested.

[0080] A generalized regressive neural network (GRNN) prediction model for the reduction period endpoint is constructed. This model is used to predict the reduction period endpoint for samples that do not meet the expected reduction period endpoint. The images of the reduction period endpoints to be detected are used as input layer features of the GRNN. A database of n reduction period sample images is selected as the training set, with corresponding reduction period stages defined as stages 1, 2, ..., n (before the reduction period has arrived). The GRNN prediction model is trained and optimized. A database of m reduction period sample images is selected as the test set to validate the GRNN prediction model.

[0081] The feature information of the copper sample images that do not meet the reduction period endpoint is input into the GRNN neural network prediction model for verification and prediction. Based on the prediction results output by the GRNN neural network prediction model, the time required for the sample to reach the reduction period endpoint is determined.

Claims

1. A method for determining the end point of the reduction period in copper smelting using an anode furnace, characterized in that, include: Surface scanning image data and cross-sectional image data of copper samples at different stages of the reduction period were obtained, wherein the surface scanning image data consisted of the mean gray-level difference, entropy, and contrast of the surface image, and the cross-sectional image data consisted of the mean gray-level difference, entropy, and contrast of the cross-sectional image. The surface scan image data and cross-sectional image data of the copper sample to be tested during the reduction period are compared with the feature database of copper samples during the reduction period to determine whether the copper sample to be tested during the reduction period is at the end of the reduction period. The feature database of copper samples during the reduction period includes: images of copper samples at various stages of the reduction period, grayscale difference mean, entropy, and contrast data of surface scan images and cross-sectional images of copper samples at different stages of the reduction period, and morphological characteristics and defect distribution of copper samples at different stages of the reduction period. If the copper sample to be tested is not at the end of the reduction period, the end of the reduction period is predicted for the copper sample to be tested using the GRNN prediction model for the end of the reduction period, and the prediction result is the time required for the sample to be tested to reach the end of the reduction period. Specifically, the GRNN prediction model is trained using data from the reduction period copper sample feature database. The GRNN prediction model is then trained and optimized using image data of the reduction period copper samples selected from the database as the training set. Finally, the GRNN prediction model is validated using image data of the reduction period copper samples selected from the database as the test set.

2. The method according to claim 1, characterized in that, The process of acquiring image information of the surface and cross-section of copper samples at different stages of the reduction period is as follows: Sufficient images of the surface and cross-section of copper samples at different stages of the reduction process were obtained using microscopic imaging equipment. The copper sample image was processed using the image processing software ImageJ to convert it into a three-dimensional image, resulting in a scanned three-dimensional image and a cross-sectional three-dimensional morphology data image of the copper sample during the reduction period. The three-dimensional images were analyzed to obtain the mean grayscale difference, entropy, and contrast of the surface and cross-section images of copper samples at different stages of the reduction period, and to determine the morphological characteristics and defect distribution of the surface and cross-section of copper samples at different stages of the reduction period.

3. The method according to claim 2, characterized in that, The image analysis and processing procedure for the copper sample surface and cross-section at different stages of the reduction period is as follows: The obtained three-dimensional images of the surface and cross-section of the copper sample during the reduction period were subjected to gray-level difference matrix operation. The images were compared pixel by pixel according to a certain pixel distance, and the gray-level difference between each pair of pixels was calculated. Using each grayscale difference value as an index, the number of occurrences of the corresponding pixel pair in the image is counted to construct a grayscale difference matrix; Based on the gray-level difference matrix, the gray-level differences between pixels in the image are calculated to extract texture features. The mean of the gray-level difference, entropy, and contrast are extracted to describe the texture features of the image.

4. The method according to claim 1, characterized in that, The process of constructing the characteristic database of copper samples during the reduction period is as follows: Two linear regression equations, y(1) and y(2), were constructed using the mean difference and contrast of the grayscale differences of all copper samples. These equations represent the linear regression equations before and after the reduction endpoint, respectively. Entropy was then used to validate the linear regression equations. The expressions for the linear equations y(1) and y(2) are as follows: （2） （3） In the formula, y(1) is the linear regression equation before the restoration endpoint, a is the linear regression coefficient of y(1), mean represents the mean of the image to be judged, b is the linear regression constant of y(1), con represents the contrast of the image to be judged, y(2) is the linear regression equation after the restoration endpoint, c is the linear regression coefficient of y(2), and d is the linear regression constant of y(2). The mean gray-level difference and contrast obtained by performing gray-level difference matrix operation on all the copper sample images acquired during the reduction period are substituted into the two linear regression equations. The residuals of the actual values ​​and the fitted values ​​calculated by the linear equations are compared. The smaller the residual, the closer it is to the reduction endpoint. The stage of the copper sample is determined based on the residual. The images of the copper sample in the reduction period are classified according to the stage of the reduction period. All classified images of different stages of the reduction period and image data are stored in the database.

5. The method according to claim 1, characterized in that, The process for determining whether the copper sample to be tested is at the end of the reduction period includes: The gray-level difference matrix operation is performed on the image of the copper sample to be tested to obtain the mean gray-level difference, entropy, and contrast of the image. Similarity is measured by calculating the cosine similarity between the image features of the copper sample to be tested and the image features of copper samples in the reduction period in the database. When the cosine similarity exceeds the set similarity threshold, the two feature vectors are considered similar, and the reduction period stage of the copper sample to be tested is determined.

6. The method according to claim 1, characterized in that, The specific steps of predicting the reduction period endpoint of the copper sample to be tested using the GRNN prediction model for the reduction period endpoint include: The feature information of the copper sample image that does not meet the reduction period endpoint is used as the input layer feature value of the GRNN neural network for verification and prediction. Based on the prediction results output by the GRNN neural network prediction model, the prediction result is the time required for the sample to be tested to reach the end of the reduction period.

Images