A detection method and system for copper wire multi-pass rolling intermediate annealing

Abstract

The present invention relates to the technical field of annealing detection, and in particular to a detection method and system for copper wire multi-pass rolling intermediate annealing. The method comprises: collecting surface features of copper wires subjected to multi-pass annealing respectively, and constructing copper wire feature models according to a number of copper wire surface features respectively, acquiring copper wire association structures by structural mapping of a number of copper wire feature models respectively, establishing copper wire dependency networks according to a number of association structures respectively, outputting single defect information of each annealing according to a number of copper wire dependency networks, acquiring a single annealing deviation by establishing a dynamic feedback loop, integrating the single annealing deviation to output a final defect detection result, and effectively solving the deficiencies in multi-source data integration and dynamic updating in a traditional copper wire rolling annealing process through the present invention. The scheme can dynamically adjust annealing parameters based on single defect information, significantly improving the accuracy and response speed of annealing quality control, and enhancing the safety and efficiency of the entire annealing process.

Description

Technical Field

[0001] The present invention relates to the technical field of annealing detection, and in particular to a detection method and system for copper wire multi-pass rolling intermediate annealing. Background Art

[0002] In the multi-pass rolling process of copper wire, annealing is a key process step, which aims to improve the physical properties of copper wire. However, with the complexity of copper wire production process, traditional detection methods face some challenges in the highly dynamic production environment. The existing technology is insufficient in the real-time prediction and precise management of copper wire defects, especially in how to comprehensively utilize multiple data sources for defect detection and production process optimization.

[0003] Although defect detection technology based on machine learning and graph neural networks has developed in recent years and can achieve more accurate prediction and analysis, the existing technology still has certain limitations in the fusion of multi-dimensional data, the updating of dynamic models and the stability of application in actual processes. Especially in the process of multi-pass rolling and intermediate annealing, how to perform effective defect detection and annealing process optimization in real time and accurately based on the geometric shape, surface features and physical properties of the copper wire is still an important problem facing the current technology.

[0004] The information disclosed in this background technology section is only intended to deepen the understanding of the overall background technology of the present disclosure, and should not be regarded as acknowledging or suggesting in any form that the information constitutes the prior art known to those skilled in the art. Summary of the invention

[0005] The invention provides a method and system for detecting intermediate annealing of copper wire during multi-pass rolling, which can effectively solve the problems in the background technology.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for detecting copper wire multi-pass rolling intermediate annealing, the method comprising:

[0008] Collect copper wire surface features for multiple annealing processes respectively to obtain a number of copper wire surface features, and construct copper wire feature models based on the copper wire surface features, each of which corresponds to each annealing process one by one;

[0009] Structural mapping is performed on the copper wire feature models to generate a plurality of copper wire association structures, and a copper wire dependency network is constructed based on the plurality of copper wire association structures;

[0010] Outputting single copper wire defect information respectively according to the plurality of copper wire dependent networks;

[0011] A dynamic feedback loop is established, and single annealing deviations are determined based on the dynamic feedback loop and the single defect detection information, and the single annealing deviations are integrated to output a final defect detection result.

[0012] Furthermore, a plurality of copper wire feature models are subjected to structural mapping to generate a plurality of copper wire associated structures, including:

[0013] According to the copper wire feature model, a plurality of copper wire surface features are respectively converted into corresponding data anchor points, wherein the copper wire surface features include copper wire geometric properties and copper wire physical properties;

[0014] Establishing links between the data anchor points according to the correlation between the copper wire geometric characteristics and the copper wire physical characteristics, and assigning link weights based on the correlation using a mathematical optimization algorithm;

[0015] Constructing a dependency feature matrix based on the data anchor point, wherein each row of the dependency feature matrix represents the physical property of the copper wire, and each column represents a different data anchor point;

[0016] A dependent adjacency matrix is ​​established according to the geometric characteristics of the copper wire and the physical characteristics of the copper wire, wherein the rows and columns of the dependent adjacency matrix are represented as different data anchor points, and the data anchor points are linked to each other to represent 1, otherwise 0, and the adjacency matrix is ​​filled;

[0017] A copper wire dependency network is constructed according to the dependency feature matrix, the dependency adjacency matrix and the link weights.

[0018] Further, outputting single copper wire defect information respectively according to the plurality of copper wire dependent networks includes:

[0019] An input layer, responsible for receiving the dependency feature matrix and the dependency adjacency matrix;

[0020] A first aggregation layer, according to the dependency adjacency matrix, uses a graph convolution algorithm to aggregate the data anchor points to obtain a first-level aggregation anchor point, wherein the aggregation is used to summarize feature information of the data anchor point and adjacent data anchor points;

[0021] The second aggregation layer performs secondary aggregation on the primary aggregation anchor point and outputs a secondary aggregation anchor point;

[0022] A fully connected layer maps the secondary aggregated anchor point data to a defect recognition output space to obtain single copper wire defect detection information;

[0023] The output layer outputs the single copper wire defect detection information.

[0024] Furthermore, a dynamic feedback loop is established, including:

[0025] Collect historical annealing detection parameters, define detection parameter thresholds based on the historical annealing detection parameters, and select a feedback control terminal based on annealing requirements;

[0026] Collecting real-time annealing detection data, using data analysis software to compare the real-time annealing detection data in combination with the detection parameter threshold, marking the real-time annealing detection data that exceeds the detection parameter threshold, and obtaining error feedback;

[0027] A closed-loop control logic is adopted to perform feedback control on the annealing detection according to the feedback error.

[0028] Furthermore, a plurality of the single annealing deviations are integrated to output a final defect detection result, including:

[0029] Extracting a single annealing deviation feature according to the single annealing deviation;

[0030] Collecting historical annealing defect information, and building a global deviation model based on the historical annealing defect information and the single annealing deviation characteristics, wherein the global deviation model uses the output of the upper-order annealing process as the input of the lower-order annealing process to predict the global cumulative deviation;

[0031] Feedback the global cumulative deviation according to the single annealing deviation, and modify the single annealing parameters;

[0032] All the single annealing deviations are integrated, the final annealing process is globally corrected, a cumulative correction result is obtained, the annealed copper wire is inspected based on the cumulative correction result, and a final defect detection result is output.

[0033] Furthermore, a plurality of the data anchor points are linked according to the correlation between the copper wire geometric characteristics and the copper wire physical characteristics, including:

[0034] Dividing historical annealing detection parameters into an annealing training group and an annealing test group, the historical annealing detection parameters including the copper wire geometric characteristics and the copper wire physical characteristics;

[0035] Establishing regression models for several physical properties of the copper wires respectively, training the regression models according to the annealing training group in combination with a regression algorithm, and evaluating them based on the annealing test group to obtain a mapping relationship between the physical properties of the copper wires and the geometric properties of the copper wires;

[0036] A link relationship between the copper wire geometric characteristics and the copper wire physical characteristics is established according to the mapping relationship.

[0037] Furthermore, the copper wire surface features are collected for each of the multiple annealing processes to obtain a number of copper wire surface features, and copper wire feature models are constructed based on the copper wire surface features, including:

[0038] The surface of the copper wire is geometrically scanned by remote sensing technology to obtain a number of point cloud geometric data, and the physical characteristics of the copper wire are synchronously collected by a detection device, and the point cloud geometric data are integrated with the physical characteristics of the copper wire to obtain a point cloud data set;

[0039] Performing data cleaning on the point cloud data set, and performing point cloud reconstruction on the cleaned point cloud data set to obtain a characteristic fusion surface, wherein the characteristic fusion surface includes the point cloud geometric data and the physical characteristics of the copper wire;

[0040] The characteristic fusion surface is smoothed, and the smoothed characteristic fusion surface is subjected to surface fitting to construct a copper wire characteristic model.

[0041] Further, performing point cloud reconstruction on the cleaned point cloud data set to obtain a feature fusion surface includes:

[0042] Constructing a copper wire feature coordinate system, making coordinate correspondence between a number of the point cloud data and the copper wire feature coordinate system, and mapping the physical properties of the copper wire with the point cloud data to construct a copper wire feature space, wherein the copper wire feature space includes all the point cloud data and the corresponding physical properties of the copper wire;

[0043] Inserting a plurality of the point cloud data into the copper wire characteristic space according to coordinates, segmenting the copper wire characteristic space according to the inserted point cloud data, and obtaining the copper wire segmentation space;

[0044] Constructing a circumscribed ball according to the copper wire cutting space, detecting whether there is the point cloud data other than the point cloud data for constructing the circumscribed ball in the circumscribed ball, and if so, flipping the copper wire segmentation space, and checking again whether there is the point cloud data other than the point cloud data for constructing the circumscribed ball, and if so, continuing to flip until there is no point cloud data other than the point cloud data for constructing the circumscribed ball, and obtaining a set of copper wire discrete points, wherein the circumscribed ball is the minimum circumscribed ball of the copper wire segmentation space;

[0045] A detection system for copper wire multi-pass rolling intermediate annealing, the system comprising:

[0046] A feature acquisition modeling module collects copper wire surface features for multiple annealing processes, obtains a number of copper wire surface features, and constructs copper wire feature models based on the copper wire surface features, each of which corresponds to each annealing process.

[0047] A dependency network construction module performs structural mapping on the copper wire feature models to generate a plurality of copper wire association structures, and constructs a copper wire dependency network based on the plurality of copper wire association structures;

[0048] A single defect output module, which outputs single copper wire defect information respectively according to the plurality of copper wire dependent networks;

[0049] The defect feedback integration module establishes a dynamic feedback loop, determines single annealing deviations based on the dynamic feedback loop and a plurality of single defect detection information, and integrates the single annealing deviations to output a final defect detection result.

[0050] Furthermore, the defect feedback integration module includes:

[0051] A deviation feature extraction unit, which extracts a single annealing deviation feature according to the single annealing deviation;

[0052] A global deviation model building unit collects historical annealing defect information, and builds a global deviation model based on the historical annealing defect information and the single annealing deviation characteristics, wherein the global deviation model uses the output of the upper-order annealing process as the input of the lower-order annealing process to predict the global cumulative deviation;

[0053] an annealing parameter correction unit, which feeds back the global cumulative deviation according to the single annealing deviation and corrects the single annealing parameter;

[0054] The detection result output unit integrates all the single annealing deviations, performs global correction on the final annealing process, obtains cumulative correction results, detects the annealed copper wire based on the cumulative correction results, and outputs the final defect detection result.

[0055] The technical solution of the present invention can achieve the following technical effects:

[0056] This solution effectively solves the deficiencies in multi-source data integration and dynamic updating in the traditional copper wire rolling annealing process. The solution can dynamically adjust annealing parameters based on single defect information, significantly improves the accuracy and response speed of annealing quality control, and enhances the safety and efficiency of the entire annealing process.

[0057] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 It is a flow chart of a method for detecting copper wire multi-pass rolling intermediate annealing;

[0060] Figure 2 Generate a flow diagram for the copper wire association structure;

[0061] Figure 3 It is a schematic diagram of the multi-pass annealing defect processing process;

[0062] Figure 4 This is the structural diagram of the detection system for intermediate annealing of copper wire during multi-pass rolling. DETAILED DESCRIPTION

[0063] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0064] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0065] Embodiment 1

[0066] like Figure 1 As shown, the present application provides a method for detecting copper wire multi-pass rolling intermediate annealing, the method comprising:

[0067] S10: collecting copper wire surface features for multiple annealing processes respectively, obtaining a number of copper wire surface features, and constructing copper wire feature models based on the several copper wire surface features, wherein each copper wire feature model corresponds to each annealing process one by one;

[0068] S20: Structural mapping is performed on a plurality of copper wire feature models to generate a plurality of copper wire association structures, and a copper wire dependency network is constructed based on the plurality of copper wire association structures;

[0069] S30: Outputting single copper wire defect information respectively according to a plurality of copper wire dependency networks;

[0070] S40: establishing a dynamic feedback loop, determining single annealing deviations based on the dynamic feedback loop and a number of single defect detection information, integrating the single annealing deviations to output a final defect detection result.

[0071] Specifically, first, in the multi-pass annealing process, high-precision scanning equipment such as laser scanners or electron microscopes can be used to carefully scan the surface of the copper wire to obtain the surface roughness, microstructure, color change and other surface features of the copper wire. According to the collected copper wire surface features, the data of the copper wire surface can be converted into a mathematical model that can represent the characteristics of each annealing process through a model analysis algorithm, and then the copper wire feature model is conceptually transformed. This transformation can be based on general data structure theory to simplify the model into basic elements and connection relationships, and then a copper wire dependency network is constructed based on the associated structure. The copper wire dependency network reveals the dependency relationship between the copper wire characteristics during the annealing process. Subsequently, the copper wire dependency network is used to analyze possible defects in each annealing stage. For example, network monitoring can be used to determine possible defect abnormal patterns in the network. Based on the analysis results, each annealing stage will output the corresponding copper wire single defect information, and then a control algorithm such as PID control is used to establish a dynamic feedback loop, and the single annealing deviation is determined based on the single defect identification information. Then, the annealing process is adjusted in real time based on the single feedback deviation, and the final defect detection result is output by integrating the single annealing deviation.

[0072] Through the technical solution of the present invention, the deficiencies in multi-source data integration and dynamic updating in the traditional copper wire rolling annealing process are effectively solved. This solution can dynamically adjust annealing parameters based on single defect information, significantly improves the accuracy and response speed of annealing quality control, and enhances the safety and efficiency of the entire annealing process.

[0073] Further, if Figure 2 As shown, several copper wire feature models are structurally mapped to generate several copper wire associated structures, including:

[0074] S21: converting a number of copper wire surface features into corresponding data anchor points according to the copper wire feature model, where the copper wire surface features include copper wire geometric properties and copper wire physical properties;

[0075] S22: establishing links between a number of data anchor points according to the correlation between the copper wire geometric characteristics and the copper wire physical characteristics, and assigning link weights based on the correlation using a mathematical optimization algorithm;

[0076] S23: construct a dependent feature matrix based on the data anchor points, where each row of the dependent feature matrix represents the physical characteristics of the copper wire and each column represents a different data anchor point;

[0077] S24: establishing a dependent adjacency matrix according to the geometric characteristics and physical characteristics of the copper wire, wherein the rows and columns of the dependent adjacency matrix are represented as different data anchor points, and the data anchor points are linked to each other to represent 1, otherwise 0, and the adjacency matrix is ​​filled;

[0078] S25: Construct a copper wire dependency network based on the dependency feature matrix, the dependency adjacency matrix, and the link weights.

[0079] As a preferred embodiment of the above embodiment, the feature points of the copper wire feature model are first converted into data anchor points, which include the geometric and physical properties of the copper wire, such as length, diameter, resistivity and hardness. Mathematical optimization algorithms, such as linear programming or network flow optimization, are used to analyze the relationship and influence between different data anchor points. This analysis is based on the correlation between the physical and geometric properties of the copper wire, so as to determine the weight of each link, ensuring that the weight can accurately reflect the degree of dependence between the two data anchor points. Then, a dependency feature matrix is ​​constructed, which takes the physical properties of the data anchor points as rows and each data anchor point as columns. This matrix is ​​used to reveal the complex relationship between different physical properties, and based on the established links and weights, an adjacency matrix is ​​formed, each row and each column represents a data anchor point, and then different data anchor points are judged to determine whether there is a link relationship between the two data anchor points. If there is a link between the two data anchor points, the corresponding matrix element is 1, otherwise it is 0. Finally, the dependency feature matrix, the dependency adjacency matrix and the link weight are combined, and a network construction technology, such as the network flow algorithm in network theory, is used to construct a complete copper wire dependency network.

[0080] Furthermore, the single copper wire defect information is outputted respectively according to several copper wire dependent networks, including:

[0081] The input layer is responsible for receiving the dependency feature matrix and the dependency adjacency matrix;

[0082] The first aggregation layer uses the graph convolution algorithm to aggregate the data anchor points according to the dependency adjacency matrix to obtain the first-level aggregation anchor points. The aggregation is used to summarize the feature information of the data anchor points and the adjacent data anchor points.

[0083] The second aggregation layer performs secondary aggregation on the first-level aggregation anchor points and outputs secondary aggregation anchor points;

[0084] The fully connected layer maps the secondary aggregated anchor point data to the defect recognition output space to obtain the single copper wire defect detection information;

[0085] Output layer, outputs single copper wire defect detection information.

[0086] In this embodiment, first, the system sets an input layer, which is used to receive the dependency feature matrix and the dependency adjacency matrix generated by the copper wire dependency network. These matrices contain detailed physical and geometric characteristic information of the copper wire at each stage. Then, a first aggregation layer is constructed. In this layer, a graph convolution algorithm is used to process data anchor points. The graph convolution algorithm can aggregate the feature information of each data anchor point and its adjacent anchor points according to the dependency adjacency matrix. This process generates a first-level aggregation anchor point. Each first-level aggregation anchor point is a new data point that integrates the characteristics of adjacent data anchor points. Then, a second aggregation layer is constructed to further perform a secondary aggregation operation on the first-level aggregation anchor point. Through the processing of this layer, the characteristics of the copper wire can be further integrated and abstracted to generate a second-level aggregation anchor point. Then, a fully connected layer is constructed. The fully connected layer is connected through a deep network to convert the aggregated data points into specific information that can indicate potential defects. Finally, the single copper wire defect detection information is presented through the output layer. This information includes the possible defect type, location and severity of the copper wire.

[0087] Going further, a dynamic feedback loop is established, including:

[0088] Collect historical annealing detection parameters, define detection parameter thresholds based on the historical annealing detection parameters, and select a feedback control terminal based on annealing requirements;

[0089] Collect real-time annealing detection data, use data analysis software combined with detection parameter thresholds to compare the real-time annealing detection data, mark the real-time annealing detection data that exceeds the detection parameter threshold, and obtain error feedback;

[0090] A closed-loop control logic is adopted to perform feedback control on the annealing detection according to the feedback error.

[0091] Specifically, first, the system will collect relevant detection parameters from the historical annealing process. These parameters may include the temperature, speed, pressure, etc. of the copper wire. Based on these historical data, the normal operating range or threshold of each parameter is defined through statistical analysis methods. At the same time, according to the specific annealing requirements, the appropriate feedback control terminal is selected, and then real-time monitoring during the annealing process is used to obtain real-time data, which is compared with the pre-set threshold. The data is processed using dedicated data analysis software such as PI System. Once any parameter exceeding the threshold is detected, the system will automatically calculate the error feedback between the set threshold. Then, closed-loop control logic can be used to adjust the annealing process to ensure that each parameter returns to its normal range. It should be noted that the thresholds need to be evaluated and adjusted regularly to ensure that they remain relevant and accurate as production conditions change.

[0092] Furthermore, if Figure 3 As shown in the figure, several single annealing deviations are integrated to output the final defect detection results, including:

[0093] S41: extracting a single annealing deviation feature according to the single annealing deviation;

[0094] S42: collecting historical annealing defect information, and building a global deviation model based on the historical annealing defect information and single annealing deviation characteristics. The global deviation model uses the output of the upper-order annealing process as the input of the lower-order annealing process to predict the global cumulative deviation.

[0095] S43: providing feedback on the global cumulative deviation according to the single annealing deviation, and correcting the single annealing parameters;

[0096] S44: Integrate all single annealing deviations, perform global correction on the final annealing process, obtain cumulative correction results, detect the annealed copper wire based on the cumulative correction results, and output the final defect detection results.

[0097] As a preferred embodiment of the above embodiment, first, the output of each annealing stage is deeply analyzed to accurately identify the annealing deviation features that are directly related to the product quality. These features include not only the parameters of a single stage, but also the interactions and influences between stages, such as changes in factors such as temperature gradient and cooling rate. Historical data and real-time feedback data are used to build a dynamic global deviation model, which not only predicts the deviation of a single annealing process, but also can analyze the mutual influence and cumulative effect between various annealing stages. This model provides a comprehensive annealing process prediction by integrating data from various stages. Based on the prediction of the global deviation model, cross-stage parameter adjustment is implemented, which includes not only adjusting the parameters of the upcoming annealing stage, but also adjusting the parameters of all subsequent stages according to the actual performance of the previous stage. Finally, the entire annealing process is globally corrected by integrating all identified single annealing deviations. The integration takes into account the specific deviations of each annealing stage and merges them to form a comprehensive cumulative correction framework. The accumulated correction results are then used to perform a comprehensive quality inspection of the annealed copper wire and output the final defect detection results.

[0098] Furthermore, several data anchor points are linked based on the correlation between the copper wire geometric characteristics and the copper wire physical characteristics, including:

[0099] Dividing historical annealing detection parameters into an annealing training group and an annealing test group, the historical annealing detection parameters include copper wire geometric properties and copper wire physical properties;

[0100] Regression models are established for several physical properties of copper wires, and the regression models are trained based on the annealing training group and the regression algorithm, and evaluated based on the annealing test group to obtain the mapping relationship between the physical properties of copper wires and the geometric properties of copper wires;

[0101] A link relationship between the copper wire geometric characteristics and the copper wire physical characteristics is established based on the mapping relationship.

[0102] In this embodiment, first, according to the collected historical annealing detection parameters, the geometric properties and physical property data of the copper wire are extracted as annealing detection parameters. For example, the geometric properties may include diameter change, surface roughness, bending degree, etc. of the copper wire, and the physical properties may include hardness, conductivity, grain size, etc. These data are divided into two groups, an annealing training group and an annealing test group. Regression models are established for the extracted physical properties of the copper wire respectively, and a mapping relationship between the physical properties and the geometric properties is constructed. The establishment of the regression model can adopt, for example, linear regression, support vector regression, or a regression model based on a gradient boosting tree. By selecting the physical properties of the copper wire and the related geometric properties as prediction variables and input variables, respectively, the regression model is trained using the annealing training group, and the model parameters are optimized so that the input geometric properties can accurately predict the physical properties. The model is verified based on the annealing test group, and the prediction error, such as the mean square error, is calculated to evaluate the accuracy and robustness of the model, thereby obtaining the mapping relationship between the physical properties and the geometric properties of the copper wire. Based on the above mapping relationship, a link is established between the physical properties and the geometric properties of the copper wire to obtain the link relationship between the physical properties and the geometric properties of the copper wire.

[0103] Furthermore, the copper wire surface features are collected for each of the multiple annealing processes to obtain a number of copper wire surface features, and copper wire feature models are constructed based on the several copper wire surface features, including:

[0104] Remote sensing technology is used to perform geometric scanning on the surface of the copper wire to obtain a number of point cloud geometric data, and the physical characteristics of the copper wire are synchronously collected through the detection equipment, and a number of point cloud geometric data and physical characteristics of the copper wire are integrated to obtain a point cloud data set;

[0105] Clean the point cloud data set, reconstruct the cleaned point cloud data set, and obtain a feature fusion surface, which includes point cloud geometric data and physical characteristics of the copper wire;

[0106] The characteristic fusion surface is smoothed, and surface fitting is performed on the smoothed characteristic fusion surface to construct a copper wire characteristic model.

[0107] Specifically, first, remote sensing technology is used to perform geometric scanning of the copper wire, such as using laser scanning or optical scanning technology to obtain point cloud geometric data of the copper wire surface. At the same time, other testing equipment, such as hardness testers and conductivity meters, are used to synchronously collect physical properties of the copper wire, such as hardness, conductivity and material composition. The collected point cloud geometric data is then integrated with the physical property data of the copper wire, and the data is formatted, and the physical property data is associated with the corresponding geometric point cloud position to form a point cloud data set containing comprehensive characteristic information. The integrated point cloud data set is cleaned to remove noise and non-related data. After cleaning, the point cloud reconstruction technology in computer graphics is used to generate a characteristic fusion surface representing the copper wire surface. The surface not only contains geometric information, but also integrates physical properties, such as the conductivity distribution or hardness distribution map of the surface. The characteristic fusion surface is smoothed to remove jagged or sharp edges that may be generated during the reconstruction process. After smoothing, surface fitting technology, such as Bezier surface technology, is used to construct a characteristic model of the copper wire.

[0108] Furthermore, the cleaned point cloud dataset is reconstructed to obtain a feature fusion surface, including:

[0109] Construct a copper wire feature coordinate system, match some point cloud data with the copper wire feature coordinate system, and map the physical properties of the copper wire with the point cloud data to construct a copper wire feature space, which contains all point cloud data and the corresponding physical properties of the copper wire;

[0110] Inserting a number of point cloud data into the copper wire characteristic space according to coordinates, segmenting the copper wire characteristic space according to the inserted point cloud data, and obtaining the copper wire segmentation space;

[0111] According to the copper wire cutting space, an external sphere is constructed, and it is detected whether there is point cloud data other than the external sphere in the external sphere. If so, the copper wire segmentation space is flipped, and it is checked again whether there is point cloud data other than the external sphere. If so, the flipping is continued until there is no point cloud data other than the external sphere, and a set of copper wire discrete points is obtained, and the external sphere is the minimum external sphere of the copper wire segmentation space;

[0112] Repeat the insertion and flipping operations of the point cloud data to obtain a number of copper wire discrete points, and generate a feature fusion surface based on the copper wire discrete points.

[0113] As a preferred embodiment of the above, first, a copper wire feature coordinate system is established, and the cleaned point cloud data is matched according to the copper wire feature coordinate system, and the physical characteristics of the copper wire are mapped to the corresponding point cloud data. In this way, a copper wire feature space containing complete geometric and physical information is formed, and then the point cloud data is inserted into the copper wire feature space according to its coordinate values, and the feature space is segmented based on these data. After segmentation, a circumscribed sphere is constructed for each segmented area to detect whether there is point cloud data outside the circumscribed sphere. If it is found that there is external point cloud data, a flip operation is performed, that is, the position of the circumscribed sphere is adjusted. The position or size is changed to ensure that all relevant point cloud data are included. If there is still point cloud data outside the circumscribed sphere after flipping, the flipping operation is continued until all point cloud data are included in a minimum circumscribed sphere. In this way, a discrete point set of the copper wire can be obtained. These discrete points accurately reflect the geometric and physical properties of the copper wire. The above insertion and flipping operations are repeated for different data segments, and finally a series of discrete points of the copper wire are obtained. Based on these discrete points, the three-dimensional reconstruction technology is used to generate a feature fusion surface. The surface comprehensively expresses the geometric shape and physical properties of the copper wire, providing an accurate basis for subsequent defect detection and quality assessment.

[0114] Embodiment 2

[0115] Based on the same inventive concept as the detection method for copper wire multi-pass rolling intermediate annealing in the aforementioned embodiment, the present invention also provides a detection system for copper wire multi-pass rolling intermediate annealing, the system comprising:

[0116] The feature acquisition modeling module collects copper wire surface features for multiple annealing processes, obtains several copper wire surface features, and constructs copper wire feature models based on the several copper wire surface features. Each copper wire feature model corresponds to each annealing process one by one.

[0117] A dependency network construction module performs structural mapping on a number of copper wire feature models, generates a number of copper wire association structures, and constructs a copper wire dependency network based on the copper wire association structures;

[0118] A single defect output module outputs single copper wire defect information according to several copper wire dependency networks;

[0119] The defect feedback integration module establishes a dynamic feedback loop, determines the single annealing deviation based on the dynamic feedback loop and a number of single defect detection information, and integrates a number of single annealing deviations to output the final defect detection result.

[0120] The above-mentioned adjustment system in the present invention can effectively realize the detection method of copper wire multi-pass rolling intermediate annealing, and the technical effect that can be achieved is as described in the above-mentioned embodiment, which will not be repeated here.

[0121] Specifically, the defect feedback integration module includes:

[0122] A deviation feature extraction unit, extracting a single annealing deviation feature according to the single annealing deviation;

[0123] The global deviation model building unit collects historical annealing defect information and builds a global deviation model based on the historical annealing defect information and the single annealing deviation characteristics. The global deviation model uses the output of the upper-order annealing process as the input of the lower-order annealing process to predict the global cumulative deviation.

[0124] The annealing parameter correction unit feeds back the global cumulative deviation according to the single annealing deviation and corrects the single annealing parameters;

[0125] The detection result output unit integrates all single annealing deviations, performs global correction on the final annealing process, obtains cumulative correction results, detects the annealed copper wire based on the cumulative correction results, and outputs the final defect detection results.

[0126] Similarly, the above-mentioned optimization schemes for the system can also respectively achieve the optimization effects corresponding to the method in Example 1, which will not be repeated here.

[0127] Although the present application has been described in conjunction with specific features and embodiments thereof, it is obvious that various modifications and combinations may be made thereto without departing from the spirit and scope of the present application. Accordingly, this specification and the accompanying drawings are merely exemplary illustrations of the present application as defined therein, and are deemed to have covered any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application is intended to include these modifications and variations.

Claims

1. A method for detecting copper wire multi-pass rolling intermediate annealing, characterized in that: The method comprises: Collect copper wire surface features for multiple annealing processes respectively to obtain a number of copper wire surface features, and construct copper wire feature models based on the copper wire surface features, each of which corresponds to each annealing process one by one; Structural mapping is performed on the copper wire feature models to generate a plurality of copper wire association structures, and a copper wire dependency network is constructed based on the plurality of copper wire association structures; Outputting single copper wire defect information respectively according to the plurality of copper wire dependent networks; A dynamic feedback loop is established, and a single annealing deviation is determined based on the dynamic feedback loop and a plurality of single copper wire defect information, and a final defect detection result is output by integrating the plurality of single annealing deviations. The dynamic feedback loop establishment method is as follows: Collect historical annealing detection parameters, define detection parameter thresholds based on the historical annealing detection parameters, and select a feedback control terminal based on annealing requirements; Collecting real-time annealing detection data, using data analysis software to compare the real-time annealing detection data in combination with the detection parameter threshold, marking the real-time annealing detection data that exceeds the detection parameter threshold, and obtaining error feedback; A closed-loop control logic is adopted to perform feedback control on the annealing detection according to the error feedback.

2. The method for detecting copper wire multi-pass rolling intermediate annealing according to claim 1, characterized in that: Structural mapping is performed on the copper wire feature models to generate a plurality of copper wire associated structures, including: According to the copper wire feature model, a plurality of copper wire surface features are respectively converted into corresponding data anchor points, wherein the copper wire surface features include copper wire geometric properties and copper wire physical properties; Establishing links between the data anchor points according to the correlation between the copper wire geometric characteristics and the copper wire physical characteristics, and assigning link weights based on the correlation using a mathematical optimization algorithm; Constructing a dependency feature matrix based on the data anchor point, wherein each row of the dependency feature matrix represents the physical property of the copper wire, and each column represents a different data anchor point; A dependent adjacency matrix is ​​established according to the geometric characteristics of the copper wire and the physical characteristics of the copper wire, wherein the rows and columns of the dependent adjacency matrix are represented as different data anchor points, and the data anchor points are linked to each other to represent 1, otherwise 0, and the adjacency matrix is ​​filled; A copper wire dependency network is constructed according to the dependency feature matrix, the dependency adjacency matrix and the link weights.

3. The method for detecting copper wire multi-pass rolling intermediate annealing according to claim 2, characterized in that: Outputting single copper wire defect information respectively according to the copper wire dependent networks includes: An input layer, responsible for receiving the dependency feature matrix and the dependency adjacency matrix; A first aggregation layer, according to the dependency adjacency matrix, uses a graph convolution algorithm to aggregate the data anchor points to obtain a first-level aggregation anchor point, wherein the aggregation is used to summarize feature information of the data anchor point and adjacent data anchor points; The second aggregation layer performs secondary aggregation on the primary aggregation anchor point and outputs a secondary aggregation anchor point; A fully connected layer maps the secondary aggregated anchor point data to a defect recognition output space to obtain single copper wire defect detection information; The output layer outputs the single copper wire defect detection information.

4. The method for detecting copper wire multi-pass rolling intermediate annealing according to claim 1, characterized in that: The final defect detection result is output by integrating several single annealing deviations, including: Extracting a single annealing deviation feature according to the single annealing deviation; Collecting historical annealing defect information, and building a global deviation model based on the historical annealing defect information and the single annealing deviation characteristics, wherein the global deviation model uses the output of the upper-order annealing process as the input of the lower-order annealing process to predict the global cumulative deviation; Feedback the global cumulative deviation according to the single annealing deviation, and modify the single annealing parameters; All the single annealing deviations are integrated, the final annealing process is globally corrected, a cumulative correction result is obtained, the annealed copper wire is inspected based on the cumulative correction result, and a final defect detection result is output.

5. The method for detecting copper wire multi-pass rolling intermediate annealing according to claim 2, characterized in that: Establishing a link between the plurality of data anchor points according to the correlation between the copper wire geometric characteristics and the copper wire physical characteristics, including: Dividing historical annealing detection parameters into an annealing training group and an annealing test group, the historical annealing detection parameters including the copper wire geometric characteristics and the copper wire physical characteristics; Establishing regression models for several physical properties of the copper wires respectively, training the regression models according to the annealing training group in combination with a regression algorithm, and evaluating them based on the annealing test group to obtain a mapping relationship between the physical properties of the copper wires and the geometric properties of the copper wires; A link relationship between the copper wire geometric characteristics and the copper wire physical characteristics is established according to the mapping relationship.

6. The method for detecting copper wire multi-pass rolling intermediate annealing according to claim 1, characterized in that: Collecting copper wire surface features in multiple annealing processes respectively to obtain a number of copper wire surface features, and constructing copper wire feature models based on the copper wire surface features respectively, including: The surface of the copper wire is geometrically scanned by remote sensing technology to obtain a number of point cloud geometric data, and the physical characteristics of the copper wire are synchronously collected by a detection device, and the point cloud geometric data are integrated with the physical characteristics of the copper wire to obtain a point cloud data set; Performing data cleaning on the point cloud data set, and performing point cloud reconstruction on the cleaned point cloud data set to obtain a characteristic fusion surface, wherein the characteristic fusion surface includes the point cloud geometric data and the physical characteristics of the copper wire; The characteristic fusion surface is smoothed, and the smoothed characteristic fusion surface is subjected to surface fitting to construct a copper wire characteristic model.

7. The method for detecting copper wire multi-pass rolling intermediate annealing according to claim 6, characterized in that: Performing point cloud reconstruction on the cleaned point cloud data set to obtain a feature fusion surface includes: Constructing a copper wire feature coordinate system, making coordinate correspondence between a number of the point cloud data and the copper wire feature coordinate system, and mapping the physical properties of the copper wire with the point cloud data to construct a copper wire feature space, wherein the copper wire feature space includes all the point cloud data and the corresponding physical properties of the copper wire; Inserting a plurality of the point cloud data into the copper wire characteristic space according to coordinates, segmenting the copper wire characteristic space according to the inserted point cloud data, and obtaining the copper wire segmentation space; Constructing a circumscribed ball according to the copper wire cutting space, detecting whether there is the point cloud data other than the point cloud data for constructing the circumscribed ball in the circumscribed ball, and if so, flipping the copper wire segmentation space, and checking again whether there is the point cloud data other than the point cloud data for constructing the circumscribed ball, and if so, continuing to flip until there is no point cloud data other than the point cloud data for constructing the circumscribed ball, and obtaining a set of copper wire discrete points, wherein the circumscribed ball is the minimum circumscribed ball of the copper wire segmentation space; Repeat the insertion and flipping operations of the point cloud data to obtain a number of the copper wire discrete points, and generate a feature fusion surface according to the number of the copper wire discrete points.

8. A detection system for copper wire multi-pass rolling intermediate annealing, characterized in that: The method for detecting the intermediate annealing of copper wire during multi-pass rolling as claimed in claim 1 is adopted, wherein the system comprises: A feature acquisition modeling module collects copper wire surface features for multiple annealing processes, obtains a number of copper wire surface features, and constructs copper wire feature models based on the copper wire surface features, each of which corresponds to each annealing process. A dependency network construction module performs structural mapping on the copper wire feature models to generate a plurality of copper wire association structures, and constructs a copper wire dependency network based on the plurality of copper wire association structures; A single defect output module, which outputs single copper wire defect information respectively according to the plurality of copper wire dependent networks; The defect feedback integration module establishes a dynamic feedback loop, determines the single annealing deviation based on the dynamic feedback loop and the single copper wire defect information, and integrates the single annealing deviations to output a final defect detection result.

9. The detection system for copper wire multi-pass rolling intermediate annealing according to claim 8, characterized in that: The defect feedback integration module includes: A deviation feature extraction unit, which extracts a single annealing deviation feature according to the single annealing deviation; A global deviation model building unit collects historical annealing defect information, and builds a global deviation model based on the historical annealing defect information and the single annealing deviation characteristics, wherein the global deviation model uses the output of the upper-order annealing process as the input of the lower-order annealing process to predict the global cumulative deviation; an annealing parameter correction unit, which feeds back the global cumulative deviation according to the single annealing deviation and corrects the single annealing parameter; The detection result output unit integrates all the single annealing deviations, performs global correction on the final annealing process, obtains cumulative correction results, detects the annealed copper wire based on the cumulative correction results, and outputs the final defect detection result.

Images