Material identification method and system in high-speed wire steel production process
In the production process of high-line steel, image recognition technology is used to obtain the number and material hanging status of the material conveying equipment, and the problem of RFID tag falling off and identifying errors in complex PF line working conditions is solved, achieving high-accurate material recognition.
Patent Information
- Application Number
- CN202510001870.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-01-02
- Publication Date
- 2025-04-25
AI Technical Summary
Due to the complex operating conditions of the PF line, the RFID tags set on the C-type hooks are at risk of falling off. The RFID tags are in a high-temperature environment, resulting in identification errors, and the induction results of the photoelectric sensors are inaccurate due to temperature changes.
The material recognition method is adopted to obtain the image of the material conveying equipment, and the target bounding box is obtained using the yolo algorithm, and combined with the k-means clustering algorithm and model library, the material hanging status of the material conveying equipment is obtained to solve the problems of RFID tag identification errors and inaccurate photoelectric sensors.
It realizes the accurate identification of the number and material hanging status of the material conveying equipment in complex PF line working conditions, avoiding identification errors caused by the fall of RFID tags and the temperature changes of the photoelectric sensor.
Smart Images
Figure CN119888587A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of steel production, and in particular to a material identification method and system in the production process of high-speed wire steel. Background Art
[0002] The purpose of high-speed wire material tracking is to track the entire process of steel billets from entering the furnace, rolling, air cooling to the PF line and record process parameters such as temperature, pressure, diameter gauge and energy consumption. The PF line uses RFID (radio frequency identification) to read the RFID tag on the C-hook to realize the interaction between wire coil information and hook number information.
[0003] The PF line area equipment system is located below the coil loading station. The C-type hook receives the coils transported by the coil transport trolley. During the transportation process, the coils are cooled, inspected, cut, trimmed, sampled, compacted and bundled, weighed and labeled. Finally, the unloading machine unloads the coils from the C-type hook, and the empty hook returns to the coil station to continue the next cycle. Generally, the C-type hook number is obtained through the RFID tag, and the state of the coil is sensed by the photoelectric sensor.
[0004] However, due to the complex working conditions of the PF line, the RFID tag installed on the C-hook is at risk of falling off. The RFID tag is also affected by signal fluctuations in a high temperature environment, resulting in RFID tag recognition errors. The photoelectric sensor responsible for perception will cause inaccurate sensing results due to temperature changes. Summary of the invention
[0005] The present application provides a material identification method and system for the high-speed wire steel production process to solve the current technical problems that due to the complex working conditions of the PF line, the RFID tag set on the C-hook is at risk of falling off, the RFID tag is also affected by signal fluctuations in a high temperature environment, resulting in RFID tag recognition errors, and the photoelectric sensor responsible for sensing the state of the material roll will cause inaccurate sensing results due to temperature changes.
[0006] The first aspect of the present application provides a method for identifying materials in a high-speed steel production process, comprising:
[0007] Acquire images of material handling equipment;
[0008] According to the image, obtaining the serial number of the material conveying equipment;
[0009] Based on the image, using the YOLO algorithm, obtain a target bounding box of the image;
[0010] Based on the target bounding box, using a k-means clustering algorithm to obtain a target cluster of the image;
[0011] Based on the target cluster, using a model library, obtaining the material hanging state of the material conveying equipment; the model library includes: images of the material conveying equipment in several different material hanging states;
[0012] According to the serial number and material hanging state of the material conveying equipment, an identification result of the material conveying equipment is obtained.
[0013] In some embodiments, the step of obtaining the serial number of the material conveying equipment according to the image includes:
[0014] According to the image, obtaining a QR code area of the material conveying equipment;
[0015] Identify the QR code and obtain the serial number of the material conveying equipment.
[0016] In some embodiments, the step of obtaining the serial number of the material conveying equipment according to the image includes:
[0017] According to the image, acquiring a character area of the material conveying equipment;
[0018] The characters are identified to obtain the serial number of the material conveying equipment.
[0019] In some embodiments, after the step of acquiring an image of the material conveying equipment, the method further comprises:
[0020] According to the image, the hook position state of the material conveying equipment is acquired; the hook position state includes: not in place and in place;
[0021] If the hook position status is in place, the hook roll status is obtained according to the image; the hook roll status includes: no detection target is identified, empty hook, and real hook;
[0022] If the hook material roll state is a real hook, the step of obtaining the serial number of the material conveying equipment according to the image is executed.
[0023] In some embodiments, the step of obtaining a target bounding box of the image based on the image using a yolo algorithm comprises:
[0024] The image is divided into grids; if a detection target of the material conveying equipment exists within the grid points of the grid, a bounding box is generated by fitting with the detection target as the center;
[0025] Using a convolutional neural network to obtain target features of the grid points;
[0026] Based on the target features, using a fully connected layer to obtain the confidence of the bounding box;
[0027] Based on the confidence level, an object bounding box of the image is obtained.
[0028] In some embodiments, the step of obtaining a target bounding box of the image based on the confidence level includes:
[0029] Based on the confidence, obtaining the bounding box with maximum confidence within the grid point;
[0030] Based on the bounding box, a non-maximum suppression algorithm is used to delete the bounding boxes with repeated confidences to obtain a target bounding box.
[0031] In some embodiments, the step of obtaining the target cluster of the image based on the target features using a k-means clustering algorithm includes:
[0032] According to the target bounding box, an initial cluster is obtained;
[0033] Calculating the distances between the initial cluster centers;
[0034] Allocating the initial cluster to other initial clusters with the smallest distance from the initial cluster to obtain an intermediate cluster;
[0035] If the absolute value of the difference between the intermediate cluster center and the initial cluster center is greater than a preset value, repeat the step of assigning the intermediate cluster to other intermediate clusters with the smallest distance to the intermediate cluster until the absolute value of the difference between the intermediate cluster centers is less than or equal to the preset value, thereby obtaining the target cluster of the image.
[0036] In some embodiments, the hanging material status includes: cooling, inspection, cutting off the head and tail, trimming, sampling, compacting and bundling, weighing and hanging labels.
[0037] The second aspect of the present application provides a material identification system for a high-speed steel production process, comprising:
[0038] A first acquisition module, used to acquire an image of the material conveying equipment;
[0039] A second acquisition module is configured to acquire a serial number of the material conveying equipment according to the image;
[0040] A third acquisition module is configured to obtain a target bounding box of the image based on the image using a Yolo algorithm;
[0041] A fourth acquisition module is configured to obtain a target cluster of the image based on the target bounding box using a k-means clustering algorithm;
[0042] A fifth acquisition module is configured to acquire the material hanging state of the material conveying equipment based on the target cluster and using a model library; the model library includes: images of a plurality of material conveying equipment in different material hanging states;
[0043] The generating module is configured to obtain the identification result of the material conveying equipment according to the serial number and the material hanging state of the material conveying equipment.
[0044] In some embodiments, the third acquisition module includes:
[0045] A segmentation unit, configured to segment the image into grids; if a detection target of the material conveying equipment exists within a grid point of the grid, a bounding box is generated by fitting with the detection target as the center;
[0046] A first acquisition unit is configured to acquire target features of the grid points using a convolutional neural network;
[0047] A second acquisition unit, based on the target feature, uses a fully connected layer to acquire the confidence of the bounding box;
[0048] A generating unit is configured to obtain a target bounding box of the image based on the confidence level.
[0049] The present application provides a material identification method and system for a high-speed steel production process, the method comprising: acquiring an image of a material conveying device; acquiring the number of the material conveying device according to the image; obtaining a target bounding box of the image using a yolo algorithm based on the image; obtaining a target cluster of the image using a k-means clustering algorithm based on the target bounding box; obtaining a material hanging state of the material conveying device using a model library based on the target cluster; the model library comprising: images of the material conveying device in several different material hanging states; obtaining an identification result of the material conveying device according to the number and the material hanging state of the material conveying device, so as to solve the problem that the RFID tag set on the C-hook has the risk of falling off due to the complex working conditions of the PF line, the RFID tag is also affected by signal fluctuations in a high temperature environment, resulting in RFID tag recognition errors, and the photoelectric sensor responsible for sensing the state of the material roll will cause inaccurate sensing results due to temperature changes. BRIEF DESCRIPTION OF THE DRAWINGS
[0050] In order to more clearly illustrate the technical solution of the present application, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.
[0051] Figure 1This is a flow chart of the material identification method in the high-speed wire steel production process in this application;
[0052] Figure 2 This is the material identification process diagram of the high-speed wire steel production process in this application;
[0053] Figure 3 This is a flow chart of the material identification system during operation of the high-speed wire steel production process in this application. DETAILED DESCRIPTION
[0054] In order to enable those skilled in the art to better understand the technical solutions in this application, the technical solutions in the embodiments of this application will be clearly and completely described below in conjunction with the drawings in the embodiments of this application. Obviously, the described embodiments are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of this application.
[0055] In some technologies, due to the complex working conditions of the PF line, the RFID (radio frequency identification) tag set on the C-hook is at risk of falling off. The RFID tag is also affected by signal fluctuations in a high temperature environment, resulting in RFID tag recognition errors. The photoelectric sensor responsible for sensing the state of the material coil will cause inaccurate sensing results due to temperature changes. In order to solve this technical problem, the present application provides a material identification method and system for the high-speed steel production process. The material identification method and system for the high-speed steel production process are described below:
[0056] Exemplarily, the purpose of high-speed wire material tracking is to realize the whole process tracking of steel billets from entering the furnace, rolling, air cooling to the PF line and the recording of process parameters such as temperature, pressure, diameter gauge and energy consumption. The PF line C-hook uses RFID radio frequency identification to read the code body on the C-hook to realize the interaction between wire coil information and hook information. The PF line (P / F line) is the key equipment in the finishing operation area of the high-speed wire production line. It is mainly used for coil processing at the coiling station, including receiving the coils transported by the coil transport trolley, and completing the processes of cooling, inspection, cutting heads and tails, trimming, sampling, compacting and bundling, weighing and hanging during the transportation of the coils. The PF line area equipment system is below the coiling station. The C-hook receives the coils transported by the coil transport trolley, and completes the processes of cooling, inspection, cutting heads and tails, trimming, sampling, compacting and bundling, weighing and hanging during the transportation of the coils. Finally, the coil is unloaded from the C-hook by the unloading machine, and the empty hook returns to the coil collection station to continue the next cycle. However, the PF line has a more complicated working condition. RFID tags are prone to being hit or falling off. The signal of the tag carrier fluctuates when it runs in a high-temperature environment for a long time. In addition, repeated movements of the coil-hanging trolley and continuous manual emptying of the hook during on-site operations can cause errors in the identification of the tag carrier, affecting the tracking of the PF line.
[0057] Depend on Figure 1 It can be seen that in view of the above problems, the first aspect of the present application provides a material identification method in the high-speed wire steel production process, comprising the following steps:
[0058] S100: Acquire an image of the material conveying equipment; the material conveying equipment includes: C-type hooks and other equipment; the C-type hook is used as an example in the embodiment of the present application. In the embodiment of the present application, a single camera is used to collect images, and the camera is installed on the opposite side of the moving direction of the C-type hook. When the C-type hook just reaches the specified area, the material hanging state and the presence of the wire coil are detected, and the identified C-type hook is marked as entering the loading state; when the C-type hook just leaves, the same detection is performed.
[0059] Exemplarily, the image acquisition device at the inspection station, i.e. the above-mentioned camera, is connected to the operating console switch via Ethernet, using a gigabit communication bandwidth; the switch is connected to the server via single-mode optical fiber, using a 10-gigabit communication bandwidth; the system server is configured with dual network cards and connected to a high-line secondary server for sending and receiving inspection information, thereby obtaining the identification results of the material conveying equipment through the inspection information.
[0060] For example, Figure 2 As shown, this is a flow chart of the image acquisition device during identification in the present application. By reading the workstation track operation signal, it is determined whether the track is running, thereby obtaining whether the C-hook is running. If the track is in operation, the image acquisition device normally captures the C-hook image, thereby obtaining the identification result of the material conveying equipment. If the track is not in operation, the image acquisition device is in standby mode, thereby reducing the power consumption of the image acquisition device.
[0061] like Figure 3 As shown, it is a flowchart for obtaining the serial number of the material conveying equipment in this application, including the following process:
[0062] After the step of acquiring the image of the material conveying equipment, the following steps are included:
[0063] S110: According to the image, the hook position status of the material conveying equipment is obtained; the hook position status includes: not in place and in place; the image of the material conveying equipment is obtained by an image acquisition device. Since the image acquisition device is installed on the opposite side of the moving direction of the C-shaped hook, by judging whether the position of the C-shaped hook is located in the middle of the image, it is possible to judge whether the hook position status is in place. Since the material conveying process can be carried out only when the hook is in place, the following steps will not be performed when the hook is not in place.
[0064] S120: If the hook position status is in place, the hook material roll status is obtained according to the image; the hook material roll status includes: no detection target is identified, empty hook, and real hook; when the C-hook is not located within the detection range of the image acquisition device, the hook material roll status is the detection target is not identified status; when the C-hook is located within the detection range of the image acquisition device, if there is no material mounted on the C-hook, the hook material roll status is the empty hook status, if there is a mounted material on the C-hook, the hook material roll status is the real hook status, and after the real hook status is detected, the following material hanging status is identified.
[0065] S130: If the hook material roll state is a solid hook, executing the step of obtaining the serial number of the material conveying equipment according to the image.
[0066] S200: According to the image, the serial number of the material conveying equipment is obtained; the C-type hook equipment is provided with a serial number and a QR code, wherein the QR code corresponds to equipment information such as the equipment serial number, etc.; currently, the C-type hook equipment adopts RFID radio frequency identification to read the RFID tag provided on the C-type hook to realize the interaction between the wire coil information and the hook number information, however, the working conditions of the PF line are relatively complex, and the RFID tag is at risk of being hit and falling off, and the RFID tag has signal fluctuations when operating in a high temperature environment for a long time, and the repeated actions of the coil hanging trolley and the manual continuous emptying of the hook during on-site operation cause RFID tag recognition errors, thereby affecting the PF line tracking.
[0067] The step of obtaining the serial number of the material conveying equipment according to the image comprises the following sub-steps:
[0068] S210: According to the image, the QR code area of the material conveying equipment is obtained; S220: The QR code is identified to obtain the serial number of the material conveying equipment. By acquiring the image of the material conveying equipment, the QR code area of the material conveying equipment is obtained, and the QR code is identified by the QR code algorithm, thereby obtaining the serial number of the material conveying equipment and confirming the equipment information of the material conveying equipment.
[0069] The step of obtaining the serial number of the material conveying equipment according to the image comprises the following sub-steps:
[0070] S230: According to the image, obtain the character area of the material conveying device; S240: Identify the character and obtain the number of the material conveying device. The C-hook device is provided with a character area such as 98, which represents the device No. 98; in the embodiment of the present application, the character area of the material conveying device can be obtained through the image of the material conveying device, thereby obtaining the number of the material conveying device and confirming the device information of the material conveying device.
[0071] S300: Based on the image, using the yolo algorithm, obtain the target bounding box of the image; the yolo algorithm is a target detection algorithm, and its core idea is to use the information of the entire image to detect the target. The yolo algorithm divides the image into grids, and each grid predicts multiple bounding boxes and corresponding category probabilities, thereby achieving end-to-end target detection.
[0072] The step of obtaining the target bounding box of the image based on the image by using the Yolo algorithm comprises the following sub-steps:
[0073] S310: Divide the image into grids; if the detection target of the material conveying equipment exists within the grid points of the grid, generate a bounding box by fitting with the detection target as the center; S320: Use a convolutional neural network to obtain the target features of the grid points; S330: Based on the target features, use a fully connected layer to obtain the confidence of the bounding box; S340: Based on the confidence, obtain the target bounding box of the image.
[0074] In this embodiment, the YOLO algorithm is as follows: divide the input image into grids: divide the input image into S×S grids, and each grid point in the grid is responsible for detecting a target; if the center point of a target is within a grid, then a bounding box will be fitted within the grid point; extract feature vectors: use a convolutional neural network to extract the feature vector of each grid point, which represents the characteristics of the target within the grid point; predict bounding boxes and categories: for each grid, use a fully connected layer to predict one or more bounding boxes, and the bounding boxes are the detection targets of the material conveying equipment.
[0075] The step of obtaining the target bounding box of the image based on the confidence comprises the following sub-steps: S341: obtaining the bounding box with the maximum confidence within the grid point based on the confidence; S342: deleting the bounding boxes with repeated confidences based on the bounding box using a non-maximum suppression algorithm to obtain the target bounding box. By obtaining the confidence of each bounding box and deleting the bounding boxes with repeated confidences, it is avoided to analyze repeated bounding boxes, thereby improving the recognition efficiency of the hanging material state.
[0076] S400: Based on the target bounding box, a k-means clustering algorithm is used to obtain a target cluster of the image; the main purpose of the k-means clustering algorithm is to divide the samples in the data set into a number of clusters so that the sum of the distances between each sample and the center or mean of the class to which it belongs is minimized. Specifically, the k-means clustering algorithm divides the data set into k clusters through an iterative process, and the samples in each cluster are as similar as possible, while the samples in different clusters are as different as possible.
[0077] The step of obtaining the target cluster of the image based on the target features by using the k-means clustering algorithm comprises the following sub-steps:
[0078] S410: Obtain an initial cluster according to the target bounding box; S420: Calculate the distance between the centers of the initial clusters; S430: Assign the initial cluster to other initial clusters with the smallest distance to the initial cluster to obtain an intermediate cluster; S440: If the absolute value of the difference between the center of the intermediate cluster and the center of the initial cluster is greater than a preset value, repeat the step of assigning the intermediate cluster to other intermediate clusters with the smallest distance to the intermediate cluster until the absolute value of the difference between the centers of the intermediate clusters is less than or equal to the preset value, that is, the absolute value of the difference between the centers of the intermediate clusters of the previous iteration and the centers of the intermediate clusters of the current iteration is less than or equal to the preset value, thereby obtaining the target cluster of the image, that is, the intermediate cluster of the current iteration.
[0079] In this embodiment, the k-means clustering algorithm is as follows: initialization: randomly select k samples, i.e., target bounding boxes, as initial cluster centers; assign samples: for each sample in the data set, calculate its distance from each cluster center, and assign each sample to the cluster represented by the cluster center closest to it; update cluster centers: for each cluster, recalculate its cluster center (i.e., the mean of all points in the cluster); iteration: repeat the above steps until the stopping condition is met; the stopping condition includes: the number of iterations is limited or the cluster center no longer changes significantly, i.e., the absolute value of the difference between the cluster center of the previous iteration and the cluster center of this iteration is less than or equal to the preset value.
[0080] S500: Based on the target cluster, using the model library, the material hanging state of the material conveying equipment is obtained; the model library includes: images of the material conveying equipment in several different hanging states; the model library is established by accumulating images of the material conveying equipment in several different hanging states, wherein each station constructs a model library separately, and each separate model library is compared with the detection target corresponding to the target cluster to obtain the hanging state of the material conveying equipment. The hanging state includes: cooling, inspection, cutting head and tail, trimming, sampling, compacting and bundling, weighing and hanging.
[0081] S600: Obtain the identification result of the material conveying equipment according to the number and the material hanging state of the material conveying equipment. The identification result includes: the number and the material hanging state of the material conveying equipment. The staff can obtain the material hanging state of each material conveying equipment through the identification result.
[0082] The present application provides a material identification method in the production process of high-wire steel. Since the RFID tags installed on the material conveying equipment are likely to be hit and fall off, the RFID tags have signal fluctuations when operating in a high-temperature environment for a long time, and the coil hanging trolleys and manual continuous emptying hooks are repeated during on-site operations, which may cause RFID tag identification errors. The present application adopts visual recognition, that is, the image of the material conveying equipment is collected by an image acquisition device to obtain the identification result of the material conveying equipment, and it can be combined with the detection result of the set RFID tag, thereby solving the problems of inaccurate coding recognition of the material conveying equipment and fluctuations in the detection of the hanging material status due to high temperature.
[0083] The second aspect of the present application provides a material identification system for the high-wire steel production process, including: a first acquisition module for acquiring an image of a material conveying device; a second acquisition module configured to acquire the number of the material conveying device based on the image; a third acquisition module configured to obtain a target bounding box of the image based on the image using the yolo algorithm; a fourth acquisition module configured to obtain a target cluster of the image based on the target bounding box using the k-means clustering algorithm; a fifth acquisition module configured to acquire the hanging state of the material conveying device based on the target cluster using a model library; the model library includes: images of several material conveying devices with different hanging states; a generation module configured to obtain the identification result of the material conveying device based on the number and hanging state of the material conveying device. The effects of the above-mentioned system embodiment during operation can be referred to the effects of the above-mentioned method embodiment, which will not be repeated here.
[0084] In this embodiment, the third acquisition module includes: a segmentation unit, which is used to segment the image into grids; if there is a detection target of the material conveying equipment in the grid points of the grid, a bounding box is generated by fitting with the detection target as the center; a first acquisition unit is configured to use a convolutional neural network to obtain the target features of the grid points; a second acquisition unit, based on the target features, uses a fully connected layer to obtain the confidence of the bounding box; a generation unit is configured to obtain the target bounding box of the image based on the confidence. The effects of the above system embodiment when running can refer to the effects of the above method embodiment, which will not be repeated here.
[0085] The above specific implementation methods further explain in detail the purpose, technical solutions and beneficial effects of the embodiments of the present application. It should be understood that the above are only specific implementation methods of the embodiments of the present application and are not used to limit the protection scope of the embodiments of the present application. Any modifications, equivalent substitutions, improvements, etc. made on the basis of the technical solutions of the embodiments of the present application should be included in the protection scope of the embodiments of the present application.
Claims
1. A material identification method in a high-speed steel production process, characterized in that: include: Acquire images of material handling equipment; According to the image, obtaining the serial number of the material conveying equipment; Based on the image, using the YOLO algorithm, obtain a target bounding box of the image; Based on the target bounding box, using a k-means clustering algorithm to obtain a target cluster of the image; Based on the target cluster, using a model library, obtaining the material hanging state of the material conveying equipment; the model library includes: images of the material conveying equipment in several different material hanging states; According to the serial number and material hanging state of the material conveying equipment, an identification result of the material conveying equipment is obtained.
2. The material identification method in the high-speed wire steel production process according to claim 1 is characterized in that: The step of obtaining the serial number of the material conveying equipment according to the image comprises: According to the image, obtaining a QR code area of the material conveying equipment; Identify the QR code and obtain the serial number of the material conveying equipment.
3. The material identification method in the high-speed wire steel production process according to claim 1 is characterized in that: The step of obtaining the serial number of the material conveying equipment according to the image comprises: According to the image, acquiring a character area of the material conveying equipment; The characters are identified to obtain the serial number of the material conveying equipment.
4. The material identification method in the high-speed wire steel production process according to claim 1 is characterized in that: After the step of acquiring the image of the material conveying equipment, the method further comprises: According to the image, the hook position state of the material conveying equipment is acquired; the hook position state includes: not in place and in place; If the hook position status is in place, the hook roll status is obtained according to the image; the hook roll status includes: no detection target is identified, empty hook, and real hook; If the hook material roll state is a real hook, the step of obtaining the serial number of the material conveying equipment according to the image is executed.
5. The material identification method in the high-speed wire steel production process according to claim 1 is characterized in that: The step of obtaining a target bounding box of the image based on the image by using the YOLO algorithm comprises: The image is divided into grids; if a detection target of the material conveying equipment exists within the grid points of the grid, a bounding box is generated by fitting with the detection target as the center; Using a convolutional neural network to obtain target features of the grid points; Based on the target features, using a fully connected layer to obtain the confidence of the bounding box; Based on the confidence level, an object bounding box of the image is obtained.
6. A material identification method in a high-speed wire steel production process according to claim 5, characterized in that: The step of obtaining a target bounding box of the image based on the confidence level comprises: Based on the confidence, obtaining the bounding box with maximum confidence within the grid point; Based on the bounding box, a non-maximum suppression algorithm is used to delete the bounding boxes with repeated confidences to obtain a target bounding box.
7. The material identification method in the high-speed wire steel production process according to claim 1 is characterized in that: The step of obtaining the target cluster of the image based on the target features by using a k-means clustering algorithm comprises: According to the target bounding box, an initial cluster is obtained; Calculating the distances between the initial cluster centers; Allocating the initial cluster to other initial clusters with the smallest distance from the initial cluster to obtain an intermediate cluster; If the absolute value of the difference between the intermediate cluster center and the initial cluster center is greater than a preset value, repeat the step of assigning the intermediate cluster to other intermediate clusters with the smallest distance to the intermediate cluster until the absolute value of the difference between the intermediate cluster centers is less than or equal to the preset value, thereby obtaining the target cluster of the image.
8. The material identification method in the high-speed wire steel production process according to claim 1 is characterized in that: The hanging material status includes: cooling, inspection, cutting head and tail, trimming, sampling, compacting and bundling, weighing and hanging labels.
9. A material identification system for high-speed steel production process, characterized in that: include: A first acquisition module, used to acquire an image of the material conveying equipment; A second acquisition module is configured to acquire a serial number of the material conveying equipment according to the image; A third acquisition module is configured to obtain a target bounding box of the image based on the image using a Yolo algorithm; A fourth acquisition module is configured to obtain a target cluster of the image based on the target bounding box using a k-means clustering algorithm; A fifth acquisition module is configured to acquire the material hanging state of the material conveying equipment based on the target cluster and using a model library; the model library includes: images of a plurality of material conveying equipment in different material hanging states; The generating module is configured to obtain the identification result of the material conveying equipment according to the serial number and the material hanging state of the material conveying equipment.
10. A material identification system for high-speed wire steel production process according to claim 9, characterized in that: The third acquisition module includes: A segmentation unit, configured to segment the image into grids; if a detection target of the material conveying equipment exists within a grid point of the grid, a bounding box is generated by fitting with the detection target as the center; A first acquisition unit is configured to acquire target features of the grid points using a convolutional neural network; A second acquisition unit, based on the target feature, uses a fully connected layer to acquire the confidence of the bounding box; A generating unit is configured to obtain a target bounding box of the image based on the confidence level.