Method and system for detecting tail angle of hot-rolled strip steel, electronic equipment and medium

Abstract

The invention provides a hot-rolled strip steel tail angle detection method and system, electronic equipment and a medium, and the method comprises the steps: obtaining a hot-rolled coiling image which comprises a steel coil main body and a strip steel tail, taking the steel coil main body and the strip steel tail as target objects, and carrying out the target detection of the hot-rolled coiling image, a plurality of steel coil main body detection frames and a plurality of strip steel tail detection frames are obtained, the plurality of steel coil main body detection frames and the plurality of strip steel tail detection frames are screened and paired to obtain a steel coil-strip tail target pair, and the steel coil-strip tail target pair comprises a steel coil main body target frame and a strip steel tail target frame. Based on the steel coil body target frame and the strip steel tail target frame, the strip steel tail angle is calculated, and coiling control is conducted according to the strip steel tail angle; real-time monitoring and closed-loop control of the belt tail angle can be achieved, the labor cost and the human error risk are reduced, the accuracy of belt tail angle recognition can be improved through screening and pairing, and then the coiling quality and efficiency are improved.

Description

Technical Field

[0001] This invention relates to the field of strip tail angle detection technology, and in particular to a method, system, electronic device and medium for detecting the tail angle of hot-rolled strip. Background Technology

[0002] In the hot rolling process of steel production, the hot rolling coiler is responsible for coiling the rolled strip steel into steel coils. The coiling quality directly affects the overall quality and appearance of the finished steel coil. During the coiling process, the accurate positioning of the strip tail is a key step to ensure the compact formation of the steel coil and prevent loosening. If the strip tail fails to stop accurately in the designated direction, the steel coil may become loose, thus affecting product quality and, in severe cases, even causing abnormal shutdown of the production line.

[0003] Currently, industrial sites still mainly rely on operators to identify the tail angle by visual observation in order to complete control operations. This manual identification method has problems such as low detection accuracy, strong subjectivity, and high labor intensity. Under the high-speed production pace, it is easy to make judgment delays or large errors, which is difficult to meet the requirements of efficiency and accuracy in modern production. Summary of the Invention

[0004] This invention provides a method, system, electronic device, and medium for detecting the tail angle of hot-rolled strip steel, in order to solve the technical problems of low detection accuracy, strong subjectivity, and high labor intensity in manual identification of strip tail angle.

[0005] This invention provides a method for detecting the tail angle of hot-rolled strip steel. The method includes: acquiring a hot-rolled coiling image, the hot-rolled coiling image including a coil body and a strip tail; performing target detection on the hot-rolled coiling image with the coil body and strip tail as target objects to obtain multiple coil body detection frames and multiple strip tail detection frames; filtering and pairing the multiple coil body detection frames and multiple strip tail detection frames to obtain a coil-strip tail target pair, the coil-strip tail target pair including a coil body target frame and a strip tail target frame; calculating the strip tail angle based on the coil body target frame and the strip tail target frame, so as to perform coiling control according to the strip tail angle.

[0006] In one embodiment of the present invention, target detection is performed on the hot-rolled coil image with the main body of the steel coil and the tail of the strip as target objects. This includes: acquiring multiple hot-rolled coil image samples; classifying and labeling the main body of the steel coil and the tail of the strip in each hot-rolled coil image sample to obtain labeled hot-rolled coil image samples, wherein the beginning end of the strip tail is located on one of a pair of opposite corners of the labeling frame, and the end end of the strip tail is located on the other corner; inputting each labeled hot-rolled coil image sample into an initial target detection model; training the initial target detection model to obtain a trained target detection model; and performing target detection on the hot-rolled coil image using the trained target detection model to obtain multiple steel coil main body detection frames and multiple strip tail detection frames.

[0007] In one embodiment of the present invention, screening and pairing of multiple coil body detection frames and multiple strip tail detection frames includes: performing a first screening of the multiple coil body detection frames and multiple strip tail detection frames according to confidence levels to obtain multiple initial screening frames for coil bodies and multiple initial screening frames for strip tails, wherein each coil body detection frame and each strip tail detection frame has a corresponding confidence level; performing a second screening of the multiple initial screening frames for coil bodies according to cross-comparison to obtain each candidate coil body frame, and then matching them according to strip tail detection frames. The initial screening frames at the tail of the steel strip are compared and compared with multiple initial screening frames at the tail of the strip to perform a second screening, resulting in candidate frames for each tail of the strip. The cross-comparison ratio of each initial screening frame of the main body of the steel coil is calculated pairwise from multiple initial screening frames of the main body of the steel coil, and the cross-comparison ratio of each initial screening frame at the tail of the strip is calculated pairwise from multiple initial screening frames of the tail of the strip. Each candidate frame of the main body of the steel coil and each candidate frame at the tail of the strip are paired according to their position information to obtain the steel coil-tail target pair. Each candidate frame of the main body of the steel coil and each candidate frame at the tail of the strip have corresponding position information.

[0008] In one embodiment of the present invention, pairing each candidate frame for the main body of a steel coil with each candidate frame for the tail of a strip steel coil according to position information includes: determining the center point position of each candidate frame for the main body of a steel coil based on the position information of each candidate frame for the main body of a steel coil; determining the center point position of each candidate frame for the tail of a strip steel coil based on the position information of each candidate frame for the tail of a strip steel coil; calculating the distance from the center point position of each candidate frame for the tail of a strip steel coil to the center point position of each candidate frame for the main body of a steel coil; calculating the corresponding radius of the main body of a steel coil based on the position information of each candidate frame for the main body of a steel coil; comparing the distance and the radius of the main body of a steel coil with a preset distance constraint condition; and determining the steel coil-tail target pair according to the comparison result.

[0009] In one embodiment of the present invention, the distance and the radius of the main body of the steel coil are compared with a preset distance constraint condition, and the steel coil-strip tail target pair is determined according to the comparison result. This includes: when there are at least two candidate strip tail frames whose distances from the center point to the center point of the current candidate steel coil body satisfy the preset distance constraint condition with the radius of the main body of the steel coil corresponding to the current candidate steel coil body, a candidate strip tail frame is selected from the at least two candidate strip tail frames to form the steel coil-strip tail target pair with the current candidate steel coil body; when there is only one candidate strip tail frame whose distance from the center point to the center point of the current candidate steel coil body satisfies the preset distance constraint condition with the radius of the main body of the steel coil corresponding to the current candidate steel coil body, the candidate strip tail frame is formed to form the steel coil-strip tail target pair with the current candidate steel coil body.

[0010] In one embodiment of the present invention, calculating the tail angle of the strip steel based on the main target frame of the steel coil and the tail target frame of the strip steel includes: calculating the distances from the center point of the main target frame of the steel coil to the two diagonals of the tail target frame of the strip steel based on the center point position of the main target frame of the steel coil and the corner point positions of the tail target frame of the strip steel, and determining the target diagonal from the two diagonals of the tail target frame of the strip steel based on the calculation results, wherein the center point position of the main target frame of the steel coil is determined based on the position information of the main target frame of the steel coil. The corner positions of the strip tail target frame are determined based on the position information of the strip tail target frame. Both the main target frame of the steel coil and the strip tail target frame have corresponding position information. The strip tail target frame is a rectangular frame. The distances from the center point of the main target frame of the steel coil to the two corner points corresponding to the diagonal of the target frame are calculated, and the strip head end position is determined from the two corner points corresponding to the diagonal of the target frame based on the calculation results. The strip tail angle is calculated based on the strip head end position and the center point position of the main target frame of the steel coil.

[0011] In one embodiment of the present invention, calculating the tail angle of the strip based on the position of the strip tip and the center point position of the target frame of the main body of the steel coil includes: obtaining the tail angle of the strip in the previous frame; calculating the lateral coordinate difference and the longitudinal coordinate difference based on the position of the strip tip and the center point position of the target frame of the main body of the steel coil, and calculating the initial tail angle of the strip in the current frame based on the lateral coordinate difference and the longitudinal coordinate difference; performing a weighted calculation based on the tail angle of the strip in the previous frame and the initial tail angle of the strip in the current frame to obtain the weighted tail angle of the strip in the current frame, and correcting the weighted tail angle of the strip in the current frame according to a preset angle compensation amount to obtain the final tail angle of the strip in the current frame, so as to perform winding control based on the final tail angle of the strip in the current frame.

[0012] This invention also provides a hot-rolled strip tail angle detection system, the system comprising: an image acquisition module for acquiring a hot-rolled coiling image, the hot-rolled coiling image including a coil body and a strip tail; an image processing module for performing target detection on the hot-rolled coiling image with the coil body and strip tail as target objects, obtaining multiple coil body detection frames and multiple strip tail detection frames; filtering and matching the multiple coil body detection frames and multiple strip tail detection frames to obtain a coil-strip tail target pair, the coil-strip tail target pair including a coil body target frame and a strip tail target frame; and calculating the strip tail angle based on the coil body target frame and the strip tail target frame, so as to perform coiling control according to the strip tail angle.

[0013] The present invention also provides an electronic device, the electronic device comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device enables the hot-rolled strip tail angle detection method as described above.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a computer processor, causes the computer to perform the hot-rolled strip tail angle detection method as described above.

[0015] The beneficial effects of this invention are as follows: This invention proposes a method, system, electronic device, and medium for detecting the tail angle of hot-rolled strip steel. This method simultaneously acquires the main body detection frame of the steel coil and the tail detection frame of the strip steel through target detection. It then filters and matches these two frames, effectively eliminating low-quality detection frames and reducing the interference of false detections on subsequent strip tail angle calculations. Through the association and matching mechanism between the main body detection frame and the tail detection frame, adaptive target correspondence is achieved under different working conditions and different steel coil sizes, thus ensuring the stability and consistency of angle calculations. Finally, coiling control is performed based on the calculated tail angle, achieving real-time monitoring and closed-loop control of the tail angle. Furthermore, this method can automatically complete tail angle identification and coiling control without manual intervention, significantly reducing labor costs and the risk of human error, improving the accuracy of tail angle detection, as well as the stability, automation level, and overall efficiency of the hot-rolling coiling process. Attached Figure Description

[0016] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention. It is obvious that the drawings described below are merely some embodiments of the invention, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0017] In the attached diagram: Figure 1 This is a schematic diagram illustrating the implementation environment of a hot-rolled strip tail angle detection method according to an embodiment of the present invention. Figure 2 A flowchart of a method for detecting the tail angle of hot-rolled strip steel according to an embodiment of the present invention; Figure 3a Infrared image of hot-rolled coil provided in an embodiment of the present invention; Figure 3b Visible light image of hot-rolled coil provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of angle calculation provided in an embodiment of the present invention; Figure 5 This is a simplified flowchart of the hot-rolled strip tail angle detection provided in an embodiment of the present invention; Figure 6 This is a flowchart illustrating the specific process for detecting the tail angle of hot-rolled strip steel according to an embodiment of the present invention. Figure 7 This is a block diagram of a hot-rolled strip tail angle detection system provided in an embodiment of the present invention; Figure 8 A hardware system architecture diagram of a hot-rolled strip tail angle detection system provided in another embodiment of the present invention; Figure 9 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention. Detailed Implementation

[0018] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments. Various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. In the absence of conflict, the following embodiments and features in the embodiments can be combined with each other.

[0019] It should be noted that the illustrations provided in the following embodiments are only schematic representations of the basic concept of the present invention. The drawings only show the components related to the present invention and are not drawn according to the actual number, shape and size of the components in the actual implementation. In the actual implementation, the form, quantity and proportion of each component can be arbitrarily changed, and the layout of the components may also be more complex.

[0020] In the following description, numerous details are explored to provide a more thorough explanation of embodiments of the invention. However, it will be apparent to those skilled in the art that embodiments of the invention may be practiced without these specific details. In other embodiments, well-known structures and devices are shown in block diagram form rather than in detail to avoid obscuring embodiments of the invention.

[0021] The embodiments of the present invention provide a method for detecting the tail angle of hot-rolled strip steel, a system for detecting the tail angle of hot-rolled strip steel, an electronic device, a computer-readable storage medium, and a computer program product, which will be described in detail below.

[0022] Please see Figure 1 , Figure 1 This is a schematic diagram illustrating the implementation environment of a hot-rolled strip steel tail angle detection method according to an embodiment of the present invention, as shown below. Figure 1 As shown, the implementation environment may include image acquisition device 110 and computer device 120, wherein, as Figure 1 As shown, the implementation environment may include an image acquisition device 110 and a computer device 120. The image acquisition device 110 may be one or more devices, and may be at least one of a camera, scanner, point cloud radar, etc., without limitation. The computer device 120 may be at least one of a GPU (Graphics Processing Unit) computer, an NPU (Neural Processing Unit) computer, a GPU computing cluster, an NPU computing cluster, a microcomputer, an embedded computer, a neural network computer, etc., without limitation. The computer device 120 can be used to automatically perform target detection on the hot-rolled coil image and determine the strip tail angle based on the target detection result. The image acquisition device 110 can acquire the hot-rolled coil image and provide it to the computer device 120 for processing.

[0023] Schematic, computer device 120 acquires a hot-rolled coiling image through image acquisition device 110, wherein the hot-rolled coiling image includes the main body of the steel coil and the tail of the strip. Taking the main body of the steel coil and the tail of the strip as target objects, target detection is performed on the hot-rolled coiling image to obtain multiple steel coil main body detection frames and multiple strip tail detection frames. The multiple steel coil main body detection frames and multiple strip tail detection frames are filtered and matched to obtain steel coil-strip tail target pairs, wherein the steel coil-strip tail target pair includes a steel coil main body target frame and a strip tail target frame. Based on the steel coil main body target frame and the strip tail target frame, the strip tail angle is calculated so as to perform coiling control according to the strip tail angle. As can be seen, the technical solution of this invention simultaneously acquires the main body detection frame and the strip tail detection frame through target detection, and filters and matches these frames to effectively eliminate low-quality detection frames and reduce the interference of false detections on subsequent strip tail angle calculations. Through the association and matching mechanism between the main body detection frame and the strip tail detection frame, adaptive target correspondence is achieved under different working conditions and different coil sizes, thereby ensuring the stability and consistency of angle calculations. Finally, coiling control is performed based on the calculated strip tail angle, achieving real-time monitoring and closed-loop control of the strip tail angle. Furthermore, this technical solution can automatically complete strip tail angle identification and coiling control without manual intervention, significantly reducing labor costs and the risk of human error, improving the accuracy of strip tail angle detection, as well as the stability, automation level, and overall efficiency of the hot rolling coiling process.

[0024] It should be noted that the hot-rolled strip tail angle detection method provided in this embodiment of the invention is generally executed by computer equipment 120.

[0025] Please see Figure 2 , Figure 2 This is a flowchart illustrating a method for detecting the tail angle of hot-rolled strip steel according to an embodiment of the present invention. This method can be applied to... Figure 1 The implementation environment is shown, and the method is specifically executed by computer equipment 120 within that implementation environment. It should be understood that this hot-rolled strip tail angle detection method can also be applied to other exemplary implementation environments and executed by equipment in other implementation environments. This embodiment does not limit the implementation environment to which the hot-rolled strip tail angle detection method is applicable. Figure 2 As shown, in an exemplary embodiment, the hot-rolled strip tail angle detection method includes at least steps S210 to S240, which are described in detail below: Step S210: Obtain the hot-rolled coil image.

[0026] In one embodiment of the present invention, the hot-rolled coiling image refers to the image of the hot-rolled coiling machine winding the strip steel into a coil, also known as a strip steel image. The hot-rolled coiling image includes the main body of the steel coil and the tail of the strip steel, wherein the tail of the strip steel is also called the strip tail. The original image of the hot-rolled coiling can be acquired by an image acquisition device set near the production line, and the original image is preprocessed to obtain the preprocessed image as the hot-rolled coiling image. The image acquisition device can be at least one of a camera, a camera, a scanner, a point cloud radar, etc. The original image can be a picture or a video, and can be a visible light image or an infrared image. The preprocessing includes at least one of scaling, cropping, rotation, flipping, grayscale processing, binarization processing, data augmentation, etc., and there are no limitations on these aspects.

[0027] In one specific embodiment of the present invention, during the hot rolling coiling process, an industrial camera can be used to capture real-time images of the strip in the coiling state as the original images. Please refer to [link to relevant documentation]. Figure 3a , Figure 3b , Figure 3a This is an infrared image of hot-rolled coil provided in an embodiment of the present invention. Figure 3b Visible light images of hot-rolled coils provided in an embodiment of the present invention. For example... Figure 3a , Figure 3b As shown, the original image can be an infrared image or a visible light image, including the main body of the steel coil and the tail of the strip. The original image can be preprocessed by the internal or edge device of the industrial visible light camera to enhance the contrast between the strip edge and tail areas. During the preprocessing process, the image can also be cropped to retain only the main body of the steel coil and the tail area of ​​the strip to reduce noise interference and improve the accuracy of subsequent recognition. The preprocessed image is then transmitted to the computer equipment as a hot-rolled coil image.

[0028] Step S220: Using the main body of the steel coil and the tail of the strip as target objects, target detection is performed on the hot-rolled coiling image to obtain multiple detection boxes for the main body of the steel coil and multiple detection boxes for the tail of the strip.

[0029] In one embodiment of the present invention, features can be extracted from the hot-rolled coil image, and the extracted features of the hot-rolled coil image can be matched according to the preset features of the main body of the steel coil and the preset features of the tail of the strip steel to identify the position, category and other information of the main body of the steel coil and the tail of the strip steel in the hot-rolled coil image as target detection results, including the detection box of the main body of the steel coil and the detection box of the tail of the strip steel.

[0030] In another embodiment of the present invention, the trained target detection model can be used to detect the main body of the steel coil and the tail of the strip as target objects, and target detection can be performed on the hot-rolled coil image to obtain target detection results, including the detection box of the main body of the steel coil and the detection box of the tail of the strip.

[0031] Each detection frame (including the main body detection frame and the strip tail detection frame) has at least corresponding location information and category. The location information of the detection frame indicates the location and area of ​​the target object, while the category indicates whether the target object within the detection frame is the main body of a steel coil or the tail of a strip. In other words, the category of the detection frame is used to distinguish whether it is a main body detection frame or a strip tail detection frame. The specific form of the category can be a label or the color of the detection frame. Of course, each detection frame can also have a corresponding confidence level to represent the probability of the target object's existence and the accuracy of the classification.

[0032] Step S230: Screen and match multiple main body detection frames of steel coils and multiple tail detection frames of strip steel to obtain steel coil-tail target pairs.

[0033] In one embodiment of the present invention, the coil-strip tail target pair includes a coil body target frame and a strip tail target frame.

[0034] Since the target detection results may contain some duplicate or low-quality detection boxes, to improve the identification accuracy of the main body of the steel coil and the tail of the strip, multiple detection boxes for the main body of the steel coil and multiple detection boxes for the tail of the strip can be filtered in the target detection results. For example, NMS (Non-Maximum Suppression) processing can be performed independently on multiple detection boxes for the main body of the steel coil and multiple detection boxes for the tail of the strip to achieve the filtering of detection boxes.

[0035] Since the screened detection frames may contain multiple main steel coil detection frames or multiple strip tail detection frames, in order to avoid interference from irrelevant detection frames, the screened main steel coil detection frames and strip tail detection frames are paired to ensure that the main steel coil target frame and the strip tail target frame in the steel coil-strip tail target pair belong to the same physical entity, thereby improving the accuracy of strip tail angle calculation.

[0036] Step S240: Based on the main target frame of the steel coil and the target frame of the tail strip in the steel coil-tail target pair, calculate the tail strip angle so as to perform winding control according to the tail strip angle.

[0037] In one embodiment of the present invention, the strip tail angle represents the spatial orientation of the strip tail relative to the main body of the coil in the current winding state, providing an angle reference for subsequent strip tail positioning and automatic control.

[0038] For a coil-strip tail target pair, the rotation angle of the strip tail relative to the coil body can be calculated based on the position information of the main target frame of the coil and the tail target frame of the strip. This angle is then used as the strip tail angle. The calculated strip tail angle is output to the host computer control system (winding control system) in real time to determine the position status of the strip tail and execute relevant decisions, such as strip tail guidance, automatic tail positioning, tail tracking, and control logic triggering. This forms a complete closed-loop control link, achieving real-time monitoring and closed-loop control of the strip tail angle.

[0039] In some embodiments, the center point position of the main body of the steel coil can be determined based on the position information of the target frame of the main body of the steel coil, and the center point position of the tail of the strip can be determined based on the position information of the target frame of the tail of the strip. A coordinate system is established with the center point position of the main body of the steel coil as the origin, and the center point position of the tail of the strip is transformed according to the established coordinate system. Based on the transformed center point position of the tail of the strip, the slope of the straight line formed by the center point position of the main body of the steel coil and the center point position of the tail of the strip is calculated, and the included angle is obtained by the arctangent function, which is used as the angle of the tail of the strip.

[0040] In other embodiments, the center point of the steel coil body can be determined based on the position information of the target frame of the steel coil body, and the center point of the strip tail can be determined based on the position information of the target frame of the strip tail. The difference between the longitudinal and transverse coordinates is calculated based on the center point of the steel coil body and the center point of the strip tail. The included angle is obtained by using the arctangent function and the difference between the longitudinal and transverse coordinates, and is used as the angle of the strip tail.

[0041] In one embodiment of the present invention, step S220 includes: acquiring multiple hot-rolled coil image samples; classifying and labeling the main body of the steel coil and the tail of the strip in each hot-rolled coil image sample to obtain each labeled hot-rolled coil image sample, wherein the head end of the strip tail is located on one of a pair of opposite corners of the labeling frame, and the tail end of the strip tail is located on the other; inputting each labeled hot-rolled coil image sample into an initial target detection model, training the initial target detection model to obtain a trained target detection model, and performing target detection on the hot-rolled coil image using the trained target detection model to obtain multiple steel coil main body detection frames and multiple strip tail detection frames.

[0042] In this embodiment, images of steel coils at different coiling stages can be pre-collected from the hot rolling coiling site as hot rolling coiling image samples. The main body of the steel coil and the tail of the strip in each hot rolling coiling image sample are respectively regarded as two independent target objects and classified and labeled using rectangular boxes. The rectangular boxes include a first real box covering the main body area of ​​the steel coil and a second real box covering the tail area of ​​the strip. The tail of the strip includes the strip tail to the strip head.

[0043] Please continue reading Figure 3a ,like Figure 3a As shown, the red box is the second ground truth box. The diagonal formed by a pair of opposite corners of the second ground truth box can connect the tail end point and the head end point of the strip, so that the second ground truth box can constrain the geometric features of the tail region of the strip. Thus, the head end point can be deduced through the boundary relationship of the ground truth box in subsequent target detection.

[0044] After the annotation is completed, multiple annotated hot-rolled coil image samples are obtained. These annotated hot-rolled coil image samples are divided into training and validation sets according to the proportion and stored for subsequent model training.

[0045] The initial object detection model is a deep convolutional network object detection algorithm, i.e., an object detection-type deep learning network. The training set is input into the initial object detection model for supervised learning training to learn the spatial distribution characteristics of the main body of the steel coil and the tail of the strip. For example, object detection-type deep learning networks include, but are not limited to, YOLO series, Faster R-CNN, RT-DETR, etc., and can also be replaced with other mainstream object detection networks according to actual needs.

[0046] After model training, a target detection model is obtained that can be used to identify the main body of the steel coil and the tail of the strip. This target detection model can simultaneously output a first detection box and a second detection box, where the first detection box is the detection box for the main body of the steel coil, and the second detection box is the detection box for the tail of the strip. During the inference phase, the trained target detection model is used to perform target detection on the hot-rolled coil image, outputting information for each detection box, including location information, category, etc., to achieve synchronous detection of the main body of the steel coil and the tail region of the strip.

[0047] In one embodiment of the present invention, before step S230, the method includes: performing coordinate mapping on the position information corresponding to the multiple steel coil main body detection frames and the multiple strip tail detection frames.

[0048] In this embodiment, the set of detection boxes output by the target detection model can be represented as follows:

[0049] in, B For the set of detection boxes, b i For the detection box, ( x i , y i () represents the x and y coordinates of the center point of the detection box in the image coordinate system. w i The width of the detection frame. h i The height of the detection frame,c i The category of the detection frame can be either "steel coil body" or "strip tail". p i The confidence level of the detection box. i The index number represents the number of the detection box in the detection box set, and there are a total of [number missing]. N One detection box. x i , y i , w i , h i The values ​​of ) are all in normalized coordinate format, and their range is limited to [0,1]. They can be calculated relative to the aspect ratio of the input image of the model.

[0050] The detection box set can be divided according to the category of the detection box to obtain the main body detection box set of steel coil and the tail detection box set of strip steel. The main body detection box set of steel coil includes multiple main body detection boxes of steel coil, and the tail detection box set of strip steel includes multiple tail detection boxes of strip steel. Coordinate mapping is performed on the multiple main body detection boxes of steel coil and the multiple tail detection boxes of strip steel in the tail detection box set, thereby obtaining the actual position information of the detection box in the original image coordinate system.

[0051] In some embodiments, for the main body detection frame of the steel coil, its center point position at the original image resolution can be calculated based on its normalized center coordinates, as follows: Equation (1) in,( x b_i , y b_i () represents the center point position of the detection frame of type steel coil body at the original image resolution. x i , y i () represents the center point of the detection box in the normalized format within the image coordinate system. W img The width of the hot-rolled coil image. H img This represents the height of the hot-rolled coil image.

[0052] For the strip tail detection frame, the corner position at the original image resolution can be calculated based on its normalized center coordinates and width and height. The calculation method is as follows: Equation (2) in,( x t_l_i ,y t_t_i ) represents the position of the top left corner of the detection frame (type: steel tail) at the original image resolution. x t_l_i , y t_b_i ) represents the position of the bottom left corner of the detection box at the original image resolution. x t_r_i , y t_t_i ) represents the position of the upper right corner of the detection box at the original image resolution. x t_r_i , y t_b_i ) represents the position of the bottom right corner of the detection box at the original image resolution. x i , y i () represents the center point of the detection box in the normalized format within the image coordinate system. w i The width of the detection frame. h i The height of the detection frame. W img The width of the hot-rolled coil image. H img This represents the height of the hot-rolled coil image.

[0053] Of course, for the main detection frame of the steel coil, the corner position at the original image resolution can also be calculated according to the calculation method of formula (2) based on its normalized center coordinates and width and height.

[0054] In some embodiments, after mapping is completed, for the set of detection frames for the main body of the steel coil, the center point position of the detection frame can be extracted as the center point position of the main body of the steel coil; for the set of detection frames for the tail of the strip, the four corner points of the detection frame can be extracted for subsequent calculation of the tail angle of the strip.

[0055] In one embodiment of the present invention, step S230 includes: performing a first screening on multiple coil body detection frames and multiple strip tail detection frames according to confidence level to obtain multiple coil body preliminary screening frames and multiple strip tail preliminary screening frames, wherein each coil body detection frame and each strip tail detection frame has a corresponding confidence level; performing a second screening on multiple coil body preliminary screening frames according to cross-comparison of the coil body preliminary screening frames to obtain each coil body candidate frame, and performing a second screening on multiple strip tail preliminary screening frames according to cross-comparison of the strip tail preliminary screening frames to obtain each strip tail candidate frame, wherein the cross-comparison ratio of each coil body preliminary screening frame is calculated pairwise from multiple coil body preliminary screening frames, and the cross-comparison ratio of each strip tail preliminary screening frame is calculated pairwise from multiple strip tail preliminary screening frames; pairing each coil body candidate frame with each strip tail candidate frame according to position information to obtain a coil-strip tail target pair, wherein each coil body candidate frame and each strip tail candidate frame has corresponding position information.

[0056] In some embodiments, a first screening is performed on multiple main coil detection frames and multiple tail strip detection frames according to confidence levels, including: for the main coil detection frames, if the confidence level of the main coil detection frame is less than a preset confidence threshold, the main coil detection frame is removed; otherwise, the main coil detection frame is retained, and the retained main coil detection frames are used as the initial screening frames for the main coil; for the tail strip detection frames, if the confidence level of the tail strip detection frame is less than a preset confidence threshold, the tail strip detection frame is removed; otherwise, the tail strip detection frame is retained, and the retained tail strip detection frames are used as the initial screening frames for the tail strip.

[0057] In some embodiments, a second screening is performed on multiple initial screening frames of the main body of the steel coil according to the intersection-union ratio (IU / U) of the initial screening frames. This includes: for each initial screening frame, determining the initial screening frame with the highest confidence level as the first candidate frame; calculating the IU / U ratio of each remaining initial screening frame with the first candidate frame; if the IU / U ratio of the initial screening frame with the first candidate frame is greater than a preset IU / U ratio threshold, then the initial screening frame is removed; otherwise, it is retained. The remaining primary screening frames for the main body of the steel coil are used as candidate frames for the main body of the steel coil. For the primary screening frames at the tail end of the strip, the primary screening frame at the tail end of the strip with the highest confidence level is determined as the first candidate frame for the tail end of the strip. The intersection-union ratio (IUR) of each of the remaining primary screening frames at the tail end of the strip with the first candidate frame at the tail end of the strip is calculated. If the IUR of the primary screening frame at the tail end of the strip with the first candidate frame at the tail end of the strip is greater than the preset IUR threshold, the primary screening frame at the tail end of the strip is removed. Otherwise, the primary screening frame at the tail end of the strip is retained. The retained primary screening frames at the tail end of the strip are used as candidate frames for the tail end of the strip.

[0058] For example, the intersection-union ratio can be calculated as follows: Equation (3) in, IoU For intersection, union, and comparison, intersection The intersection of the two detection boxes. union The union of the two detection boxes. S A and S B These are the areas of the two detection boxes, ( x 1，A , y 1，A )and( x 2，A , y 2，A ) represents the top left and bottom right corner positions of a detection box. x 1，B , y 1，B )and( x 2，B , y 2，B ) represents the top left and bottom right corner positions of another detection box. +1.0 indicates the boundary correction term when taking the closed interval of coordinates, which is used to ensure that the area calculation of the detection box region includes the endpoint pixels, thereby avoiding the case of zero area under discrete pixel coordinates.

[0059] It should be understood that the initial screening frame, candidate frame, and target frame are merely different naming conventions used to avoid confusion when referring to detection frames at different processing stages. Regardless of whether it is an initial screening frame, candidate frame, or target frame, they are essentially still detection frames. Therefore, the two detection frames referred to in equation (3) are the first candidate frame of the main body of the steel coil and the remaining initial screening frame of the main body of the steel coil, or the first candidate frame of the tail of the strip steel and the remaining initial screening frame of the tail of the strip steel.

[0060] After completing the first and second screenings, at least one candidate frame for the main body of a steel coil and at least one candidate frame for the tail of a strip are obtained. If there is only one candidate frame for the main body of a steel coil and one candidate frame for the tail of a strip, the candidate frame for the main body of the steel coil can be directly used as the target frame for the main body of the steel coil, and the candidate frame for the tail of the strip can be used as the target frame for the tail of the strip, forming a steel coil-tail target pair. If there are two or more candidate frames for the main body of a steel coil, and / or two or more candidate frames for the tail of a strip, the position information of one candidate frame for the main body of a steel coil and one candidate frame for the tail of a strip can be used to determine whether the candidate frame for the main body of the steel coil and the candidate frame for the tail of the strip satisfy geometric constraints. If the geometric constraints are satisfied, the pairing is successful, and the candidate frame for the main body of the steel coil is used as the target frame for the main body of the steel coil, and the candidate frame for the tail of the strip is used as the target frame for the tail of the strip, forming a steel coil-tail target pair, thereby ensuring that the target frame for the main body of the steel coil and the target frame for the tail of the strip in the steel coil-tail target pair belong to the same physical entity.

[0061] For example, geometric constraints can be distance constraints, overlap constraints, relative position constraints, etc., and there are no restrictions here.

[0062] In one embodiment of the present invention, pairing each candidate frame for the main body of a steel coil with each candidate frame for the tail of a strip steel coil according to position information includes: determining the center point position of each candidate frame for the main body of a steel coil based on the position information of each candidate frame for the main body of a steel coil; determining the center point position of each candidate frame for the tail of a strip steel coil based on the position information of each candidate frame for the tail of a strip steel coil; calculating the distance from the center point position of each candidate frame for the tail of a strip steel coil to the center point position of each candidate frame for the main body of a steel coil; and calculating the corresponding steel coil main body radius based on the position information of each candidate frame for the main body of a steel coil; comparing the distance and the steel coil main body radius with preset distance constraints; and determining the steel coil-tail target pair based on the comparison results.

[0063] In some embodiments, for each steel coil body candidate frame b j It can be obtained from its location information ( x bj , y bj , w j , h j Extract the center point position () x bj , y bj ) is used as the center point of the steel coil body, and based on its position information ( x bj , y bj , w j , h j Width in ) w j and height h j Estimate the corresponding steel coil body radius R i .

[0064] In some embodiments, for each strip tail candidate frame t i It can be based on its location information ( x t_l_i , y t_t_i , x t_r_i , y t_b_i ), calculate the coordinates of the midpoint of its diagonal, which is also the center point of the candidate frame for the tail of the strip, as follows: Equation (4) in,( x t_m_i , y t_m_i ( ) represents the coordinates of the midpoint of the diagonal of the candidate frame for the tail section of the strip. x t_l_i Let x be the x-coordinate of the left corner point of the candidate frame for the steel tail section. x t_r_i Let x be the x-coordinate of the right corner point of the candidate frame for the steel tail section. y t_t_i Let be the ordinate of the top corner of the candidate frame for the steel tail section. y t_b_i The coordinates of the bottom corner of the candidate frame for the strip tail are given.

[0065] In some embodiments, the Euclidean distance from the center point of the strip tail candidate frame (coordinates of the midpoint of the diagonal of the strip tail candidate frame) to the center point of the coil body candidate frame (center point of the coil body) can be calculated based on the center point of the strip tail candidate frame (coordinates of the midpoint of the diagonal of the strip tail candidate frame) and the center point of the coil body candidate frame (center point of the coil body). The calculation method is as follows: Equation (5) in, d ij The Euclidean distance from the center point of the candidate frame for the tail section of the strip (the midpoint of the diagonal of the candidate frame for the tail section of the strip) to the center point of the candidate frame for the main body of the coil (the center point of the main body of the coil). x t_m_i , y t_m_i ) represents the center point of the candidate frame for the strip tail (coordinates of the midpoint of the diagonal of the candidate frame for the strip tail). x bj , y bj ) represents the center point position of the candidate frame for the main body of the steel coil (center point position of the main body of the steel coil).

[0066] In some embodiments, the distance and the radius of the main body of the steel coil are compared with preset distance constraints, and the steel coil-tail target pair is determined according to the comparison result. This includes: if the Euclidean distance between the center point of a candidate frame for the tail of a strip (coordinates of the midpoint of the diagonal of the candidate frame for the tail of the strip) and the center point of a candidate frame for the main body of a steel coil (center point of the main body of the steel coil) satisfies the preset distance constraint with the radius of the main body of the steel coil corresponding to the candidate frame for the main body of the steel coil, then the candidate frame for the main body of the steel coil and the candidate frame for the tail of the strip are considered to have a valid matching relationship in space. Therefore, the candidate frame for the main body of the steel coil and the candidate frame for the tail of the strip are respectively used as the target frame for the main body of the steel coil and the target frame for the tail of the strip to form a steel coil-tail target pair.

[0067] For example, the preset distance constraints are as follows: d ij ≤ R i +Δ in, d ij The Euclidean distance is the distance from the center point of the candidate frame for the tail section of the strip (coordinates of the midpoint of the diagonal of the candidate frame for the tail section of the strip) to the center point of the candidate frame for the main body of the coil (center point of the main body of the coil). R i Δ is the radius of the steel coil body corresponding to the candidate frame of the steel coil body, and Δ is the preset distance tolerance threshold, which can be preset based on experience with the rolling machine and image resolution.

[0068] In one embodiment of the present invention, the distance and the radius of the main body of the steel coil are compared with preset distance constraints, and the steel coil-strip tail target pair is determined according to the comparison result. This includes: when there are at least two candidate strip tail frames whose distances from the center point to the center point of the current candidate steel coil body satisfy the preset distance constraints with the radius of the main body of the steel coil corresponding to the current candidate steel coil body, one candidate strip tail frame is selected from the at least two candidate strip tail frames to form a steel coil-strip tail target pair with the current candidate steel coil body; when there is only one candidate strip tail frame whose distance from the center point to the center point of the current candidate steel coil body satisfies the preset distance constraints with the radius of the main body of the steel coil corresponding to the current candidate steel coil body, the candidate strip tail frame is formed into a steel coil-strip tail target pair with the current candidate steel coil body.

[0069] In this embodiment, the current candidate frame of the main body of the steel coil refers to the candidate frame of the main body of the steel coil that is currently being compared to determine whether it meets the preset distance constraint. It can be understood as a certain candidate frame of the main body of the steel coil. For a certain candidate frame of the main body of the steel coil, when there are multiple candidate frames of the tail of the strip that meet the preset distance constraint, the candidate frame of the tail of the strip with the smallest distance can be selected as the final matching result of the candidate frame of the main body of the steel coil. The two are combined to form a unique and valid steel coil-tail target pair, which is used for the subsequent calculation of the tail angle of the strip.

[0070] It should be understood that the candidate frame for the tail of the strip with the smallest distance refers to the candidate frame for the tail of the strip with the smallest distance from its center point to the center point of the current candidate frame for the main body of the steel coil.

[0071] In some embodiments, comparing the distance and the radius of the main body of the steel coil with a preset distance constraint condition specifically includes: generating a set of main bodies of steel coils based on multiple candidate frames of main bodies of steel coils, and generating a set of tails of strip steel based on multiple candidate frames of tails of strip steel; taking a candidate frame of main bodies of steel coils from the set of main bodies of steel coils as the current candidate frame of main bodies of steel coils, and traversing the candidate frames of tails of strip steel in the set of tails of strip steel. If the distance from the center point of a candidate frame of tails of strip steel to the center point of the current candidate frame of main bodies of steel coils satisfies the preset distance constraint condition with the radius of the main body of the steel coil corresponding to the current candidate frame of main bodies of steel coils, then the candidate frame of tails of strip steel is determined as a candidate frame of tails of strip steel, until all candidate frames of tails of strip steel in the set of tails of strip steel have been traversed; if there is only one candidate frame of tails of strip steel, then the current candidate frame of main bodies of steel coils and the radius of the main body of steel coils corresponding to the current candidate frame of main bodies of steel coils satisfy the preset distance constraint condition, then the candidate frame of tails of strip steel is determined as a candidate frame of tails of strip steel, until all candidate frames of tails of strip steel in the set of tails of strip steel have been traversed; if there is only one candidate frame of tails of strip steel, then the current candidate frame of main bodies of steel coils and the radius of the main body of strip steel are compared with the radius of the main body of strip steel. The process involves identifying the undetermined strip tail frame and forming a coil-strip tail target pair. The current coil main candidate frame is removed from the coil main set, and the undetermined strip tail frame is also removed from the strip tail set. If multiple undetermined strip tail frames exist, the current coil main candidate frame and the undetermined strip tail frame with the smallest distance are combined to form a coil-strip tail target pair. The current coil main candidate frame is then removed from the coil main set, and the undetermined strip tail frame with the smallest distance is also removed from the strip tail set. If no corresponding undetermined strip tail frame exists for the current coil main candidate frame, the current coil main candidate frame is removed from the coil main set, and the next coil main candidate frame is taken from the coil main set as the new current coil main candidate frame. This comparison operation is repeated until a coil-strip tail target pair is obtained.

[0072] It should be understood that the strip tail undetermined frame with the smallest distance refers to the strip tail undetermined frame whose center point is the smallest distance from the center point of the current steel coil main body candidate frame.

[0073] In one embodiment of the present invention, the strip tail angle is calculated based on the main target frame of the steel coil and the tail target frame of the strip, including: calculating the distances from the center point of the main target frame of the steel coil to the two diagonals of the tail target frame of the strip based on the center point position of the main target frame of the steel coil and the corner point positions of the tail target frame of the strip, and determining the target diagonal from the two diagonals of the tail target frame of the strip based on the calculation results, wherein the center point position of the main target frame of the steel coil is determined based on the position information of the main target frame of the steel coil, and the corner point positions of the tail target frame of the strip are determined based on the position information of the tail target frame of the strip, and both the main target frame of the steel coil and the tail target frame of the strip have corresponding position information, and the tail target frame of the strip is a rectangular frame; calculating the distances from the center point position of the main target frame of the steel coil to the two corner points corresponding to the target diagonal, and determining the strip head end position from the two corner points corresponding to the target diagonal based on the calculation results; and calculating the strip tail angle based on the strip head end position and the center point position of the main target frame of the steel coil.

[0074] In this embodiment, by calculating the vertical distance from the center point of the main target frame of the steel coil to the two diagonals of the tail target frame of the strip, the target diagonal is determined as the main direction line of the tail of the strip, and the position of the strip tip is selected accordingly. This geometric constraint can accurately identify the strip tip, so that the tail angle of the strip can be accurately calculated based on the center point of the main target frame of the steel coil and the position of the strip tip.

[0075] Please see Figure 4 , Figure 4 This is a schematic diagram of angle calculation provided in an embodiment of the present invention, as shown below. Figure 4 As shown, XY is the image coordinate system, the green detection box is the main target box of the steel coil, and the red detection box is the tail target box of the strip. A local polar coordinate system can be established based on the spatial relationship between the main target box of the steel coil and the tail target box of the strip. Specifically, this includes setting the center point position of the main target box of the steel coil (…). x c , y c (as the origin of the local coordinate system) O The local coordinate system reference zero angle direction is defined by the vertical upward direction of the hot-rolled coil image. y The positive direction of the axis, with clockwise direction as the positive direction of the angle, is denoted as . θ The coordinates of the four corner points of the target frame at the tail of the strip are marked as the top left corner. x tl , y tl ), top right corner ( x tr , y tr (bottom left corner) x bl , y bl bottom right corner x br , y br ).

[0076] A geometric constraint and temporal correlation mechanism is introduced to adaptively optimize the detection results and calculate the real-time angle information of the strip tail angle. Specifically, this includes representing the two diagonals of the strip tail target box as straight line segments. L 1. L 2, of which, L 1 represents the diagonal line from the top left corner to the bottom right corner, that is ( x tl , y tl )to( x br , y brThe diagonal of ) L 2 represents the diagonal line from the top right corner to the bottom left corner, that is ( x tr , y tr )to( x bl , y bl The diagonals of ); for each diagonal L k ( k =1,2), assuming the positions of its two corner points are ( x k1 , y k1 ), ( x k2 , y k2 The center point of the main target frame of the steel coil. O ( x c , y c ) to the diagonal L k vertical distance D k The calculation method is as follows: Equation (6) in, D k The center point of the target frame of the steel coil is located at the diagonal. L k vertical distance, ( x k1 , y k1 ), ( x k2 , y k2 () is the diagonal L k The corner position, ( x c , y c () represents the center point of the target frame of the steel coil. The perpendicular distances from the center point of the main target frame of the steel coil to the two diagonals of the target frame at the tail of the strip are denoted as follows: D 1. D 2. Comparison D 1 and D 2. When the conditions are met D m =max( D 1, D 2), mWhen the condition is ∈{1,2}, choose D m Corresponding diagonal L m As the main direction line of the strip tail, that is, selecting D 1 and D The maximum value in 2 is used as the target diagonal, i.e., the main direction line of the strip tail. Please continue reading Figure 4 ,like Figure 4 As shown, for the selected main direction line of the strip tail, the positions of its two corner points are denoted as follows: V a and V b Then calculate the center point position of the target frame of the steel coil. O arrive V a , V b The distance between these two corner points is calculated as follows: Equation (7) in, d a The distance from the center point to the corner of the target frame of the steel coil. V a distance, d b The distance from the center point to the corner of the target frame of the steel coil. V b The distance, ( x a , y a () is the corner point V a Location, ( x b , y b () is the corner point V b Location, ( x c , y c () represents the center point of the target frame of the steel coil.

[0077] when d a ≥ d b Time to take V a Otherwise take V b That is, to d a and db The corner point corresponding to the maximum value is determined as the leading endpoint, i.e., the leading endpoint. P h ( x h , y h ); Based on the center point of the main target frame of the steel coil O ( x c , y c ) and leading endpoint P h ( x h , y h ) position, calculate the angle of the strip tail.

[0078] It should be noted that, Figure 4 The schematic diagram for calculating the strip tail angle is simplified to aid understanding of the features of the coil body and strip tail. The actual coil body, strip tail, and corresponding inspection frame can be referenced elsewhere. Figure 3a .

[0079] In one embodiment of the present invention, the strip tail angle is calculated based on the position of the strip head endpoint and the center point position of the main target frame of the steel coil, including: obtaining the strip tail angle of the previous frame; calculating the lateral coordinate difference and the longitudinal coordinate difference based on the position of the strip head endpoint and the center point position of the main target frame of the steel coil, and calculating the initial strip tail angle of the current frame based on the lateral coordinate difference and the longitudinal coordinate difference; performing a weighted calculation based on the strip tail angle of the previous frame and the initial strip tail angle of the current frame to obtain the weighted strip tail angle of the current frame, and correcting the weighted strip tail angle of the current frame according to a preset angle compensation amount to obtain the final strip tail angle of the current frame, so as to perform winding control based on the final strip tail angle of the current frame.

[0080] In this embodiment, the calculation step of the tail angle of the current frame strip includes: based on the center point position of the target frame of the steel coil body. O ( x c , y c ) and leading endpoint P h ( x h , y h The position of the lead end is determined, and the coordinate difference between the lead end and the center point of the main target frame of the steel coil is calculated as follows: Equation (8) Where, Δ x The difference in horizontal coordinates, Δ y The difference between the vertical coordinates, ( x h , y h ) represents the position of the leading endpoint, ( x c , y c () represents the center point of the target frame of the steel coil. like Figure 4 As shown, since the y-axis direction in the image coordinate system is opposite to that in the mathematical coordinate system, in order to ensure that the angle calculation direction is consistent, the vertical coordinate difference can also be reversed. The calculation method is as follows: Δ y '=-Δ y = y c - y h Equation (9) Where, Δ y ' represents the difference in vertical coordinates after reverse processing, Δ y The difference is the vertical coordinate. y c This represents the ordinate of the center point of the target frame for the steel coil. y h The ordinate of the position of the leading endpoint; Based on the lateral coordinate difference and the reverse-processed longitudinal coordinate difference, the polar coordinate angle of the current frame strip tail relative to the center of the main body of the steel coil is calculated as follows: θ =arctan2(Δ y ',Δ x =arctan2( y c - y h , x h - x c Formula (10) in, θ Δ represents the polar coordinate angle of the current frame strip tail relative to the center of the main body of the steel coil. y ' represents the difference in vertical coordinates after reverse processing, Δ x The difference is the horizontal coordinate. x h , y h ) represents the position of the leading endpoint, which is ( x c , y c() represents the center point of the target frame of the steel coil. radians as angles θ Convert to standardized angle values ​​within the range of 0° to 360°. The specific conversion method is as follows: θ '=( θ ×180 / π +360) mod 360-degree (11) in, θ 'This is the initial angle of the strip tail in the current frame.' θ The polar coordinate angle of the strip tail relative to the center of the main body of the steel coil in the current frame; To suppress short-term noise and angle fluctuations, the initial value of the strip tail angle can be filtered and compensated in real time to align with the actual strip tail angle under actual working conditions. For example, EWMA (Exponentially Weighted Moving Average) or an adaptive filtering algorithm can be used, as detailed below: θ t = α · θ ' t +(1- α )· θ t-1 + C Equation (12) in, θ t This is the filtered strip tail angle signal of the current frame, which is the final strip tail angle of the current frame. θ ' t This is the unfiltered real-time strip tail angle signal, i.e., the initial strip tail angle of the current frame. θ t-1 This is the filtered tail angle signal of the previous frame, i.e., the angle of the strip tail in the previous frame. α These are the filter coefficients, used to adjust the weights between the current angle and historical angles. α ∈[0,1], C This is a preset angle compensation amount, which is related to the specific operating conditions of the winding machine. It is used to correct systematic deviations caused by changes in operating conditions, thereby adjusting the angle weighting value. θ t The expected size.

[0081] Of course, other appropriate filtering algorithms can be selected based on the actual operating status of the winding machine; no restrictions are imposed here.

[0082] The tail angle signal of the current frame will be calculated in real time. θ tThe output is sent to the coiling control system for automatic tail positioning, tail tracking, and control logic triggering, enabling real-time monitoring and closed-loop control of the coil tail posture, thereby improving the stability and control accuracy of strip tail angle detection.

[0083] This embodiment calculates the tail angle by using the matched effective steel coil-tail target pair, and introduces filtering and compensation mechanisms to smooth and correct the angle signal, which can significantly improve the stability and accuracy of the angle output.

[0084] The technical solution of the present invention will be illustrated below by providing a specific embodiment in actual work.

[0085] Please see Figure 5 , Figure 5 This is a simplified flowchart of the hot-rolled strip tail angle detection according to an embodiment of the present invention, as shown below. Figure 5 As shown, the simplified process for detecting the tail angle of hot-rolled strip steel is as follows: S1. Acquire images of the strip in the coiling state using an industrial camera, and preprocess the images to enhance the contrast between the strip edge and the tail area, thus obtaining a hot-rolled coiled image. S2. Input the hot-rolled coil image into the trained main body-strip tail joint detection model, i.e., the target detection model. This model is used to simultaneously detect the main body area of ​​the steel coil and the tail area of ​​the strip, and output the position information of the corresponding main body detection box and the tail detection box of the strip. S3. Based on the spatial relationship between the main detection frame of the steel coil and the tail detection frame of the strip, a local polar coordinate system is established, and a geometric constraint and temporal correlation mechanism is introduced to adaptively optimize the detection results and calculate the real-time angle information of the tail of the strip. S4. Output the calculated strip tail angle information to the host computer control system to determine the strip tail position status in real time and realize closed-loop control and related decisions.

[0086] Please see Figure 6 , Figure 6 This is a flowchart illustrating the specific process for detecting the tail angle of hot-rolled strip steel according to an embodiment of the present invention, as follows: Figure 6 As shown, the specific process for detecting the tail angle of hot-rolled strip is as follows: Image frames of the acquired strip steel image are preprocessed. The preprocessed image is then input into the main body-strip tail detection model as a hot-rolled coil image. The model's detection results are then checked to determine whether there are detection boxes for the main body of the coil and the tail of the strip steel. If both types of detection boxes exist, the next step is performed; otherwise, the next image frame of the strip steel image is identified. After simultaneously detecting the main body detection frame of the steel coil and the tail detection frame of the strip, the position information of the main body detection frame of the steel coil and the tail detection frame of the strip is mapped by coordinates. Then, the mapped main body detection frame of the steel coil and the tail detection frame of the strip are filtered by NMS and confidence. The selected candidate main body frame of the steel coil and the candidate tail of the strip are deduplicated and matched to construct the steel coil-tail target pair. The steel coil-tail target pair includes a steel coil body target frame and a strip tail target frame. Extract the center point of the matching target frame of the steel coil body and the corner point of the target frame of the strip tail. Calculate the vertical distance from the center point of the target frame of the steel coil body to the two diagonals of the target frame of the strip tail. Take the diagonal with the largest vertical distance as the main direction line of the strip tail (strip tail). The corner point on the main direction line of the strip tail that is farthest from the center point of the target frame of the steel coil body is the end point of the strip head. The angle of the strip tail is calculated based on the center point of the target frame of the steel coil body and the head end point. Angle filtering and compensation are then performed, and the final filtered and compensated angle signal is output to the winding tail-fixing system for control execution.

[0087] For detailed procedures of this embodiment, please refer to the descriptions in the foregoing embodiments; they will not be repeated here. By using a target detection model combined with deep learning and image post-processing technology, efficient identification and angle analysis of the tail angle of the steel coil are achieved. Compared with traditional manual monitoring methods, it has higher identification accuracy, faster response speed, and higher degree of automation, effectively improving the positioning accuracy during the steel coil winding process, reducing human error, and improving overall production efficiency and intelligence level.

[0088] Please see Figure 7 , Figure 7 This is a block diagram of a hot-rolled strip tail angle detection system according to an embodiment of the present invention. This system can be applied to... Figure 1 The implementation environment shown is specifically configured in computer device 120. This system can also be applied to other exemplary implementation environments and specifically configured in other devices; this embodiment does not limit the implementation environment to which the system is applicable.

[0089] like Figure 7As shown, the exemplary hot-rolled strip tail angle detection system includes: an image acquisition module 710 for acquiring a hot-rolled coil image, which includes a coil body and a strip tail; an image processing module 720 for performing target detection on the hot-rolled coil image with the coil body and strip tail as target objects, obtaining multiple coil body detection frames and multiple strip tail detection frames; filtering and matching the multiple coil body detection frames and multiple strip tail detection frames to obtain a coil-strip tail target pair, which includes a coil body target frame and a strip tail target frame; and calculating the strip tail angle based on the coil body target frame and the strip tail target frame to perform coiling control according to the strip tail angle.

[0090] In this embodiment, the image acquisition module 710 can be an image acquisition device such as a camera, scanner, or point cloud radar, or a signal acquisition system for collecting signals from the aforementioned image acquisition device; the image processing module 720 can be a GPU computer, NPU computer, GPU computing cluster, NPU computing cluster, microcomputer, embedded computer, neural network computer, image processor, etc., and there are no limitations here. The image acquisition module 710 and the image processing module 720 can be independent electronic devices or can be configured in the same electronic device, and there are no limitations here.

[0091] Please see Figure 8 , Figure 8 A hardware system architecture diagram of a hot-rolled strip tail angle detection system provided in another embodiment of the present invention is shown below. Figure 8 As shown, the hot-rolled strip tail angle detection system includes: The image acquisition and preprocessing unit includes an industrial camera (such as an integrated smart camera), a supplementary lighting device (light source), and an edge computing module. It is used to acquire strip images in real time during the hot rolling coiling process. After the acquired images are preliminarily processed and converted by A / D through the camera's internal or edge devices, they are transmitted to the image processing and analysis unit. The image processing and analysis unit, as the core part of the system, includes an image annotation module, a model training module, a target detection module, and an angle analysis module. This unit uses a deep learning model to identify targets in the acquired images, extract the spatial positional relationship between the steel coil body and the tail of the strip, and calculate the relative rotation angle information of the tail of the strip. The communication and transmission unit consists of an industrial switch, photoelectric conversion equipment, and communication cables, and is used for high-speed and stable transmission of image data and analysis results between modules. The business management unit, including the management server, operating terminal and supporting software platform, is responsible for functions such as visualization of test results, parameter configuration, data storage management and historical record query. The data storage unit can be a local storage device (such as a hard disk recorder) or a network storage system to save data such as image information, detection logs and angle results, which will facilitate later traceability and analysis. Electrical control unit: including industrial field distribution boxes, control modules (such as PLC, Programmable Logic Controller) and cable laying system, used to power the entire system and ensure the electrical safety and stability of equipment operation.

[0092] The image acquisition and preprocessing unit is an example of the image acquisition module 710, and the image processing and analysis unit is an example of the image processing module 720.

[0093] It should be noted that the hot-rolled strip tail angle detection system and the hot-rolled strip tail angle detection method provided in the above embodiments belong to the same concept. The specific operation methods of each module have been described in detail in the method embodiments and will not be repeated here. In practical applications, the hot-rolled strip tail angle detection system provided in the above embodiments can be assigned to different functional modules as needed, that is, the internal structure of the system can be divided into different functional modules to complete all or part of the functions described above. This is not a limitation here.

[0094] In one embodiment of the present invention, an electronic device is also provided, comprising: one or more processors; and a storage device for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the electronic device enables the hot-rolled strip tail angle detection method provided in the above embodiments.

[0095] Please see Figure 9 , Figure 9 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. It should be noted that... Figure 9 The electronic device 900 shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0096] like Figure 9 As shown, the electronic device 900 includes a processor 901, a memory 902, and a communication bus 903; the communication bus 903 is used to connect the processor 901 and the memory 902; the processor 901 is used to execute a computer program stored in the memory 902 to implement one or more methods in the above embodiments.

[0097] In one embodiment of the present invention, a computer-readable storage medium is also provided, on which a computer program is stored. When executed by a computer processor, the computer program causes the computer to perform the hot-rolled strip tail angle detection method as described above. This computer-readable storage medium may be included in the electronic device described in the above embodiments, or it may exist independently and not incorporated into the electronic device.

[0098] In one embodiment of the present invention, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the hot-rolled strip tail angle detection method provided in the various embodiments described above.

[0099] The electronic device provided in this embodiment of the invention includes a processor, a memory, a transceiver, and a communication interface. The memory and the communication interface are connected to the processor and the transceiver and complete communication between them. The memory is used to store computer programs, the communication interface is used to perform communication, and the processor and the transceiver are used to run the computer programs, so that the electronic device performs the various steps of the above method.

[0100] In embodiments of the present invention, the memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk storage device.

[0101] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0102] As will be understood by those skilled in the art, the computer-readable storage medium in the embodiments of the present invention can implement all or part of the steps of the above method embodiments by hardware related to computer programs. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM (Read-Only Memory), RAM (Random Access Memory), magnetic disks, or optical disks.

[0103] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method of detecting a tail end angle of a hot-rolled strip, characterized by, The method includes: Acquire a hot-rolled coil image, the hot-rolled coil image including the main body of the steel coil and the tail of the strip; Using the main body of the steel coil and the tail of the strip as target objects, target detection is performed on the hot-rolled coil image to obtain multiple detection boxes for the main body of the steel coil and multiple detection boxes for the tail of the strip. Multiple main body detection frames and multiple tail detection frames of strip steel are screened and matched to obtain steel coil-tail target pairs, wherein each steel coil-tail target pair includes a main body target frame of steel coil and a tail target frame of strip steel. Based on the main target frame of the steel coil and the tail target frame of the strip, the tail angle of the strip is calculated so as to perform winding control according to the tail angle of the strip.

2. The method of detecting the tail end angle of a hot-rolled strip steel according to claim 1, characterized by, Target detection is performed on the hot-rolled coil image, with the main body of the steel coil and the tail of the strip as the target objects, including: Acquire multiple hot-rolled coil image samples; The main body of the steel coil and the tail of the strip in each hot-rolled coil image sample are classified and labeled to obtain each labeled hot-rolled coil image sample. The head end of the strip tail is located on one of the two opposite corners of the label frame, and the tail end of the strip tail is located on the other. Each labeled hot-rolled coil image sample is input into the initial target detection model, and the initial target detection model is trained to obtain the trained target detection model. The trained target detection model is then used to perform target detection on the hot-rolled coil image to obtain multiple main body detection boxes of steel coils and multiple tail detection boxes of strip steel.

3. The tail angle detection method for hot-rolled strip steel according to claim 1 or 2, characterized in that, Screening and matching of multiple main body inspection frames for steel coils and multiple tail inspection frames for strip steel, including: The main body detection frames of multiple steel coils and the tail detection frames of multiple strip steels were screened for the first time according to the confidence level, resulting in multiple preliminary screening frames of the main body of steel coils and multiple preliminary screening frames of the tail of strip steels. Each main body detection frame of steel coils and each tail detection frame of strip steels has a corresponding confidence level. A second screening is performed by comparing and contrasting multiple primary screening frames for the main body of the steel coil to obtain candidate frames for each main body of the steel coil. Similarly, a second screening is performed by comparing and contrasting multiple primary screening frames for the tail of the strip to obtain candidate frames for each tail of the strip. The cross-comparison ratio of each primary screening frame for the main body of the steel coil is calculated pairwise from multiple primary screening frames for the main body of the steel coil, and the cross-comparison ratio of each primary screening frame for the tail of the strip is calculated pairwise from multiple primary screening frames for the tail of the strip. Each main body candidate frame of a steel coil is paired with each tail candidate frame of a strip steel according to the location information to obtain the steel coil-strip tail target pair, wherein each main body candidate frame of a steel coil and each tail candidate frame of a strip steel have corresponding location information.

4. The method of detecting the tail end angle of a hot-rolled strip steel according to claim 3, characterized by, The candidate frames for the main body of each steel coil are paired with the candidate frames for the tail of each strip steel coil according to the location information, including: The center point of each candidate frame of the main body of each steel coil is determined based on the position information of each candidate frame of the tail of each strip. Calculate the distance from the center point of each strip tail candidate frame to the center point of each coil body candidate frame, and calculate the corresponding coil body radius based on the position information of each coil body candidate frame; The distance and the radius of the main body of the steel coil are compared with the preset distance constraints, and the target pair of steel coil-tail is determined based on the comparison results.

5. The method of detecting the tail end angle of a hot-rolled strip steel according to claim 4, characterized by, The distance and the radius of the main body of the steel coil are compared with preset distance constraints. Based on the comparison results, the steel coil-tail target pair is determined, including: When there are at least two candidate strip tail frames whose distances from the center point to the center point of the current coil body candidate frame satisfy the preset distance constraint condition with respect to the radius of the corresponding coil body of the current coil body candidate frame, a candidate strip tail frame is selected from the at least two candidate strip tail frames to form the coil-strip tail target pair with the current coil body candidate frame. When there exists only one candidate strip tail frame whose distance from the center point to the center point of the current coil body candidate frame satisfies the preset distance constraint condition with the radius of the corresponding coil body of the current coil body candidate frame, the candidate strip tail frame and the current coil body candidate frame constitute the coil-tail target pair.

6. The method for detecting the tail angle of hot-rolled strip steel according to claim 1, characterized in that, Based on the target frame of the main body of the steel coil and the target frame of the tail of the strip, the angle of the tail of the strip is calculated, including: Based on the center point of the main target frame of the steel coil and the corner points of the tail target frame of the strip, the distances from the center point of the main target frame of the steel coil to the two diagonals of the tail target frame of the strip are calculated. Based on the calculation results, the target diagonal is determined from the two diagonals of the tail target frame of the strip. The center point of the main target frame of the steel coil is determined based on the position information of the main target frame of the steel coil, and the corner points of the tail target frame of the strip are determined based on the position information of the tail target frame of the strip. Both the main target frame of the steel coil and the tail target frame of the strip have corresponding position information. The tail target frame of the strip is a rectangular frame. Calculate the distances from the center point of the target frame of the steel coil to the two corner points corresponding to the diagonal of the target, and determine the head end point position from the two corner points corresponding to the diagonal of the target based on the calculation results; The tail angle of the strip is calculated based on the position of the strip tip and the center point of the main target frame of the steel coil.

7. The method for detecting the tail angle of hot-rolled strip steel according to claim 6, characterized in that, Based on the position of the strip's head end and the center point position of the main target frame of the steel coil, the angle of the strip's tail is calculated, including: Get the angle of the steel strip tail in the previous frame; Based on the position of the strip tip and the center point of the target frame of the main body of the steel coil, calculate the lateral coordinate difference and the longitudinal coordinate difference, and calculate the initial current frame strip tail angle based on the lateral coordinate difference and the longitudinal coordinate difference. The current frame strip tail angle is obtained by weighting the previous frame strip tail angle and the initial current frame strip tail angle. The weighted current frame strip tail angle is then corrected according to a preset angle compensation amount to obtain the final current frame strip tail angle, and the winding control is performed based on the final current frame strip tail angle.

8. A system for detecting the tail angle of hot-rolled strip steel, characterized in that, The system includes: The image acquisition module is used to acquire hot-rolled coil images, which include the main body of the steel coil and the tail of the strip. The image processing module is used to perform target detection on the hot-rolled coiling image, with the main body of the steel coil and the tail of the strip as target objects, to obtain multiple detection frames for the main body of the steel coil and multiple detection frames for the tail of the strip; to filter and match the multiple detection frames for the main body of the steel coil and the multiple detection frames for the tail of the strip to obtain a steel coil-tail target pair, wherein the steel coil-tail target pair includes a target frame for the main body of the steel coil and a target frame for the tail of the strip; and to calculate the angle of the tail of the strip based on the target frame for the main body of the steel coil and the target frame for the tail of the strip, so as to perform coiling control according to the angle of the tail of the strip.

9. An electronic device, characterized in that, The electronic device includes: One or more processors; A storage device for storing one or more programs, which, when executed by one or more processors, cause the electronic device to implement the hot-rolled strip tail angle detection method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, It stores a computer program, which, when executed by the computer's processor, causes the computer to perform the hot-rolled strip tail angle detection method as described in any one of claims 1-7.

Images