Square billet on-line fixed-length and fixed-weight adjusting method based on visual technology

Abstract

The invention discloses a square billet on-line fixed-length and fixed-weight adjusting method based on a visual technology, which comprises the following steps of: acquiring an end face image of a square billet through a high-resolution industrial camera, calculating the area of the end face, combining parameters such as steel rolling weighing data, pulling speed, steel grade components and the like, establishing an associated database, and dynamically adjusting the cutting fixed-length length of the square billet. The problem that in the prior art, the fixed-length and fixed-weight precision is insufficient due to changes of steel grade components, the pulling speed and the like is solved, high-precision fixed-length and fixed-weight control is achieved, and the device has the advantages of being non-contact, resistant to high temperature, high in accuracy, high in adaptability and the like. The machine vision technology is applied to end face area detection and calculation and introduced into fixed length and fixed weight of a continuous casting billet, the fixed length of the square billet is dynamically adjusted according to the change value of the end face area of the square billet, the target of the same standard fixed weight of each square billet needed by a steel rolling mill is achieved, and the target of 2kg / t control precision can be achieved.

Description

Technical Field

[0001] This invention relates to the field of automation application technology, and in particular to a method for online length and weight adjustment of billets based on vision technology. Background Technology

[0002] In the continuous casting process of steel enterprises, the control of billet length and weight is a key link in ensuring steel quality and production efficiency. Currently, the length and weight adjustment methods used in the continuous casting process of steel enterprises in China include: 1. Infrared length control, weighing of individual billets, and manual adjustment of billet length; 2. Encoder length control, weighing of individual billets, and manual adjustment of billet length; 3. Laser length control, weighing of individual billets, and manual adjustment of billet length; 4. Based on infrared length control, encoder, or laser length control, introducing machine-aided algorithms, weighing of individual billets, and manual adjustment of billet length.

[0003] The first three methods all involve determining the required fixed-length weight for rolling different specifications; calculating the billet length of the continuous casting machine production section based on the fixed-length weight; cutting each billet to the fixed-length during continuous casting; weighing the hot billet on the roller conveyor when it is delivered to the steel rolling mill, comparing it with the standard weight, and manually reducing or increasing the fixed-length of a certain flow; the fourth method involves using the casting speed and casting temperature of each flow of the networked continuous casting machine to simulate and calculate the density of each flow of billet under hot conditions, thereby providing simulated values ​​and adjusting the billet length.

[0004] However, existing methods for adjusting length and weight generally have the following drawbacks: (1) High dependence on manual labor: The first three methods rely entirely on operators to judge the adjustment amount based on the weighing data, which requires long-term continuous monitoring, resulting in high labor costs and susceptibility to subjective judgment. When the continuous casting machine speed fluctuates greatly or the steel composition changes, the lag and experience error of manual adjustment will cause fluctuations in the fixed weight accuracy. Currently, it can only achieve the fixed length and fixed weight control accuracy of 5Kg / t steel, which is difficult to meet the needs of high-precision production.

[0005] (2) Significant errors in machine-assisted algorithms: Although the fourth method introduces algorithm simulation, factors such as wear of copper pipes in continuous casting machines and changes in casting speed in industrial field applications will cause the actual cross-sectional area of ​​the billet to deviate from the theoretical value. The algorithm cannot accurately capture this change in real time, resulting in a large deviation between the calculated length and weight and the actual value. The error even exceeds that of manual adjustment. Summary of the Invention

[0006] To overcome the above-mentioned shortcomings of the prior art, the present invention provides a method for online length and weight adjustment of billets based on vision technology.

[0007] The technical solution adopted by this invention to solve its technical problem is: a method for online length and weight adjustment of billets based on vision technology, comprising: Image acquisition: Real-time acquisition of images of the end face of the continuously cast billet using a camera; Image preprocessing and analysis: Preprocessing and semantic segmentation of the end face image to accurately calculate the area of ​​the billet end face; Data correlation and density calculation: Based on the formula: Actual density = Weighing weight / (End face area × Fixed length), calculate the density of the current billet; obtain the rolling weight, casting speed and steel composition of the current billet from the network; Fixed length adjustment: Construct a dynamic relational database of "end face area - drawing speed - tundish temperature - steel grade - density"; dynamically adjust the fixed length of the next billet based on the target fixed weight value, real-time end face area and matching density value: adjusted fixed length = target fixed weight / (end face area × matching density); transmit the adjusted fixed length to the cutting equipment for precise cutting.

[0008] As a further improvement of the present invention, the camera is a high-resolution industrial camera.

[0009] As a further improvement of the present invention: the image acquisition includes: an infrared sensor acquiring the billet positioning signal and transmitting the signal to a high-resolution industrial camera, triggering the high-resolution industrial camera to capture the end face of the billet.

[0010] As a further improvement of the present invention: the high-resolution industrial camera is installed inside a water-cooled protective cover above the roller conveyor table.

[0011] As a further improvement of the present invention, the image acquisition also includes: obtaining the internal and external parameter matrix of the equipment through high-resolution industrial camera calibration before shooting, eliminating camera distortion and establishing the mapping relationship between the image coordinate system and physical space.

[0012] As a further improvement of the present invention: the image preprocessing and analysis includes: preprocessing the captured image, using a multi-quality processing mechanism to filter out invalid samples and underexposed images, and performing brightness normalization processing on valid samples; inputting the preprocessed image into a deep semantic segmentation network based on deep learning, performing pixel-level end-face region segmentation, extracting contours and performing morphological processing, and accurately calculating the geometric area of ​​the billet end face.

[0013] As a further improvement of the present invention, the multi-quality processing mechanism includes filtering out invalid samples with abnormal trigger positions by judging the position of the square billet in the image, and removing underexposed images with insufficient brightness based on image analysis.

[0014] As a further improvement of the present invention: the extraction of contour and morphological processing includes: using a semantic segmentation model to identify the end face region of the target image, obtaining the corresponding rounded quadrilateral region, binarizing and morphologically processing the segmentation result to eliminate interference, extracting the complete outer contour and performing smoothing processing.

[0015] As a further improvement of the present invention, the geometric area of ​​the end face of the billet is calculated using a connected domain analysis algorithm.

[0016] As a further improvement of the present invention: the data association and density calculation includes: assigning a virtual number to each billet, tracking its weighing weight at the rolling mill, calculating the billet density based on the weighing weight, end face area and fixed length, and simultaneously acquiring data such as the casting speed and steel composition during the production of the billet via the network.

[0017] As a further improvement of the present invention: the construction logic of the dynamic relational database includes: Establish a mapping relationship between the density and end face area of ​​square billets of the same steel grade, casting speed, and tundish temperature; Once the accumulated data reaches the threshold, the fixed length is calculated directly by calling the historical matching density value.

[0018] As a further improvement of the present invention: the fixed length adjustment is interlocked with the infrared fixed length device, the encoder fixed length device or the laser fixed length device.

[0019] As a further improvement of the present invention: the water-cooled protective cover is provided with a camera hole with a lens, and the water-cooled protective cover is resistant to high temperatures greater than 800°C.

[0020] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention uses a high-resolution industrial camera to capture images of the billet end face, calculates the end face area, and combines parameters such as rolling weight, drawing speed, and steel composition to establish a dynamic correlation database. This allows for dynamic adjustment of the billet cutting length, solving the problem of insufficient accuracy in cutting length and weight due to changes in steel composition and drawing speed in existing technologies, and achieving high-precision cutting length and weight control.

[0021] 2. This invention applies machine vision technology to end face area detection and calculation, and introduces it into the fixed length and weight of continuously cast billets. Based on the change value of the end face area of ​​the billet, the fixed length of the billet is dynamically adjusted to achieve the goal of the same standard fixed weight for each billet required by the steel rolling mill, and can achieve the control accuracy target of 2kg / t. Attached Figure Description

[0022] Figure 1 A schematic diagram of the structure taken by a 2D camera.

[0023] Figure 2 A real-world image taken with a 2D camera.

[0024] Figure 3 This is a diagram illustrating the installation of a camera bracket.

[0025] Figure 4 This is a schematic diagram of the water-cooled protective cover.

[0026] Figure 5 The logic block diagram for fixed length and weight.

[0027] Figure 6 This is a structural block diagram of the present invention.

[0028] Notes: 1. Camera bracket; 2. Billet; 3. Roller conveyor; 4. Water-cooled protective cover; 5. Water inlet; 6. Shooting hole with lens; 7. 2D camera. Detailed Implementation

[0029] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below in conjunction with specific embodiments and corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention. It should be noted that the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to these processes, methods, products, or devices.

[0030] Please see Figure 1-6 A method for online length and weight adjustment of billets based on vision technology includes: Image acquisition: Real-time acquisition of images of the end face of continuously cast billets using a high-resolution industrial camera; Image preprocessing and analysis: Preprocessing and semantic segmentation of the end face image to accurately calculate the area of ​​the billet end face; Data correlation and density calculation: Based on the formula: Actual density = Weighing weight / (End face area × Fixed length), calculate the density of the current billet; obtain the rolling weight, casting speed and steel composition of the current billet from the network; Fixed length adjustment: Construct a dynamic relational database of "end face area - drawing speed - tundish temperature - steel grade - density"; dynamically adjust the fixed length of the next billet based on the target fixed weight value, real-time end face area and matching density value: adjusted fixed length = target fixed weight / (end face area × matching density); transmit the adjusted fixed length to the cutting equipment for precise cutting.

[0031] By acquiring images of the billet end face using a high-resolution industrial camera and calculating the end face area, and combining these with parameters such as rolling weight, drawing speed, and steel composition, a dynamic correlation database is established to dynamically adjust the billet cutting length. This solves the problem of insufficient length and weight accuracy caused by changes in steel composition and drawing speed in existing technologies, and achieves high-precision length and weight control.

[0032] By applying machine vision technology to end face area detection and calculation, and introducing it into the length and weight determination of continuously cast billets, a control accuracy target of 2 kg / t can be achieved.

[0033] In some embodiments, the image acquisition includes: an infrared sensor acquiring a billet positioning signal and transmitting the signal to a high-resolution industrial camera, triggering the high-resolution industrial camera to capture images of the billet's end face.

[0034] In some embodiments, the high-resolution industrial camera is mounted inside a water-cooled protective cover above the roller conveyor table.

[0035] In some implementations, the high-resolution industrial camera is a 2D camera.

[0036] In some embodiments, the image acquisition further includes: obtaining the internal and external parameter matrix of the equipment through calibration of a high-resolution industrial camera before shooting, eliminating camera distortion, and establishing a mapping relationship between the image coordinate system and physical space.

[0037] In some embodiments, the image preprocessing and analysis includes: preprocessing the captured image, using a multi-level quality processing mechanism to filter out invalid samples and underexposed images, and performing brightness normalization processing on valid samples; inputting the preprocessed image into a deep semantic segmentation network based on deep learning, performing pixel-level end-face region segmentation, extracting contours and performing morphological processing, and accurately calculating the geometric area of ​​the billet end face.

[0038] In some implementations, the multiple quality processing mechanism includes filtering out invalid samples with abnormal trigger positions by determining the position of the billet in the image, and removing underexposed images with insufficient brightness based on image analysis.

[0039] In some implementations, the extraction of contours and morphological processing includes: using a semantic segmentation model to identify the end face region of the target image, obtaining the corresponding rounded quadrilateral region, binarizing and morphologically processing the segmentation result to eliminate interference, extracting the complete outer contour and performing smoothing processing.

[0040] In some implementations, the geometric area of ​​the end face of the billet is calculated using a connected component analysis algorithm.

[0041] In some implementations, the data association and density calculation includes: assigning a virtual number to each billet, tracking its weighing weight at the rolling mill, calculating the billet density based on the weighing weight, end face area, and fixed length, and simultaneously acquiring data such as the casting speed and steel composition during the production of the billet via network connection.

[0042] In some implementations, the construction logic of the dynamic relational database includes: Establish a mapping relationship between the density and end face area of ​​square billets of the same steel grade, casting speed, and tundish temperature; Once the accumulated data reaches the threshold, the fixed length is calculated directly by calling the historical matching density value.

[0043] In some embodiments, the length adjustment is interlocked with an infrared length adjustment device, an encoder length adjustment device, or a laser length adjustment device.

[0044] In some embodiments, the water-cooled protective cover is provided with a camera aperture with a lens, and the water-cooled protective cover is resistant to high temperatures greater than 800°C.

[0045] During normal continuous casting machine production, the area of ​​the end face of each billet is accurately calculated. After the billet is cut to the standard length, it is hot-sent to the rolling mill for weighing. The weight of each billet is divided by the actual end face area and the standard length at the time of cutting to obtain the density of the billet. The casting speed and steel composition during the production of this billet are obtained through networking. The casting speed, steel composition, standard weight required for billet rolling, and the actual accurately calculated end face area are linked to form a corresponding relationship. Using the actual accurately calculated end face area of ​​the billet as the main condition, the standard length that the billet should be cut into is calculated. The cutting device is guided to cut according to the calculated standard length, and the data is compared and adjusted with the weighing data of the rolling mill.

[0046] This invention primarily addresses the issue of billet density variation errors caused by changes in the cross-sectional area of ​​billets during continuous casting, resulting from variations in steel composition, casting speed, and copper tube wear. It effectively adjusts the fixed length of billets dynamically based on changes in the end-face area, achieving the goal of consistent standard weight for each billet required by the rolling mill, with the standard weight per flow controlled at 2 kg / t.

[0047] This invention has the advantages of being non-contact, high temperature resistant, highly accurate, and highly adaptable. 1. Non-contact: The industrial camera is installed above the billet running, which does not affect the normal production rhythm of the continuous casting machine; 2. High temperature resistance: The industrial camera is installed inside a water-cooled protective cover, so it is not afraid of billets with a surface temperature of over 800 degrees passing underneath it; 3. High accuracy: The area of ​​the blank end face is accurately calculated, providing a reliable basis for length and weight adjustment; 4. High adaptability: It can be matched with existing infrared length measuring devices, encoder length measuring devices, or laser length measuring devices; 5. Rapid and accurate weight adjustment: Matching the billet weighing weight, the density value of the billet of the same steel grade, the same drawing speed, and the same temperature is automatically matched with the end face area calculation. Based on the required standard weight, the corresponding fixed length is quickly obtained and transmitted to the fixed length cutting equipment to achieve precise fixed length and weight cutting. Through data association and automatic calculation, the fixed length can be quickly and accurately adjusted, improving production efficiency and product quality.

[0048] Implementation Case 1: A method for online length and weight adjustment of billets based on vision technology includes: A high-resolution industrial camera was selected to quickly and clearly acquire images of the end face of the billet. The vision camera is connected to the vision workstation in the control room via a gigabit Ethernet port, and the vision algorithm is controlled by an authorized dongle to run the algorithm software. Considering the available space and to avoid interference from workers and overhead cranes, the vision camera was designed to be mounted above the roller conveyor table. Figure 1 As shown.

[0049] Once the continuously cast billet is in place, the sensor is triggered to send a signal to the camera to take a picture, as shown in the image. Figure 2 As shown, the vision camera is connected to the vision workstation in the field control room via a gigabit network port, and the vision algorithm authorizes the encryption dongle to control the operation of the algorithm software.

[0050] After calculating the end face area of ​​this billet, a virtual number for this billet is determined online, and the weighing data of this billet at the rolling mill is tracked. The weight value of this billet is obtained, divided by the end face area of ​​this billet and the fixed-length cutting length, to obtain the density of this billet. The casting speed and steel grade of this billet during production are obtained online. The real-time weighing value of the billet is compared with the weight of the standard billet to obtain the difference, which is then divided by the end face area of ​​this billet to determine whether to reduce or increase the real-time fixed-length value of the next billet. The new fixed-length calculated by the computer is imported into the fixed-length system, and the next billet is cut according to the new fixed-length, and the same steel is obtained in a cycle. By establishing a database relating the calculated area and weighing data of the billet end face obtained at different times during the same casting pass to the casting speed, we can obtain the density values ​​of billets with different end faces and casting speeds for the same steel grade, along with data on cutting length and rolling weight. Furthermore, by establishing a one-to-one correspondence between the density fluctuation data of this steel grade under the same casting speed conditions and changes in end face area, we can create a logical link for accurately calculating the end face area and precisely adjusting the fixed length under different casting speeds. For the standard single billet weight of this steel grade, as long as the end face area is measured and the casting speed of this billet is interlocked, the fixed length of this billet can be precisely and automatically adjusted.

[0051] Implementation Case 2: The method for online length and weight adjustment of billets based on vision technology includes the following steps: 1. For example Figure 1 As shown, when the billet 2 reaches the positioning point of the 2D camera 7 via the roller conveyor 3, the infrared sensor detects and sends a signal to instruct the 2D camera 7 to take pictures of the end face of the billet 2. Before taking pictures, the internal and external parameter matrix of the equipment is obtained through high-precision camera calibration to eliminate camera distortion and establish the mapping relationship between the image coordinate system and the physical space.

[0052] 2. For example Figure 5 As shown, the server analyzes and calculates the end face area of ​​the billet. The calculation of the end face area of ​​the billet is as follows: Image preprocessing: A multi-layered quality control mechanism is employed: firstly, invalid samples with abnormal trigger positions are filtered out by determining the billet position in the image; secondly, underexposed images with insufficient brightness (too low billet temperature, too dark brightness) are removed based on image analysis. Brightness normalization is then applied to the remaining valid samples to eliminate imaging differences under different billet brightness levels.

[0053] Semantic segmentation: The preprocessed image input is based on a deep semantic segmentation network based on deep learning, which achieves pixel-level segmentation of the end face region through multi-scale feature fusion.

[0054] Contour extraction and area calculation: A semantic segmentation model is used to identify the end face region of the target image, distinguishing the contour region (foreground) from the background and obtaining the corresponding rounded quadrilateral regions. The segmentation results are binarized and morphologically processed to eliminate burr areas. The complete outer contour of the target region is extracted and smoothed. Morphological operations (such as opening and closing operations) are used to eliminate interference such as oxide layer and metal reflection on the surface of the billet, ensuring contour continuity. Morphological edge smoothing is applied to the segmentation results, and the end face geometric area is accurately calculated using a connected component analysis algorithm.

[0055] 3. For example Figure 5As shown, the billet density is calculated based on the standard weight, end face area, and standard length. The calculation formula is: Standard weight / End face area / Standard length = Billet density. When the first billet produced by the continuous casting machine passes through the weighing device of the rolling mill, the weighing weight is fed back. At this time, the actual density of this billet is calculated according to the formula: Actual billet density = Rolling mill weighing weight / End face area / Standard length. Each billet is virtually numbered to prevent confusion. Each billet is recorded online with the casting speed, tundish temperature, and steel grade during production. Based on the actual density value, end face area, and standard weight, the actual adjusted billet length is calculated: Standard weight / End face area / Actual density = Adjusted billet length, and this is correlated one-to-one with the casting speed, tundish temperature, and steel grade during production.

[0056] 4. For example Figure 6 As shown: When the same data accumulates to a certain level, when the 2D camera takes a picture of the billet end face and calculates the end face area, the density data of billets with the same steel grade, the same casting speed, the same tundish temperature, and the same standard weight can be extracted from the database. The required fixed length can be calculated, and the fixed length can be modified and adjusted through the server. After the billet is cut to the fixed length by the cutting machine, the rolling weight data is compared with the standard weight to obtain the comparison result, which is then stored in the server database.

[0057] Implementation Case 3: Taking the annual production of 3 million tons of billets by a steel company as an example, the online billet length and weight adjustment method based on vision technology includes: Image acquisition and processing: When the billet reaches the positioning point of the 2D camera through the roller conveyor, infrared detection triggers the camera to capture the end face image. After calibration and distortion elimination, the image is preprocessed to screen out abnormal images. The valid images are normalized in brightness and semantically segmented to extract the contour and calculate the end face area.

[0058] Data Calculation and Correlation: After the first billet is cut, it is sent to the rolling mill for weighing. The actual density is calculated based on the weighing data, and a correlation is established by combining data such as casting speed and steel grade. Each subsequent billet is virtually numbered, and relevant data is recorded.

[0059] Fixed-length adjustment: Based on the difference between real-time weighing data and the standard weight, the fixed-length adjustment value for the next billet is calculated and imported into the fixed-length system for cutting. After accumulating data, density data under the same conditions is quickly extracted from the database to achieve automatic adjustment of the fixed-length.

[0060] Using this method, the company achieved a fixed length and weight control accuracy of 2 kg / t, generating an annual profit of approximately 5 million yuan.

[0061] The main functions of this invention are: This invention uses a high-resolution industrial camera to capture images of the billet end face, calculates the end face area, and combines this with parameters such as rolling mill weight, casting speed, and steel composition to establish a dynamic relational database. This allows for dynamic adjustment of the billet cutting length, solving the problem of insufficient cutting and weight accuracy caused by variations in steel composition and casting speed in existing technologies. This achieves high-precision cutting and weight control. By applying machine vision technology to end face area detection and calculation, and introducing it into the cutting and weight control of continuously cast billets, the invention dynamically adjusts the billet length based on changes in the billet end face area, achieving the goal of consistent standard weight for each billet required by the rolling mill, reaching a control accuracy target of 2 kg / t.

[0062] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. The device embodiments described above are merely illustrative; for example, the division of units can be a logical functional division, and in actual implementation, there may be other division methods. For instance, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the displayed or discussed mutual coupling, direct coupling, or communication connection may be through some interfaces; the indirect coupling or communication connection between units or modules may be electrical or other forms.

[0063] The units described as separate components may or may not be physically separate. Similarly, the components of the control device may or may not be physical units; they may be located in one place or distributed across multiple units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0064] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A method for online length and weight adjustment of billets based on vision technology, characterized in that: include: Image acquisition: Real-time acquisition of images of the end face of the continuously cast billet using a camera; Image preprocessing and analysis: Preprocessing and semantic segmentation of the end face image to accurately calculate the area of ​​the billet end face; Data association and density calculation: Calculate the density of the current billet based on the weighing weight, end face area and fixed length, and obtain the rolling weight, casting speed and steel composition of the current billet through the network; Fixed length adjustment: Construct a dynamic correlation database of "end face area - drawing speed - tundish temperature - steel grade - density"; Based on the target fixed weight value, real-time end face area and matching density value, dynamically adjust the fixed length of the next billet, and transmit the adjusted fixed length to the cutting equipment for precise cutting.

2. The method for online length and weight adjustment of billets based on vision technology according to claim 1, characterized in that: The image acquisition includes: an infrared sensor acquiring the billet's arrival signal and transmitting the signal to a high-resolution industrial camera, triggering the high-resolution industrial camera to capture images of the billet's end face.

3. The method for online length and weight adjustment of billets based on vision technology according to claim 2, characterized in that: The high-resolution industrial camera is installed inside a water-cooled protective cover above the roller conveyor table.

4. The method for online length and weight adjustment of billets based on vision technology according to claim 3, characterized in that: The image acquisition also includes: obtaining the internal and external parameter matrix of the equipment through high-resolution industrial camera calibration before shooting, eliminating camera distortion, and establishing the mapping relationship between the image coordinate system and physical space.

5. The method for online length and weight adjustment of billets based on vision technology according to claim 1, characterized in that: The image preprocessing and analysis includes: preprocessing the captured image, using a multi-quality processing mechanism to filter out invalid samples and underexposed images, and performing brightness normalization processing on valid samples; inputting the preprocessed image into a deep semantic segmentation network based on deep learning, performing pixel-level end-face region segmentation, extracting contours and performing morphological processing, and accurately calculating the geometric area of ​​the billet end face.

6. The method for online length and weight adjustment of billets based on vision technology according to claim 5, characterized in that: The multi-quality processing mechanism includes filtering out invalid samples with abnormal trigger positions by judging the position of the billet in the image, and removing underexposed images with insufficient brightness based on image analysis.

7. The method for online length and weight adjustment of billets based on vision technology according to claim 5, characterized in that: The extraction of contours and morphological processing includes: using a semantic segmentation model to identify the end face region of the target image, obtaining the corresponding rounded quadrilateral region, binarizing and morphologically processing the segmentation result to eliminate interference, extracting the complete outer contour and performing smoothing processing; the calculation of the geometric area of ​​the end face of the billet adopts a connected component analysis algorithm.

8. The method for online length and weight adjustment of billets based on vision technology according to claim 1, characterized in that: The data association and density calculation include: assigning a virtual number to each billet, tracking its weighing weight at the rolling mill, and calculating the current billet density according to the formula: actual density = weighing weight / (end face area × fixed length); at the same time, acquiring data such as the casting speed and steel composition of the billet during production via the network.

9. The method for online length and weight adjustment of billets based on vision technology according to claim 1, characterized in that: The construction logic of the dynamic relational database includes: Establish a mapping relationship between the density and end face area of ​​square billets of the same steel grade, casting speed, and tundish temperature; Once the accumulated data reaches the threshold, the fixed length is calculated directly by calling the historical matching density value.

10. The method for online length and weight adjustment of billets based on vision technology according to claim 1, characterized in that: The adjusted fixed length = target fixed weight / (end face area × matching density).

Images