Electrolytic tank short-circuit plate block identification method and device based on infrared image and medium

Abstract

The application provides an electrolytic cell short-circuit plate block recognition method and device based on an infrared image, and a medium, which comprises the following steps: acquiring all the plate blocks in the electrolytic cell in real time to obtain an infrared image to be recognized; adopting a YOLOv8-OBB network model to extract a short-circuit feature of the infrared image to be recognized, so as to locate a short-circuit plate block; judging whether the short-circuit plate block is located in a rectangular detection region constructed in advance, if yes, acquiring an intersection point of the short-circuit plate block and a boundary line segment in the rectangular detection region, and calculating a plate block number of the short-circuit plate block based on the intersection point to realize recognition of the short-circuit plate block. The application can effectively inhibit complex environmental interference, accurately recognize short-circuit thermal features of different degrees and different forms, has high detection accuracy, low false alarm rate and high calculation efficiency, and can realize reliable, real-time and automatic monitoring of an electrolysis process.

Description

Technical Field

[0001] The embodiments of this application relate to the field of image processing, and in particular to a method, apparatus and medium for identifying short-circuit plates in an electrolytic cell based on infrared images. Background Technology

[0002] In large-scale industrial production such as refining, electrolytic cells typically contain a dense array of anode and cathode plates, immersed in an electrolyte. By applying direct current, target metal ions are reduced and deposited on the cathode surface (e.g., zinc) or dissolved at the anode and deposited as pure metal at the cathode (e.g., copper). The current efficiency and stability of this process have a decisive impact on the yield, quality, and production cost of the final product. However, in actual operation, electrical short circuits between the electrodes within the electrolytic cell are a common and serious problem. Short circuits can be caused by various factors, including but not limited to: insufficient spacing due to manufacturing or installation deviations of the electrodes; physical deformation (bending, twisting) of the electrodes during use; bridging caused by uneven growth and shedding of anode sludge or cathode deposits (e.g., copper nodules); and the accumulation of conductive impurities between the electrodes. Once a short circuit occurs, most of the current bypasses the normal electrochemical reaction path, forming a direct circuit through the short circuit point, generating a strong Joule heating effect.

[0003] Currently, methods for detecting electrode short circuits in industrial settings have many limitations. For example, direct detection by operators using handheld detectors or contact measurement techniques with installed sensors are susceptible to subjective factors or sensor performance issues, making it difficult to meet the accuracy and sensitivity requirements of refined management. Existing technologies also include short-circuit detection algorithms based on infrared image processing, such as fixed or adaptive threshold segmentation and inter-frame difference methods. These methods exhibit poor robustness against complex interference and weak signals, making it difficult to effectively distinguish between real short circuits and background noise or normal hot spots, resulting in high false alarm and false negative rates. Furthermore, some traditional target detection strategies, such as using sliding windows to traverse the image and extracting features from each window for classification, may improve detection accuracy to some extent, but the computational load is enormous, making it difficult to meet the real-time detection requirements of industrial settings. Summary of the Invention

[0004] To address the aforementioned technical problems, this invention proposes a method, device, and medium for identifying short-circuit plates in electrolytic cells based on infrared images. The intelligent analysis method based on infrared thermal imaging enables efficient identification of short circuits in electrolytic cells. This invention can effectively suppress interference from complex environments, accurately identify short-circuit thermal characteristics of different degrees and forms, and possesses high detection accuracy, low false alarm rate, and high computational efficiency, thereby achieving reliable, real-time, and automated monitoring of the electrolysis process.

[0005] In a first aspect, the present invention proposes a method for identifying short-circuit plates in electrolytic cells based on infrared images, specifically including the following steps: Real-time acquisition of infrared images of all plates in the electrolytic cell to be identified; The YOLOv8-OBB network model is used to extract short-circuit features from the infrared image to be identified, so as to locate the short-circuit plate where the short circuit occurs. Determine whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, obtain the intersection point of the short-circuit plate and the boundary line segment of the rectangular detection area, and calculate the plate number of the short-circuit plate based on the intersection point to realize the identification of the short-circuit plate.

[0006] In one embodiment, the step of pre-constructing the rectangular detection region includes: Two plates are selected in the electrolytic cell as the starting plate and the ending plate, respectively, and the area from the starting plate to the ending plate is a rectangular detection area.

[0007] In one embodiment, before the step of acquiring the infrared image of all plates to be identified in the electrolytic cell in real time, the method includes: identifying whether the starting plate and the ending plate of the rectangular detection area are plates at both ends of the electrolytic cell; if so, the area between the plates at both ends of the electrolytic cell is taken as the rectangular detection area; if not, the plate number of the edge plates at both ends of the rectangular detection area and the number of plates in the rectangular detection area are identified, and the area between the two edge plates is taken as the rectangular detection area.

[0008] In one embodiment, the YOLOv8-OBB network model is an improved model that introduces oriented bounding boxes (OBBs) into the YOLOv8 network model, and replaces the C2F module in the YOLOv8 network model with a C2F_CBAM module. The CBAM module includes a channel attention (CAM) module and a spatial attention (SAM) module. The infrared image to be identified is processed by the C2F module in the C2F_CBAM module to extract feature maps, and then input to the CBAM module. The feature map is directly adjusted by adjusting the feature channel weights. Then, based on the SAM module, the spatial location of the thermal feature region generated by the short circuit in the image to be identified is adaptively obtained to locate the short-circuited plate.

[0009] In one embodiment, the loss function of the YOLOv8-OBB network model is generated by introducing a direction consistency constraint. Specifically, the global direction of the electrolytic cell can be obtained by calculating the direction of the initial or final plate. And setting the orientation of each plate in all the plates of the electrolytic cell to Then the loss function of the direction consistency constraint. Represented as: in, To detect the number of short-circuited plates.

[0010] In one embodiment, the step of using the YOLOv8-OBB network model to extract short-circuit features from the infrared image to be identified in order to locate the short-circuit plate where the short circuit has occurred also includes processing the results after short-circuit feature extraction by non-maximum suppression (NMS), and generating a detection result corresponding to each plate after removing duplicate alarms.

[0011] In one embodiment, the step of determining whether the short-circuited plate is located within a pre-constructed rectangular detection area, and if so, obtaining the intersection point of the short-circuited plate with the boundary line segment of the rectangular detection area, and calculating the plate number of the short-circuited plate based on the intersection point to achieve the identification of the short-circuited plate specifically includes: Calculate the ratio L of the intersection area of ​​the oriented bounding box OBB of the short-circuit plate and the rectangular detection region in the area of ​​the short-circuit plate. If the ratio L is greater than a set threshold, the short-circuit plate is located in the pre-constructed rectangular detection region. Obtain the intersection point of the short-circuit plate with the boundary line segment in the rectangular detection area. If the vector product of the intersection point and the two endpoints of the boundary line segment in the same direction is greater than 0, then the intersection point is located on the boundary line segment. The plate number of the short-circuit plate is calculated to identify the short-circuit plate. If the intersection point is not on the boundary line segment, then the short-circuit plate is either the starting plate or the ending plate.

[0012] In one embodiment, the step of calculating the segment number of the short-circuit segment specifically includes: dividing the boundary segment into N equally divided intervals, each interval corresponding to a segment, and calculating the distance from the intersection point to the node of the starting segment corresponding to the boundary segment. and the distance of the boundary line segment. The section number M of the short-circuit section is calculated using the following formula: in, This indicates rounding down, and K represents the starting plate number.

[0013] In a second aspect, the present invention also provides an infrared image-based electrolytic cell short-circuit plate identification device, used to implement the infrared image-based electrolytic cell short-circuit plate identification method as described in the first aspect, comprising: The acquisition module is used to acquire infrared images of all plates in the electrolytic cell in real time. The positioning module is used to extract short-circuit features from the infrared image to be identified using the YOLOv8-OBB network model in order to locate the short-circuit plate where the short circuit has occurred. The identification module is used to determine whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, it obtains the intersection point of the short-circuit plate and the boundary line segment of the rectangular detection area, and calculates the plate number of the short-circuit plate based on the intersection point to realize the identification of the short-circuit plate.

[0014] Thirdly, the present invention also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect.

[0015] The short-circuit plate localization solution provided by this invention, based on the OBB (Oriented Bounding Box) algorithm, achieves high-precision localization of target plates by accurately calculating the intersection area ratio, intersection point location, and plate number of the short-circuit plate and the rectangular detection area. This method brings the following significant benefits: by calculating the intersection area ratio and intersection point location, it ensures that only short-circuit plates located within the rectangular detection area are located, effectively avoiding false alarms and improving the reliability of the localization results. The technical solution of this invention can quickly determine the plate number, making it suitable for real-time detection scenarios and significantly reducing the time cost of short-circuit localization. Furthermore, the intersection ratio threshold and plate division method can be flexibly adjusted according to actual needs, enabling this method to adapt to different industrial detection environments and application scenarios. Attached Figure Description

[0016] Figure 1 This is a flowchart of the method for identifying short-circuit plates in an electrolytic cell based on infrared images proposed in this invention.

[0017] Figure 2 This is a schematic diagram of the rectangular detection area in the infrared image-based electrolytic cell short-circuit plate identification method proposed in this invention.

[0018] Figure 3 This is a schematic diagram of the C2F_CBAM module in the YOLOv8-OBB network model of the short-circuit plate identification method for electrolytic cells based on infrared images proposed in this invention.

[0019] Figure 4 This is a schematic diagram showing the short-circuit plate located within a rectangular detection area in the electrolytic cell short-circuit plate identification method based on infrared images proposed in this invention.

[0020] Figure 5 This is a schematic diagram showing that the short-circuit plate is located outside the rectangular detection area in the short-circuit plate identification method for electrolytic cells based on infrared images proposed in this invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0022] To achieve the objectives of this invention, the infrared image acquisition system provided by this invention includes: Infrared detector: Converts infrared signals into electrical signals.

[0023] Infrared detector drive circuit: converts the electrical signal output by the infrared detector into a raw 14-bit parallel digital signal.

[0024] FPGA processing circuit: Processes the input 14-bit parallel digital signal in real time, performs infrared image preprocessing, and converts the infrared image energy data into grayscale data using the AGC algorithm. Outputs a digital video stream conforming to the protocol requirements of the USB interface chip.

[0025] ARM processing circuit: It is programmed with a supporting embedded Linux system and image and video computer vision processing service programs. It converts infrared AD data into RTSP and outputs RTSP protocol infrared video streams.

[0026] Figure 1 This is a flowchart of the method for identifying short-circuit plates in an electrolytic cell based on infrared images proposed in this invention. Figure 1 As shown, this invention proposes a method for identifying short-circuit plates in electrolytic cells based on infrared images, specifically including the following steps: S101. Acquire the infrared images of all plates to be identified in the electrolytic cell in real time; S102. Use the YOLOv8-OBB network model to extract short-circuit features from the infrared image to be identified, so as to locate the short-circuit plate where the short circuit occurs; S103. Determine whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, obtain the intersection point of the short-circuit plate and the boundary line segment in the rectangular detection area, and calculate the plate number of the short-circuit plate based on the intersection point to realize the identification of the short-circuit plate.

[0027] In this embodiment, as Figure 2 As shown, before performing infrared thermal imaging detection, the area to be detected above the electrolytic cell must first be determined to ensure that the infrared camera can fully cover the electrode area where a short circuit may occur. The specific setup method is as follows: like Figure 2As shown, the area to be detected is defined as a rectangular detection area ABCD, where AD is the starting plate, BC is the ending plate, and the number of plates is N. The starting plate AD and the ending plate BC are located at opposite ends of the electrolytic cell, and N represents the total number of plates uniformly distributed between AD and BC. This rectangular area covers all plates in the electrolytic cell that may experience short circuits.

[0028] When the field of view of a single preset point (i.e., the shooting position of a single infrared camera) cannot cover the entire trench surface, multiple preset points can be stitched together. The detection area of ​​each preset point should follow a unified setting logic, that is, it should include the starting plate AD, the ending plate BC, and the corresponding number of plates N. There should be appropriate overlap between the detection areas of multiple preset points to avoid detection blind spots and ensure the consistency of subsequent short-circuit plate positioning. To achieve this goal, the shooting range of each preset point can be calculated based on the field of view (FOV) and installation height of the infrared camera, and the area stitching can be completed through geometric transformation.

[0029] Therefore, before acquiring the infrared image to be identified, a rectangular detection area needs to be established first. Then, the specific range of the rectangular detection area needs to be determined. If a single preset point completely covers the area, the starting and ending plates are the plates at both ends of the electrolytic cell, and the starting and ending plates are numbered from 0 to 100. The number of any short-circuited plate located within the area directly corresponds to its number within 0 to 100. If a single preset point cannot completely cover the area, and multiple preset points are used, if the detection area formed by the starting and ending plates at both ends of the detection area corresponds to the number 26 to 48, and if the short-circuited plate detected in this area is the 5th plate, then the actual corresponding plate number is 26 + 5 = 31, that is, the short-circuited plate number output externally is 31.

[0030] like Figure 3As shown, in this embodiment, the YOLOv8-OBB network model is an improved model that introduces oriented bounding boxes (OBBs) on the basis of the YOLOv8 network model. The C2F (Cross Stage Partial) module in the YOLOv8 network model is replaced with the C2F_CBAM module. The CBAM (Convolutional Block Attention Module) module includes a channel attention (CAM) module and a spatial attention (SAM) module. The infrared image to be identified is input to the CBAM module after feature map extraction by the C2F module in the C2F_CBAM module. Since the image to be detected is an infrared image, in this embodiment, it is not necessary to go through the CAM module. The feature channel weights of the feature map are directly adjusted, which can increase the inference speed of the model. Then, based on the SAM module, the spatial location of the thermal feature region generated by the short circuit in the image to be identified is obtained to locate the short-circuited block where the short circuit occurs.

[0031] In this embodiment, to improve the detection capability for small-sized short-circuit targets, a smaller target detection output is added to the YOLOv8 OBB Head network. Short-circuit thermal features in infrared images may be small in size or close together, making them difficult for traditional detection heads to distinguish, especially when multiple consecutive electrode plates are short-circuited. The implementation method for the new detection head is as follows: a higher-resolution feature layer is introduced into the Feature Pyramid Network (FPN), and an additional detection head is built on top of this. This detection head focuses on detecting smaller-scale targets, thereby improving the model's detection rate for small short-circuit features or adjacent targets, while maintaining the ability to detect medium and large targets.

[0032] In this embodiment, considering that the short-circuit directions of the electrode plates in the electrolytic cell are basically consistent (usually along the length of the electrolytic cell), the loss function of the YOLOv8-OBB network model is generated by introducing a direction consistency constraint. Specifically, the global direction of the electrolytic cell can be obtained by calculating the direction of the initial plate or the termination plate. And setting the orientation of each plate in all the plates of the electrolytic cell to Then the loss function of the direction consistency constraint. Represented as: in, To detect the number of short-circuited plates, this loss term penalizes the deviation between the detection direction and the global direction, causing the model to tend to predict bounding box directions with higher consistency, thereby reducing the false alarm rate of false hotspots such as cover plate wrinkles. The final loss function is a weighted sum of classification loss, boundary regression loss, distribution focal loss, orientation loss, and consistency loss.

[0033] It should be noted that after using the YOLOv8-OBB network model to extract short-circuit features from the infrared image to be identified and completing target detection, it is necessary to locate the specific section where the short circuit occurred based on the detection results. In this embodiment, the results after short-circuit feature extraction are processed by NMS (Non-Maximum Suppression), and after removing duplicate alarms, a detection result corresponding to each section is generated.

[0034] In this embodiment, when a short-circuit plate is identified, it is necessary to identify the short-circuit plate by number. Specifically, this process involves determining whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, the intersection point of the short-circuit plate and the boundary line segment in the rectangular detection area is obtained, and the plate number of the short-circuit plate is calculated based on the intersection point to achieve the identification of the short-circuit plate.

[0035] In this embodiment, the short-circuit plate output by the YOLOv8-OBB network model has a directional bounding box (OBB), which is used to determine whether the short-circuit plate is located in the rectangular detection area.

[0036] like Figure 4 and 5 As shown in the preceding content, Figure 4 The diagram illustrates the detection of a short-circuited plate. As can be seen, the detected short-circuited plate is located outside the rectangular detection area. Figure 5 This shows that the short-circuit plate is located within the rectangular detection area.

[0037] In this embodiment, the location of the short-circuit plate within the rectangular detection area is determined by calculating the ratio of the intersection area between the short-circuit plate and the rectangular detection area. The specific steps are as follows: (1) Based on the center point of the short-circuit plate ,width ,high And direction D, generate its oriented bounding box OBB, (2) Calculate the intersection area of ​​the short-circuit plate OBB and the rectangular detection area ABCD. .

[0038] (3) Calculate the intersection area The total area of ​​the short-circuit plate (i.e. × ) proportion : A predetermined threshold T is set. If the intersection ratio is lower than the threshold, the short-circuited plate is determined to be outside the rectangular detection area, and the subsequent positioning steps are terminated; if the intersection ratio is greater than or equal to the threshold, the subsequent steps continue. In this embodiment, the above process identifies whether the detected short-circuited plate is located within the rectangular detection area. If it is located within the rectangular detection area, i.e., it is a short-circuited plate that meets the intersection ratio condition, the intersection point of the short-circuited plate and the boundary line segment in the rectangular detection area is obtained. If the vector product of the intersection point and the two endpoints of the boundary line segment in the same direction is greater than 0, then the intersection point is located on the boundary line segment. The plate number of the short-circuited plate is calculated to identify the short-circuited plate. If the intersection point is not on the boundary line segment, then the short-circuited plate is either the starting plate or the ending plate.

[0039] The following example illustrates how, in this embodiment, segment AB within the rectangular detection area is used as the boundary line segment. The specific method is as follows: (1) Taking the center point of the short-circuit plate As the base point, direction As a vector, construct a line from The ray that sets off.

[0040] (2) Calculate the intersection point of the ray and the upper boundary line segment AB of the rectangular detection area. .

[0041] (3) Intersection point Apply constraints to ensure that it lies within the range of line segment AB, i.e. The coordinates must satisfy: like If the short circuit occurs outside the AB range, it needs to be determined whether the short-circuited plate is located on the starting or ending plate: if If the intersection is successful, it is considered the termination plate; otherwise, it is considered the starting plate. A valid intersection point is obtained. Then, based on its position on the upper boundary line segment AB of the rectangular detection area, the segment number where the short-circuit target is located is calculated. The specific steps are as follows: (1) Divide the line segment AB evenly into N equal intervals (N is the number of segments in the area), with each interval corresponding to one segment; (2) Calculate Distance from point to the starting point A of line segment : (3) Calculate the distance of line segment AB. : Sure The section number of the section where the point is located is M: in, This indicates rounding down, and K represents the starting plate number.

[0042] Through the above steps, the OBB detection results of the short-circuit plate are accurately mapped to the specific plate number within the rectangular detection area, thus completing the location of the short-circuit plate.

[0043] This embodiment also provides an infrared image-based electrolytic cell short-circuit plate identification device, used to implement the infrared image-based electrolytic cell short-circuit plate identification method as described in the first aspect, including: The acquisition module is used to acquire infrared images of all plates in the electrolytic cell in real time. The positioning module is used to extract short-circuit features from the infrared image to be identified using the YOLOv8-OBB network model in order to locate the short-circuit plate where the short circuit has occurred. The identification module is used to determine whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, it obtains the intersection point of the short-circuit plate and the boundary line segment of the rectangular detection area, and calculates the plate number of the short-circuit plate based on the intersection point to realize the identification of the short-circuit plate.

[0044] This embodiment also provides a computer-readable storage medium. One or more non-volatile computer-readable storage media containing computer-executable instructions, which, when executed by one or more processors, cause the processors to perform the steps of a method for filtering trailing point clouds. A computer program product containing instructions, which, when run on a computer, causes the computer to perform a method for filtering trailing point clouds.

[0045] Any references to memory, storage, database, or other media used in the embodiments of this application may include non-volatile and / or volatile memory. Suitable non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM), which is used as external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM).

[0046] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0047] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A method for identifying short-circuited plates in an electrolytic cell based on infrared images, characterized in that, Specifically, the steps include the following: Two plates are selected in the electrolytic cell as the starting plate and the ending plate, and the area from the starting plate to the ending plate is a rectangular detection area. Identify whether the starting and ending plates of the rectangular detection area are plates at both ends of the electrolytic cell. If so, the area between the plates at both ends of the electrolytic cell is taken as the rectangular detection area. If not, identify the plate number of the edge plates at both ends of the rectangular detection area and the number of plates in the rectangular detection area, and take the area between the two edge plates as the rectangular detection area. Real-time acquisition of infrared images of all plates in the electrolytic cell to be identified; A YOLOv8-OBB network model is used to extract short-circuit features from the infrared image to be identified, in order to locate the short-circuited plate. The YOLOv8-OBB network model is an improved model based on the YOLOv8 network model, which introduces oriented bounding boxes (OBB). The C2F module in the YOLOv8 network model is replaced with a C2F_CBAM module, which includes a channel attention (CAM) module and a spatial attention (SAM) module. The infrared image to be identified is processed by the C2F module in the C2F_CBAM module to extract feature maps, and then input into the CBAM module. The feature map is directly adjusted by adjusting the feature channel weights. Then, based on the SAM module, the spatial location of the thermal feature region generated by the short circuit in the image to be identified is adaptively obtained to locate the short-circuited plate. In this case, the orientation consistency constraint is introduced according to the orientation of the plate in the electrolytic cell to generate the loss function of the YOLOv8-OBB network model, which is used to predict the orientation of the bounding box with higher consistency. Determine whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, obtain the intersection point of the short-circuit plate and the boundary line segment of the rectangular detection area, and calculate the plate number of the short-circuit plate based on the intersection point to realize the identification of the short-circuit plate. The loss function of the YOLOv8-OBB network model is generated by introducing a directional consistency constraint. Specifically, the global direction of the electrolyzer can be obtained by calculating the direction of the initial or final plate. And setting the orientation of each plate in all the plates of the electrolytic cell to Then the loss function of the direction consistency constraint. Represented as: in, To detect the number of short-circuited plates.

2. The method for identifying short-circuit plates in an electrolytic cell based on infrared images according to claim 1, characterized in that, The step of using the YOLOv8-OBB network model to extract short-circuit features from the infrared image to be identified in order to locate the short-circuit plate where the short circuit has occurred also includes processing the results after short-circuit feature extraction by non-maximum suppression (NMS), and generating detection results corresponding to each plate after removing duplicate alarms.

3. The method for identifying short-circuit plates in an electrolytic cell based on infrared images according to claim 2, characterized in that, The steps of determining whether the short-circuited plate is located within a pre-constructed rectangular detection area, and if so, obtaining the intersection point of the short-circuited plate with the boundary line segment of the rectangular detection area, and calculating the plate number of the short-circuited plate based on the intersection point to achieve the identification of the short-circuited plate, specifically include: Calculate the ratio L of the intersection area of ​​the oriented bounding box OBB of the short-circuit plate and the rectangular detection region in the area of ​​the short-circuit plate. If the ratio L is greater than a set threshold, the short-circuit plate is located in the pre-constructed rectangular detection region. Obtain the intersection point of the short-circuit plate with the boundary line segment in the rectangular detection area. If the vector product of the intersection point and the two endpoints of the boundary line segment in the same direction is greater than 0, then the intersection point is located on the boundary line segment. The plate number of the short-circuit plate is calculated to identify the short-circuit plate. If the intersection point is not on the boundary line segment, then the short-circuit plate is either the starting plate or the ending plate.

4. The method for identifying short-circuit plates in an electrolytic cell based on infrared images according to claim 3, characterized in that, The steps for calculating the segment number of the short-circuit segment specifically include: dividing the boundary segment into N equally divided intervals, each interval corresponding to one segment, and calculating the distance from the intersection point to the node of the starting segment corresponding to the boundary segment. and the distance of the boundary line segment. The section number M of the short-circuit section is calculated using the following formula: in, This indicates rounding down, and K represents the starting plate number.

5. An infrared image-based electrolytic cell short-circuit plate identification device, used to implement the infrared image-based electrolytic cell short-circuit plate identification method as described in any one of claims 1-4, characterized in that, include: The acquisition module is used to acquire infrared images of all plates in the electrolytic cell in real time. The positioning module is used to extract short-circuit features from the infrared image to be identified using the YOLOv8-OBB network model in order to locate the short-circuit plate where the short circuit has occurred. The identification module is used to determine whether the short-circuit plate is located in a pre-constructed rectangular detection area. If so, it obtains the intersection point of the short-circuit plate and the boundary line segment of the rectangular detection area, and calculates the plate number of the short-circuit plate based on the intersection point to realize the identification of the short-circuit plate.

6. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1 to 4.

Images