Battery string detection method, device and equipment

Abstract

The invention provides a battery string detection method, device and equipment. The method comprises the following steps: acquiring a to-be-detected image shot in the transmission process of a target battery string; in response to the acquired to-be-detected image, splicing the to-be-detected image with a battery string image of the target battery string to obtain a spliced image, the battery string image being formed by splicing the shot to-be-detected image in the transmission process of the target battery string; a battery piece sub-image is obtained from the spliced image, the battery piece sub-image comprises a target battery piece, and the target battery piece is a battery piece which is not detected; and performing defect detection on the target battery piece based on the battery piece sub-image. According to the scheme, the appearance of the battery string is detected while the battery string collects data, so that the detection time is effectively shortened, and the detection efficiency and the real-time performance are improved.

Description

Technical Field

[0001] The present application relates to the field of machine vision technology, and in particular to a battery string detection method, device and equipment. Background Art

[0002] The string soldering process of the battery string can include the following steps: 1. Clean the string soldering template before string soldering to prevent tin slag and tin wire from falling and cracking the battery cells. 2. During the string soldering process, take the battery cells from the waiting string soldering area and distribute them to the string soldering template. Distribute the battery cells to the string soldering template from left to right, and then solder each battery cell from the right side to the left side of the string soldering template. 3. Clean the tin wire and tin slag on the battery string after the string soldering is completed.

[0003] In the process of string welding, in addition to welding each battery cell, a cutting process may also be involved. For example, when a battery string includes 15 battery cells, cutting is performed between the 15th battery cell and the 16th battery cell, so that the 1st to 15th battery cells form a battery string. Cutting is performed between the 30th battery cell and the 31st battery cell, so that the 16th to 30th battery cells form a battery string, and so on.

[0004] During the string welding process of battery strings, various defects are prone to occur in the battery strings, such as damaged appearance, dirty appearance, poor welding quality, etc. Therefore, it is necessary to perform appearance inspection on the battery strings, promptly detect defective battery strings, assist in the inspection and adjustment of the string welding equipment, and prevent defective battery strings from entering the market.

[0005] In order to perform visual inspection on the battery string, it is necessary to use manual visual inspection to observe whether the battery string has defects. However, since the battery string is long and there are many defect inspection items, manual visual inspection cannot detect all defects of the battery string, and the inspection results of manual visual inspection may have errors, that is, the inspection results are inaccurate. Manual visual inspection requires contact with the battery string, which may cause secondary damage to the battery string. Summary of the invention

[0006] The present application provides a method for detecting a battery string, wherein a target battery string includes at least one battery cell, and the method includes: acquiring an image to be detected that is taken during the transmission of the target battery string; in response to acquiring the image to be detected, splicing the image to be detected with a battery string image of the target battery string to obtain a spliced ​​image, wherein the battery string image is spliced ​​from the images to be detected that are taken during the transmission of the target battery string; acquiring a battery cell sub-image from the spliced ​​image, wherein the battery cell sub-image includes a target battery cell, and the target battery cell is a battery cell that has not been inspected; and performing defect detection on the target battery cell based on the battery cell sub-image.

[0007] The present application provides a detection device for a battery string, wherein a target battery string includes at least one battery cell, and comprises: an acquisition module, used to acquire an image to be detected taken during the transmission of the target battery string; a splicing module, used to splice the image to be detected with a battery string image of the target battery string in response to acquiring the image to be detected to obtain a spliced ​​image, wherein the battery string image is spliced ​​from the images to be detected taken during the transmission of the target battery string; a battery cell sub-image is acquired from the spliced ​​image, wherein the battery cell sub-image includes a target battery cell, and the target battery cell is a battery cell that has not been detected; and a detection module, used to perform defect detection on the target battery cell based on the battery cell sub-image.

[0008] The present application provides a battery string detection system, including: an image acquisition device, a battery string detection device, a control device, an MES device and a centralized control device; wherein: the image acquisition device is used to acquire an image to be detected during the transmission process of a target battery string; the battery string detection device is used to execute the above-mentioned battery string detection method based on the image to be detected, and obtain a defect detection result corresponding to the target battery string; the control device is used to control the transmission device and the image acquisition device; the MES device is used to obtain the defect detection result corresponding to the target battery string, and process the target battery string based on the defect detection result; the centralized control device is used to obtain the defect detection result corresponding to the target battery string, so that a user can manually detect whether the target battery string has defects based on the defect detection result, obtain the manual detection result of the target battery string, and return the manual detection result to the battery string detection device.

[0009] The present application provides an electronic device, comprising: a processor and a machine-readable storage medium, wherein the machine-readable storage medium stores machine-executable instructions that can be executed by the processor; the processor is used to execute the machine-executable instructions to implement the above-mentioned battery string detection method.

[0010] The present application provides a computer program product, which may include a computer program. When the computer program is executed by a processor, the above-mentioned battery string detection method is implemented.

[0011] The present application provides a machine-readable storage medium, which stores machine-executable instructions that can be executed by a processor; wherein the processor is used to execute the machine-executable instructions, and when the machine-executable instructions are executed, the above-mentioned battery string detection method is implemented.

[0012] It can be seen from the above technical solutions that in the embodiment of the present application, the image of the battery string is captured by a camera, and whether the battery string has defects is determined based on the image, so that the defective battery string can be discovered in time. The defects of the battery string are detected by image, and the fast and accurate defect detection of extra-long battery strings and regular-length battery strings is achieved, avoiding the subjectivity and errors of manual visual inspection, effectively improving the quality and efficiency of detection, and reducing labor costs. The detection process does not require contact with the battery string, and will not cause secondary damage to the battery string, thus achieving non-destructive detection. It can be applied to the appearance inspection of battery strings of various sizes. While collecting data on the battery string, the appearance of the battery string can be inspected, which effectively shortens the detection time and improves the detection efficiency and real-time performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 is a flow chart of a method for detecting a battery string in one embodiment of the present application;

[0014] Figure 2 is a schematic diagram of the system structure of battery string detection in one embodiment of the present application;

[0015] Figure 3A is a schematic diagram of an appearance image of an ultra-long battery string in one embodiment of the present application;

[0016] Figure 3B is a schematic diagram of an appearance image of a battery string of conventional length in one embodiment of the present application;

[0017] Figure 4 is a schematic diagram of the structure of a battery string detection system in one embodiment of the present application;

[0018] Figure 5 is a flow chart of a method for detecting a battery string in one embodiment of the present application;

[0019] Figure 6 is a flow chart of a method for detecting a battery string in one embodiment of the present application;

[0020] Figure 7 It is a schematic diagram of the battery string splicing process in one embodiment of the present application;

[0021] Figure 8 is a schematic structural diagram of a battery string detection device in one embodiment of the present application;

[0022] Fig. 9 It is a hardware structure diagram of an electronic device in one embodiment of the present application. DETAILED DESCRIPTION

[0023] In an embodiment of the present application, a method for detecting a battery string is provided, wherein the target battery string includes at least one battery cell. Figure 1FIG. 4 is a flow chart of the method, which may include:

[0024] Step 101: Acquire an image to be detected that is taken during the transmission process of a target battery string.

[0025] Step 102 , in response to obtaining the image to be detected, splicing the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image, wherein the battery string image is spliced ​​by the images to be detected taken during the transmission process of the target battery string.

[0026] Step 103: Obtain a cell sub-image from the stitched image, the cell sub-image may include a target cell, and the target cell is a cell that has not been inspected. For example, the target cell may be a complete cell that has not been inspected.

[0027] Step 104: perform defect detection on the target cell based on the cell sub-image.

[0028] Exemplarily, the image to be detected includes an image to be detected of the upper surface of the target battery string acquired by the front camera and / or an image to be detected of the lower surface of the target battery string acquired by the back camera; the front camera is located above the target battery string, and the back camera is located below the target battery string. The image to be detected on the upper surface is used to perform defect detection on the upper surface of the target battery string; the image to be detected on the lower surface is used to perform defect detection on the lower surface of the target battery string. For example, based on the image to be detected on the upper surface of the target battery string acquired by the front camera, steps 101 to 104 can be used to obtain the defect detection result of the upper surface of the target battery string. For example, based on multiple images to be detected on the upper surface of the target battery string, the defect detection results of all target battery cells can be obtained, and the defect detection results of all target battery cells constitute the defect detection result of the upper surface of the target battery string. For example, based on the image to be detected on the lower surface of the target battery string acquired by the back camera, steps 101 to 104 can be used to obtain the defect detection result of the lower surface of the target battery string. For example, based on the image to be inspected of the upper surface of the target battery string obtained by the front camera, the defect detection result of the upper surface of the target battery string can be obtained. Based on the image to be inspected of the lower surface of the target battery string obtained by the rear camera, the defect detection result of the lower surface of the target battery string can be obtained, that is, the defect detection results of the upper surface and the lower surface are obtained at the same time.

[0029] Exemplarily, the front detection data of the upper surface of the target battery string can also be obtained, and the front detection data includes information on whether there are defects on the upper surface of the target battery string. If there are defects on the upper surface, the front detection data also includes the defect type and defect area of ​​the upper surface of the target battery string. In addition, the back detection data of the lower surface of the target battery string can also be obtained. The back detection data includes information on whether there are defects on the lower surface of the target battery string. If there are defects on the lower surface, the back detection data also includes the defect type and defect area of ​​the lower surface of the target battery string. The front detection data and the back detection data are summarized to obtain the appearance detection data of the target battery string, so that the user can manually detect whether there are defects on the target battery string based on the appearance detection data, and obtain the manual detection results of the target battery string.

[0030] Exemplarily, after defect detection of the target cell based on the cell image, the defect detection result of the target cell string can also be obtained, and the defect detection result includes at least one of the following: the result obtained when defect detection of the target cell based on the cell image (i.e., the detection result of step 104), front detection data, back detection data, appearance detection data, and manual detection results. Then, an operation control instruction can be generated according to the defect detection result, and the operation control instruction is used to control the target cell string to flow in a direction corresponding to the defect detection result. For example, if the defect detection result is that there is no defect, the target cell string is controlled to flow to the market. If the defect detection result is that there is a defect, the target cell string is controlled not to flow to the market.

[0031] Exemplarily, in the transportation process of multiple battery strings (the battery string sequence to be inspected may include multiple battery strings, the transportation process of multiple battery strings is the transportation process of the battery string sequence, and the battery string sequence is transported by a transmission device), each battery string passes through the field of view of the camera in turn, and the target battery string is any battery string (i.e., the battery string currently in the field of view of the camera). When the target battery string passes through the field of view of the camera, the camera can acquire an image to be inspected for the target battery string.

[0032] Exemplarily, the image to be detected is spliced ​​with the battery string image of the target battery string to obtain a spliced ​​image, which may include but is not limited to: determining whether the number of spliced ​​images of the battery string image is 0; if so, and the image to be detected has the string head feature of the target battery string, the image to be detected may be updated to the battery string image of the target battery string, and the number of spliced ​​images may be updated to 1; if not, and the number of spliced ​​images does not reach the number threshold, the image to be detected may be spliced ​​with the battery string image of the target battery string to obtain a spliced ​​image, and the spliced ​​image may be updated to the battery string image of the target battery string, and the number of spliced ​​images may be increased by 1. For example, the number threshold may be determined based on the configured maximum splicing number.

[0033] Exemplarily, after determining whether the number of stitched images of the battery string image is 0, if not, and the number of stitched images reaches a number threshold, it can be determined whether the image to be detected has a string tail feature of the target battery string; if the string tail feature does not exist, the image to be detected and the battery string image of the target battery string can be stitched together to obtain a stitched image, and the stitched image is updated to the battery string image of the target battery string, and the number of stitched images is increased by 1; if the string tail feature exists, the image to be detected and the battery string image of the target battery string can be stitched together to obtain a stitched image, and the stitched image is updated to the complete battery string image of the target battery string, the complete battery string image is output, and the number of stitched images is updated to 0.

[0034] Exemplarily, the spliced ​​image is updated to a complete battery string image of the target battery string. Before outputting the complete battery string image, if there is a string end feature, it can also be determined whether a string end signal is received, and the string end signal can indicate that the last battery cell of the target battery string is transported to the camera's field of view; if so, the spliced ​​image can be updated to a complete battery string image of the target battery string, and the operation of outputting the complete battery string image can be executed; if not, abnormal information can be output.

[0035] Exemplarily, the stitched image is updated to a complete battery string image of the target battery string. After the complete battery string image is output, a sub-image of each battery cell can be determined from the complete battery string image, and the defect detection result of the battery cell can be added to the sub-image of the battery cell.

[0036] Exemplarily, before splicing the image to be detected with the battery string image of the target battery string to obtain the spliced ​​image, it can also be determined whether the image to be detected is a valid image; if it is a valid image, the operation of splicing the image to be detected with the battery string image of the target battery string to obtain the spliced ​​image is performed; if it is an invalid image, the image to be detected is discarded; wherein, if the brightness mean value of the battery cell area in the image to be detected is less than the brightness threshold, at least one of the presence of a string head feature in the image to be detected and the presence of a string tail feature in the image to be detected is established, then the image to be detected is a valid image; otherwise, the image to be detected is an invalid image.

[0037] Exemplarily, the edge pixel points of the target battery string can also be determined from the image to be detected. If a first edge straight line and a second edge straight line are fitted based on these edge pixel points, and the distance between the first edge straight line and the second edge straight line is greater than the distance threshold, and the average brightness of the transition area between the first edge straight line and the second edge straight line is greater than the first brightness threshold, and the average brightness of the battery cell area in the image to be detected is less than the second brightness threshold, then it can be determined that the image to be detected has string head features and string tail features; wherein the second brightness threshold can be less than the first brightness threshold.

[0038] Exemplarily, obtaining a cell sub-image from the stitched image may include but is not limited to: inputting the stitched image to a first target detection model, and having the first target detection model output a first predicted position of a main grid line area of ​​the cell; and determining an area corresponding to the first predicted position as the cell sub-image.

[0039] Exemplarily, defect detection of a target cell based on a cell sub-image may include, but is not limited to: inputting the cell sub-image into a second target detection model, and having the second target detection model output a second predicted position and a predicted category; determining an area corresponding to the second predicted position as a defect area of ​​the cell sub-image, and determining the defect type of the defect area based on the predicted category.

[0040] It can be seen from the above technical solutions that in the embodiment of the present application, the image of the battery string is captured by a camera, and whether the battery string has defects is determined based on the image, so that the defective battery string can be discovered in time. The defects of the battery string are detected by image, and the fast and accurate defect detection of extra-long battery strings and regular-length battery strings is achieved, avoiding the subjectivity and errors of manual visual inspection, effectively improving the quality and efficiency of detection, and reducing labor costs. The detection process does not require contact with the battery string, and will not cause secondary damage to the battery string, thus achieving non-destructive detection. It can be applied to the appearance inspection of battery strings of various sizes. While collecting data on the battery string, the appearance of the battery string can be inspected, which effectively shortens the detection time and improves the detection efficiency and real-time performance.

[0041] The technical solution of the embodiment of the present application is described below in conjunction with specific application scenarios.

[0042] In an embodiment of the present application, a battery string detection method is proposed, in which an image of the battery string is captured by a camera, and whether the battery string has defects is determined based on the image. Figure 2 The figure shows a schematic diagram of the system structure of battery string detection. Multiple battery cells are placed on the string welding device, that is, multiple battery cells are distributed to the string welding template of the string welding device, so that the string welding device can weld each battery cell in sequence from the right side to the left side of the string welding template.

[0043] Camera 1 is deployed on the upper side of the string welding device, and camera 1 is used to collect images of the upper surface of the battery cell. Camera 2 is deployed on the lower side of the string welding device, and camera 2 is used to collect images of the lower surface of the battery cell. Alternatively, camera 1 may be deployed only on the upper side of the string welding device, or camera 2 may be deployed only on the lower side of the string welding device.

[0044] Take a battery string including 15 cells, and a total of M cells (such as 1500 cells, etc.) as an example for explanation, then the string welding equipment performs welding operations on cell 1, cell 2, ..., cell M in order from the right to the left. In addition, cutting is performed between the 15th cell and the 16th cell, and the 1st to 15th cells form a battery string. Cutting is performed between the 30th cell and the 31st cell, and the 16th to 30th cells form a battery string, and so on.

[0045] Based on this, each battery string passes through the field of view of the camera (camera 1 and / or camera 2) in turn. When battery string a1 (such as the 1st to 15th battery cells) passes through the field of view of the camera, the camera can capture an image of battery string a1, and determine whether battery string a1 has defects based on the image of battery string a1. When battery string a2 (such as the 16th to 30th battery cells) passes through the field of view of the camera, the camera can capture an image of battery string a2, and determine whether battery string a2 has defects based on the image of battery string a2, and so on.

[0046] Considering the limited field of view of the camera, a complete image of the battery string is not obtained through a single image acquisition. Instead, multiple images are acquired and stitched together to obtain a complete image of the battery string.

[0047] Through image acquisition and image stitching, the appearance image of an extra-long battery string or the appearance image of a regular-length battery string can be obtained. An extra-long battery string refers to a battery string whose length is greater than a threshold, while a regular-length battery string refers to a battery string whose length is not greater than a threshold. Figure 3A FIG. 1 is a schematic diagram of an appearance image of an ultra-long battery string. Based on the appearance image of the ultra-long battery string, fast and accurate defect detection of the ultra-long battery string can be achieved. Figure 3B As shown, it is a schematic diagram of the appearance image of a battery string of a conventional length. Based on the appearance image of the battery string of a conventional length, rapid and accurate defect detection of the battery string of a conventional length can be achieved.

[0048] For example, see Figure 4 The figure shows a schematic diagram of the battery string detection system structure. The battery string detection system may include but is not limited to: image acquisition equipment, battery string detection equipment, control equipment, MES (Manufacturing Execution System, production execution system) equipment and centralized control equipment. In addition, the battery string detection system is used to detect defects in the battery string, and the image acquisition equipment may be a camera.

[0049] For battery strings, the battery string sequence to be detected (such as 1500 battery cells) may include multiple battery strings (such as 100 battery strings), and each battery string may include multiple battery cells (such as 15 battery cells). All battery cells are located in a transmission device (the transmission device is a string welding device, or other equipment with a transmission function, and the present embodiment takes the string welding device as an example), and the string welding device transports each battery cell in the battery string sequence from right to left (or from left to right), so that each battery cell passes through the camera's field of view in turn, that is, each battery string in the battery string sequence passes through the camera's field of view in turn. The battery string can be a photovoltaic battery string or other types of battery strings, and there is no restriction on the type of battery string.

[0050] For the camera, the camera can be a high-resolution camera or a low-resolution camera. Take a high-resolution camera as an example. The camera is located in a fixed position and takes pictures of the battery string that moves to the field of view of the camera to obtain an image of the battery string (one battery string will correspond to multiple images, that is, multiple images are obtained after multiple shots). Since the camera is a high-resolution camera, a high-definition image of the battery string can be obtained. The high-definition image ensures that the fine details on the surface of the battery string can be clearly displayed, which is convenient for defect detection.

[0051] A camera 1 may be deployed on the upper side of the string welding device, and the camera 1 may be referred to as a front camera, and the front camera is used to capture images of the upper surface of the battery cell. And / or, a camera 2 may be deployed on the lower side of the string welding device, and the camera 2 may be referred to as a back camera, and the back camera is used to capture images of the lower surface of the battery cell.

[0052] For the battery string detection equipment, the camera sends the collected image to the battery string detection equipment, and the battery string detection equipment performs defect detection based on the image, that is, executes the battery string detection method to complete the battery string defect detection. The battery string detection equipment combines image processing technology and artificial intelligence algorithms to realize automatic recognition and classification of battery string defects, thereby improving detection efficiency and accuracy. The battery string detection equipment can also perform statistics and analysis on the detected defect data to help production management personnel promptly discover the source and cause of defective processes, optimize production processes, and reduce production costs. In the subsequent process, the battery string detection method is introduced in detail, and the statistics and analysis process of defect data will not be repeated in this embodiment.

[0053] For the control device, the control device can be a PLC (Programmable Logic Controller), an FPGA (Field Programmable Gate Array), a CPLD (Complex Programmable Logic Device), etc., and there is no restriction on the type of the control device. The control device is used to control the string welding device and the camera, such as linking the string welding device and the camera. For example, the control device can control the moving speed of the string welding device and control the interval between two adjacent camera acquisitions. For example, assuming that the field of view of the camera is 10 cm, and the interval between two adjacent camera acquisitions is 1 second, then the moving speed of the string welding device needs to be controlled to be 10 cm / 1 second, so that the camera can continuously capture images and there is no overlapping area between adjacent images.

[0054] Of course, the above is only an example of controlling the string welding equipment and the camera, and there is no limitation on this control method, as long as the camera continuously captures images and there is no overlapping area between adjacent images.

[0055] For MES equipment, the battery string inspection equipment can send the defect detection results of the battery string to the MES equipment, and the MES equipment obtains the defect detection results corresponding to the battery string and processes the battery string based on the defect detection results. For example, the MES equipment can receive visual inspection results (such as defect detection results) and collect production data to provide strong support for subsequent quality analysis, defect tracing, etc.

[0056] For example, MES equipment can dynamically monitor defects in battery strings, detect problems in a timely manner and take measures, providing strong support for subsequent quality analysis, defect tracing, etc.

[0057] For the centralized control device, the centralized control device can obtain the defect detection results corresponding to the battery string. For example, for a battery string with defects, the battery string detection device can send the defect detection results of the battery string to the centralized control device. The centralized control device interacts with the user, and the user decides whether the battery string has defects (that is, the user manually detects whether the battery string has defects based on the defect detection results and obtains the manual detection results of the battery string). The centralized control device can obtain the manual detection results of the battery string and return the manual detection results (such as the battery string has defects, or the battery string does not have defects) to the battery string detection device. The battery string detection device sends the corresponding results to the string welding control device (PLC) according to the re-judgment results sent by the centralized control device, thereby determining the flow direction of the battery string. For example, the battery string detection device can obtain the defect detection results of the battery string (such as manual detection results), generate operation control instructions according to the defect detection results, and send the operation control instructions to the string welding control device (PLC). The string welding control device (PLC) controls the flow direction of the battery string in the direction corresponding to the defect detection results according to the operation control instructions. For example, if the defect detection result shows that there is no defect, the target battery string is controlled to flow into the market; if the defect detection result shows that there is a defect, the target battery string is controlled not to flow into the market.

[0058] For example, if a centralized control device is deployed in the system, the battery string detection device sends the manual detection results of the battery string to the MES device. If no centralized control device is deployed in the system, the battery string detection device sends the defect detection results analyzed by the battery string detection device itself to the MES device.

[0059] In the above application scenario, a battery string detection method is proposed in the embodiment of the present application, see Figure 5 FIG. 4 is a flow chart of the method, which may include the following steps:

[0060] Step 501: Capture an image through a front camera. The image captured by the front camera is an image of the upper surface of the battery string. The image captured by the front camera is called a front image (i.e., the image to be detected of the upper surface of the battery string acquired by the front camera). Capture an image through a back camera. The image captured by the back camera is an image of the lower surface of the battery string. The image captured by the back camera is called a back image (i.e., the image to be detected of the lower surface of the battery string acquired by the back camera).

[0061] Step 502: stitch the front image and stitch the back image. For example, the battery string detection device can obtain the front image from the front camera and stitch the front image. The battery string detection device can obtain the back image from the back camera and stitch the back image.

[0062] Step 503: Perform battery string inspection based on the front stitching image, and perform battery string inspection based on the back stitching image. For example, after the battery string inspection device stitches the front images to obtain the front stitching image, it performs battery string inspection on the upper surface of the battery string based on the front stitching image to detect whether there are defects on the upper surface of the battery string, that is, the image to be inspected on the upper surface is used to perform defect inspection on the upper surface of the battery string. After the battery string inspection device stitches the back images to obtain the back stitching image, it performs battery string inspection on the lower surface of the battery string based on the back stitching image to detect whether there are defects on the lower surface of the battery string, that is, the image to be inspected on the lower surface is used to perform defect inspection on the lower surface of the battery string.

[0063] Step 504: After obtaining the front complete battery string image of the battery string, cache the front complete battery string image. After obtaining the back complete battery string image of the battery string, cache the back complete battery string image.

[0064] Exemplarily, steps 501 to 504 are implemented by a battery string detection device, that is, the battery string detection device implements the detection of the battery string based on steps 501 to 504. The detection process of the battery string is described later.

[0065] Step 505 : Count the front detection data of the battery string, and count the back detection data of the battery string.

[0066] For example, after performing a battery string inspection on the upper surface of the battery string based on the front stitching image, the battery string inspection device counts the front inspection data of the battery string based on the battery string inspection result. The front inspection data may include whether there are defects on the upper surface of the battery string. When there are defects on the upper surface, it may include the defect type, defect area, etc., that is, the front inspection data includes statistical data and analysis data of the upper surface defects.

[0067] After performing battery string inspection on the lower surface of the battery string based on the back stitching image, the battery string inspection device counts the back surface inspection data of the battery string based on the battery string inspection result. The back surface inspection data may include whether there are defects on the lower surface of the battery string. When there are defects on the lower surface, the back surface inspection data may include the defect type, defect area, etc., that is, the back surface inspection data includes statistical data and analysis data of the defects on the lower surface.

[0068] Step 506: The battery string detection device aggregates the front detection data of the battery string and the back detection data of the battery string to obtain appearance detection data of the battery string, and caches the appearance detection data.

[0069] Step 507: The battery string detection device outputs appearance detection data.

[0070] For example, when completing the inspection of a battery string, the battery string inspection device may cache the image data and appearance inspection data of the battery string, and output the appearance inspection data for further analysis and processing.

[0071] Step 508: The centralized control device obtains the image data and appearance inspection data of the battery string. The centralized control device can display the image data and appearance inspection data to the user, and the user decides whether the battery string has defects, that is, the centralized control device can obtain the manual inspection results of the battery string (such as whether the battery string has defects or not), that is, the user manually inspects whether the battery string has defects based on the appearance inspection data.

[0072] Step 509: The battery string detection device obtains the manual detection result of the battery string, and outputs the manual detection result of the battery string to the MES device, and the MES device processes based on the defect detection result.

[0073] In the embodiment of the present application, a battery string detection method is proposed, and the processing process of step 501 to step 504 is described. In this embodiment, taking one camera (the camera may be a back camera or a front camera) to collect images as an example, the processing flow when two cameras collect images at the same time is similar, but it is based on image processing in two directions. Figure 6 FIG. 4 is a flow chart of the method, which may include:

[0074] Step 601: When a target battery string passes through the field of view of a camera, an image of the target battery string to be inspected is collected by the camera. The target battery string may be any battery string in a battery string sequence.

[0075] Exemplarily, all battery strings to be detected (i.e., all battery strings placed on the string welding device) can be referred to as a battery string sequence, i.e., a battery string sequence includes multiple battery strings. For each battery string, the battery string includes multiple battery cells (e.g., 15 battery cells). The string welding device can transport each battery string in the battery string sequence, i.e., each battery string in the battery string sequence passes through the field of view of the camera in turn, and the battery string currently passing through the field of view of the camera is recorded as a target battery string, i.e., the target battery string passes through the field of view of the camera.

[0076] For example, the string welding device transports the battery string according to a fixed rule (such as a fixed moving speed), and each time the string welding device moves a certain distance, the camera is triggered to collect the image to be detected of the corresponding moving distance. During the image acquisition process, the camera continuously collects the images to be detected, and there is no overlapping area between two adjacent images to be detected.

[0077] For example, the camera captures image 1 to be detected for the first time, and image 1 to be detected covers the 0th to 10th centimeters of the battery string. When the string welding equipment transports the 10th to 20th centimeters of the battery string to the camera's field of view, the camera is triggered to capture image 2 to be detected, and image 2 to be detected covers the 10th to 20th centimeters of the battery string. When the string welding equipment transports the 20th to 30th centimeters of the battery string to the camera's field of view, the camera is triggered to capture image 3 to be detected, and image 3 to be detected covers the 20th to 30th centimeters of the battery string. And so on, the camera continuously captures multiple images to be detected.

[0078] In order to achieve the above-mentioned synchronous control, the moving speed can be directly configured in the string welding device, and the interval length between two adjacent acquisitions can be configured in the camera to achieve the above-mentioned synchronous control. Alternatively, the string welding device and the camera can be controlled by the control device, such as controlling the moving speed of the string welding device and controlling the interval length between two adjacent acquisitions of the camera. For example, assuming that the field of view of the camera is 10 cm, and the interval length between two adjacent acquisitions of the camera is 1 second, then the moving speed of the string welding device needs to be controlled to be 10 cm / 1 second.

[0079] Exemplarily, the field of view of the camera can also be controlled so that the field of view of the camera is larger than the width of the battery cell. For example, when the width of the battery cell is M, the field of view of the camera needs to be larger than M, so that the image to be detected collected by the camera can include at least one battery cell. For example, if the width of the battery cell is 8 cm, the field of view of the camera is 10 cm. If the width of the battery cell is 10 cm, the field of view of the camera is 12 cm. By analogy, the field of view of the camera can be larger than M.

[0080] If the moving direction of the string welding equipment corresponds to the horizontal direction, the field of view of the camera is the field of view in the horizontal direction. If the moving direction of the string welding equipment corresponds to the vertical direction (height direction), the field of view of the camera is the field of view in the vertical direction, that is, the field of view of the camera is the field of view in the height direction.

[0081] For example, the interval between two adjacent battery cells in the same battery string is M1 (i.e., the distance between the centers of two adjacent battery cells, which is taken as the width of the battery cell), and the interval between two adjacent battery cells in different battery strings is M2, and M1 is smaller than M2, that is, the widths of different battery cells may be different. Therefore, the field of view of the camera needs to be larger than the maximum width of the battery cell (such as M2).

[0082] Step 602: When the image to be detected is obtained, determine whether the image to be detected is a valid image.

[0083] If yes, then execute step 603. For example, if the image to be detected is a valid image, it means that there are battery cells in the image to be detected, and operations such as splicing and detection can be performed based on the image to be detected. If not, then discard the image to be detected. For example, if the image to be detected is an invalid image, it means that there are no battery cells in the image to be detected, and operations such as splicing and detection cannot be performed based on the image to be detected, and the image to be detected is discarded.

[0084] Exemplarily, if the average brightness of the battery cell area in the image to be detected is less than the brightness threshold, and at least one of the string head feature and the string tail feature exists in the image to be detected, then the image to be detected is a valid image; otherwise, if the average brightness of the battery cell area in the image to be detected is not less than the brightness threshold, the image to be detected does not have a string head feature, and the image to be detected does not have a string tail feature, then the image to be detected is an invalid image.

[0085] Regarding how to determine the battery cell area in the image to be detected, please refer to the subsequent steps. After obtaining the battery cell area in the image to be detected, the brightness values ​​of all pixels in the battery cell area are counted, and the brightness mean is calculated based on the brightness values ​​of all pixels, and then the brightness mean is compared to see if it is less than the brightness threshold. The brightness threshold can be configured based on experience, and there is no restriction on this brightness threshold. Since the brightness of the battery cell area is relatively small, if the brightness mean is less than the brightness threshold, it means that the image quality of the image to be detected is good. Conversely, if the brightness mean is not less than the brightness threshold, it means that the image quality of the image to be detected is poor.

[0086] Regarding how to determine whether the image to be detected has a string head feature, and how to determine whether the image to be detected has a string tail feature, please refer to the subsequent steps, and will not be repeated here.

[0087] Exemplarily, after obtaining the image to be detected, before determining whether the image to be detected is a valid image, the image to be detected may be preprocessed to determine whether the preprocessed image is a valid image. For example, the image to be detected may be subjected to noise removal, brightness enhancement, white balance removal, etc., without limitation.

[0088] Step 603: when the image to be detected is obtained, determine whether the number of spliced ​​images of the battery string image is 0.

[0089] Exemplarily, in the initial state, the number of stitched images of the battery string image is a preset value (such as 0, and 0 is taken as an example below), indicating that the processing of the previous battery string has been completed, and the current battery string (i.e., the target battery string) has not been processed yet, and the image to be detected is the first image of the target battery string.

[0090] Each time an image to be detected is obtained, the number of stitched images is adjusted, that is, a certain value (such as 1, 1 is used as an example in the following) is added to the number of stitched images, that is, 1 is added to the number of stitched images.

[0091] If the number of stitched images of the battery string image is not 0, it means that when the image to be detected is obtained, the target battery string has been processed, and the image to be detected is not the first image for the target battery string.

[0092] If yes, that is, the number of stitched images is 0, step 604 may be executed.

[0093] If not, that is, the number of stitched images is not 0, step 606 may be executed.

[0094] Step 604: determine whether the image to be detected has a string head feature of the target battery string.

[0095] Exemplarily, if the number of stitched images is 0, it means that the image to be detected is the first image for the target battery string, and it is necessary to determine whether the image to be detected has the string head feature of the target battery string. If the image to be detected has the string head feature of the target battery string, it is determined that the image to be detected is the first image for the target battery string, and step 605 is executed. If the image to be detected does not have the string head feature of the target battery string, it is determined that the image to be detected is not the first image for the target battery string, and the number of stitched images can be kept at 0, waiting to obtain the next image to be detected, and when the next image to be detected is obtained, steps 602 and 603 are repeated.

[0096] Step 605: update the image to be detected to the battery string image of the target battery string, and update the number of stitched images to 1. The number of stitched images being 1 indicates that one image to be detected of the target battery string has been obtained.

[0097] For example, when the image to be detected is the first image of the target battery string, since the battery string image of the target battery string has not been cached, image stitching cannot be performed. Therefore, the image to be detected is directly updated to the battery string image of the target battery string. In this case, considering that the image to be detected may not have a complete battery cell (there are some battery cells of the previous battery string and some battery cells of the target battery string), detection is not performed based on the image to be detected, that is, whether the target battery string has defects is not detected yet.

[0098] In addition, when the first image of the target battery string is obtained, the number of stitched images can be updated to 1, and the next image to be detected is waited to be obtained, and the above steps are repeated based on the next image to be detected.

[0099] Step 606: Determine whether the number of stitched images of the battery string image reaches a number threshold.

[0100] Exemplarily, a quantity threshold may be preconfigured, and the quantity threshold may be determined based on a configured maximum splicing quantity, such as the quantity threshold may be less than the maximum splicing quantity, such as the quantity threshold may be the difference between the maximum splicing quantity and 1, or the quantity threshold may be the difference between the maximum splicing quantity and 2.

[0101] The maximum splicing quantity indicates the total number of battery cells in the target battery string. For example, if the target battery string includes 15 battery cells, the maximum splicing quantity may be 15. Therefore, the quantity threshold may be 13 or 14.

[0102] When obtaining the image to be detected, if the number of stitched images does not reach the number threshold, it means that the image to be detected will not be the last image of the target battery string, and the image to be detected of the target battery string can continue to be obtained, that is, wait for the next image to be detected to be obtained, and repeat the above steps. Based on this, step 607 can be executed.

[0103] When the image to be detected is obtained, if the number of stitched images reaches the number threshold, it means that the image to be detected may be the last image of the target battery string. Based on this, step 610 can be executed.

[0104] Step 607: splice the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image, update the spliced ​​image to the battery string image of the target battery string, and increase the number of spliced ​​images by 1.

[0105] For example, when the first image to be detected of the target battery string is obtained, the image to be detected is updated to the battery string image a1 of the target battery string, and the number of stitched images is updated to 1. When the second image to be detected of the target battery string is obtained, the image to be detected is stitched with the battery string image a1 to obtain a stitched image, and the stitched image is updated to the battery string image a2 of the target battery string, and the number of stitched images is increased by 1, that is, the number of stitched images is 2, indicating that two images to be detected of the target battery string have been obtained.

[0106] When the image to be detected is spliced ​​with the battery string image a1, the image to be detected may be spliced ​​behind the battery string image a1, that is, serial splicing, which may also be called hard splicing.

[0107] When the third image to be detected of the target battery string is obtained, the image to be detected is spliced ​​with the battery string image a2 to obtain a spliced ​​image, the spliced ​​image is updated to the battery string image a3 of the target battery string, and the number of spliced ​​images is updated to 3. Similarly, assuming that the number threshold is 13, when the 13th image to be detected of the target battery string is obtained, the image to be detected is spliced ​​with the battery string image a12 to obtain a spliced ​​image, the spliced ​​image is updated to the battery string image a13 of the target battery string, and the number of spliced ​​images is updated to 13.

[0108] Step 608: when the spliced ​​image is obtained, a cell sub-image is acquired from the spliced ​​image. The cell sub-image may include a target cell, and the target cell is a complete cell that has not been inspected.

[0109] For example, when the image to be detected is spliced ​​with the battery string image a1 to obtain a spliced ​​image (battery string image a2), if the spliced ​​image includes only one battery cell b1, then the battery cell sub-image c1 of the battery cell b1 is obtained from the spliced ​​image, that is, the target battery cell is the battery cell b1; if the spliced ​​image includes battery cell b1 and battery cell b2, then the battery cell sub-image c1 of the battery cell b1 and the battery cell sub-image c2 of the battery cell b2 are obtained from the spliced ​​image, and the following takes the case where only one battery cell b1 is included as an example.

[0110] When the image to be inspected is spliced ​​with the battery string image a2 to obtain a spliced ​​image (battery string image a3), if the spliced ​​image includes battery cell b1 and battery cell b2, since battery cell b1 is a battery cell that has been inspected (that is, the previous inspection process has completed the inspection of battery cell b1), the target battery cell is battery cell b2. In this way, a battery cell sub-image c2 of battery cell b2 can be obtained from the spliced ​​image.

[0111] By analogy, when the image to be detected is spliced ​​with the battery string image a12 to obtain a spliced ​​image (battery string image a13), the battery cell sub-image c12 of the battery cell b12 can be obtained from the spliced ​​image.

[0112] Exemplarily, when obtaining a cell sub-image from a stitched image, the stitched image can be input into a first target detection model, which outputs a first predicted position of a cell main grid area, and the area corresponding to the first predicted position can be determined as the cell sub-image.

[0113] For example, a first initial detection model may be obtained, and the first initial detection model may be a deep learning model or another type of model. The first initial detection model may be trained using training data to obtain a trained first target detection model, and there is no restriction on the training process.

[0114] Based on the trained first target detection model, when the spliced ​​image is obtained, the spliced ​​image can be input to the first target detection model, and the first target detection model processes the spliced ​​image to obtain the first predicted position of the battery cell main grid area (i.e., the harpoon area), and the first target detection model outputs the first predicted position of the battery cell main grid area. Assuming that the battery cell main grid area is a rectangular area, the first predicted position can be the coordinates of the four corner points of the battery cell main grid area, or the coordinates of a corner point of the battery cell main grid area + length + width, or the coordinates of the center point of the battery cell main grid area + length + width. There is no restriction on this first predicted position, as long as it can represent the battery cell main grid area.

[0115] After the first predicted position is obtained, the area corresponding to the first predicted position (such as the main grid line area of ​​the cell corresponding to the coordinates of the four corner points) can be determined as the cell sub-image.

[0116] When the stitched image includes multiple battery cells, the first target detection model can output multiple first prediction positions, and the last first prediction position can be selected from the multiple first prediction positions. The last first prediction position corresponds to the battery cell image of the target battery cell, which is a battery cell image that has not yet been detected, and the previous first prediction positions correspond to the battery cell image of the battery cell that has completed detection.

[0117] In the above embodiment, it involves determining the battery cell area within the image to be detected. Based on this, the image to be detected can be input into the first target detection model, and the first target detection model outputs the first predicted position of the battery cell main grid line area, and the area corresponding to the first predicted position can be determined as the battery cell area.

[0118] Step 609: perform defect detection on the target cell based on the cell sub-image.

[0119] For example, after obtaining the cell sub-image c1, the target cell b1 can be inspected for defects based on the cell sub-image c1, such as whether the target cell b1 has defects such as appearance damage, appearance dirtiness, poor welding quality, abnormal shape (such as bending), abnormal cell spacing, etc. After obtaining the cell sub-image c2, the target cell b2 can be inspected for defects based on the cell sub-image c2, and so on.

[0120] Exemplarily, when performing defect detection on a target cell, the cell sub-image can be input to the second target detection model, and the second target detection model outputs a second predicted position and a predicted category. The area corresponding to the second predicted position can be determined as the defect area of ​​the cell sub-image (i.e., the defect area has a defect), and the defect type of the defect area is determined based on the predicted category. In addition to the second predicted position and the predicted category, the second target detection model can also output the size of the defect area.

[0121] For example, a second initial detection model may be obtained, and the second initial detection model may be a deep learning model or a neural network, or other types of models. The second initial detection model may be trained using training data to obtain a trained second target detection model, and there is no restriction on the training process.

[0122] Based on the trained second target detection model, when a cell wafer image is obtained, the cell wafer image can be input to the second target detection model, and the second target detection model can be processed based on the cell wafer image. If there is a defect in the cell wafer image, the second target detection model can obtain a second predicted position and a predicted category, and output the second predicted position and the predicted category. If there is no defect in the cell wafer image, the second target detection model can output a detection result that there is no defect.

[0123] Assuming that the defective area in the cell wafer image is a rectangular area, the second predicted position can be the coordinates of the four corner points of the defective area, or the coordinates of one corner point of the defective area + length + width, or the coordinates of the center point of the defective area + length + width. There is no restriction on this, as long as it can represent the defective area. The prediction category can represent the defect type of the defective area. For example, when the prediction category is 1, it means that the defective area has an appearance damage defect. When the prediction category is 2, it means that the defective area has an appearance dirt defect. When the prediction category is 3, it means that the defective area has a poor welding quality defect, and so on.

[0124] After obtaining the second predicted position and predicted category, the area corresponding to the second predicted position (such as the main grid line area of ​​the cell corresponding to the four corner point coordinates) can be determined as the defect area of ​​the cell sub-image, and then the defect type of the defect area can be determined based on the predicted category.

[0125] Exemplarily, after obtaining the cell image, after inputting the cell image to the second target detection model, the cell image can also be preprocessed, and the preprocessed cell image can be input to the second target detection model. By preprocessing the cell image, the image features of the battery string can be enhanced and the interference of other pollution on the battery string can be reduced. For example, the cell image can be de-noised, the brightness can be enhanced, the white balance can be removed, etc., and there is no restriction on this.

[0126] Exemplarily, in addition to the detection method of inputting the cell sub-image into the second target detection model, the cell sub-image can also be directly analyzed to determine whether the target cell has defects. For example, the cell spacing between the target cell and the adjacent cell can be counted based on the cell sub-image. If the cell spacing is greater than a threshold, it means that the target cell has defects. Of course, this is just an example. In summary, defect detection can be performed by combining models and non-models.

[0127] Exemplarily, in the above method, a sliding window detection method of stitching and detecting is adopted (a dynamic window of unfixed length is used to perform a complete battery cell appearance inspection on the battery string), that is, when a stitched image of part of the images to be inspected is obtained, the battery cell sub-image is directly obtained from the stitched image, and the inspection is performed based on the battery cell sub-image, rather than obtaining the battery cell image from the stitched image when the stitched image of all the images to be inspected is obtained. For example, when the image to be inspected is spliced ​​with the battery string image a1 to obtain the spliced ​​image, the battery cell sub-image c1 is directly obtained for inspection. When the image to be inspected is spliced ​​with the battery string image a2 to obtain the spliced ​​image, the battery cell sub-image c2 is directly obtained for inspection.

[0128] By adopting a sliding window detection method of splicing and detecting at the same time, it can be applicable to the appearance detection of battery strings of various sizes (that is, the sizes of different battery strings can be different). While collecting data on the battery strings, the appearance of the battery strings can be inspected, which effectively shortens the detection time and improves the detection efficiency and real-time performance.

[0129] Step 610: Determine whether the image to be inspected has a string end feature of the target battery string.

[0130] Exemplarily, if the number of stitched images reaches the number threshold, it means that the image to be detected may be the last image for the target battery string, and it is necessary to determine whether the image to be detected has the end-of-string feature of the target battery string. If the image to be detected has the end-of-string feature of the target battery string, it is determined that the image to be detected is the last image for the target battery string, and step 611 is executed. If the image to be detected does not have the end-of-string feature of the target battery string, it is determined that the image to be detected is not the last image for the target battery string, and step 607 is executed, that is, the image to be detected is stitched with the battery string image of the target battery string to obtain a stitched image, the stitched image is updated to the battery string image of the target battery string, and the number of stitched images is increased by 1, and a cell sub-image is obtained from the stitched image, and defect detection is performed on the target cell based on the cell sub-image.

[0131] Step 611: In the case where the end-of-string feature exists, determine whether an end-of-string signal is received.

[0132] For example, a control device (such as a PLC) is used to control the string welding device, and can know which battery cell the string welding device transports to the camera's field of view. If the control device knows that the string welding device transports the last battery cell of the target battery string to the camera's field of view, it sends a string end signal to the battery string detection device, and the string end signal indicates that the last battery cell of the target battery string is transported to the camera's field of view.

[0133] In the case of the presence of the string end feature, the battery string detection device determines whether the string end signal is received. If the string end signal is received, it is confirmed that the image to be detected is the last image for the target battery string, that is, through double confirmation (string end feature + string end signal), it is determined that the image to be detected is the last image for the target battery string, and step 612 is executed. If the string end signal is not received, it is confirmed that the image to be detected is not the last image for the target battery string, that is, the results of the string end feature and the string end signal are inconsistent, and therefore, abnormal information can be output, which indicates that the results of the string end feature and the string end signal are inconsistent, and the user needs to repair the string end logic or the string welding control device. This embodiment does not limit this repair process.

[0134] Step 612: splice the image to be inspected with the battery string image of the target battery string to obtain a spliced ​​image, update the spliced ​​image to a complete battery string image of the target battery string, and output the complete battery string image. Also, obtain a cell sub-image from the spliced ​​image, the cell sub-image may include the target cell, and perform defect detection on the target cell based on the cell sub-image.

[0135] For example, if the image to be detected is the last image of the target battery string, after the spliced ​​image is obtained, the spliced ​​image is the complete battery string image of the target battery string, and the complete battery string image can be cached. Figure 3A or Figure 3B Shown is a schematic diagram of a complete battery string image.

[0136] After the spliced ​​image is obtained, a cell sub-image may be obtained from the spliced ​​image, and defect detection may be performed on the target cell based on the cell sub-image, see steps 308 and 309 .

[0137] Exemplarily, after obtaining the complete battery string image and the defect detection result of each battery cell, the sub-image of each battery cell can be determined from the complete battery string image (during the defect detection process, the sub-image of each battery cell, i.e., the above-mentioned battery cell sub-image, has been obtained). In this way, for each battery cell, the defect detection result of the battery cell is added to the sub-image of the battery cell. The defect detection result can indicate whether the battery cell has defects. If defects exist, it can also indicate the defect area and defect type.

[0138] Exemplarily, if the image to be detected is the last image of the target battery string, the number of stitched images may be updated to 0, and then the process returns to step 602, that is, the above steps are repeated based on the image to be detected.

[0139] Obviously, the image to be detected is used as the last image of the target battery string to complete the image stitching of the target battery string. In addition, the image to be detected can be used as the first image of the next battery string. In the stitching process of the next battery string, the image to be detected needs to be used as the first image for stitching.

[0140] In a possible implementation, after obtaining the image to be detected, it can be determined whether the image to be detected has the string head feature of the target battery string, or it can be determined whether the image to be detected has the string tail feature of the target battery string. In order to determine whether the image to be detected has the string head feature and string tail feature of the target battery string, the following method can be used: determine the edge pixel points of the target battery string from the image to be detected. For example, the features of the edge pixel points (such as brightness features, texture features, gradient features, etc., and there is no restriction on the type of features) are different from the features of the non-edge pixel points. Therefore, for each pixel point in the image to be detected, if the feature of the pixel point is similar to the feature of the edge pixel point, the pixel point is determined as an edge pixel point. If the feature of the pixel point is similar to the feature of the non-edge pixel point, the pixel point can be determined as a non-edge pixel point. In this way, the edge pixel points of the target battery string are determined from the image to be detected.

[0141] If a first edge straight line and a second edge straight line are fitted based on these edge pixel points, and the distance between the first edge straight line and the second edge straight line is greater than a distance threshold (which can be configured based on experience), and the average brightness of the transition area between the first edge straight line and the second edge straight line is greater than the first brightness threshold, and the average brightness of the battery cell area in the image to be detected is less than the second brightness threshold (the second brightness threshold can be less than the first brightness threshold), then it is determined that the image to be detected has a string head feature and a string tail feature, and the first edge straight line (the edge straight line at the front side of the image) is the string tail straight line of the previous battery string, and the second edge straight line (the edge straight line at the back side of the image) is the string head straight line of the next battery string. For example, if the image to be detected has a string head feature and a string tail feature, then the image to be detected has a string tail straight line of the previous battery string and a string head straight line of the next battery string. Based on this, if two edge straight lines cannot be fitted based on these edge pixel points, it is determined that the image to be detected does not have a string head feature and a string tail feature.

[0142] If two edge straight lines, i.e., a first edge straight line and a second edge straight line, are fitted based on these edge pixel points, considering that the distance between the end straight line of the previous battery string and the beginning straight line of the next battery string is relatively large, it is necessary to determine whether the distance between the first edge straight line and the second edge straight line is greater than a distance threshold. If not, it is determined that the image to be detected does not have a string beginning feature and a string end feature.

[0143] If the distance between the first edge line and the second edge line is greater than the distance threshold, considering that the average brightness of the transition area between the first edge line and the second edge line (the middle area between the two battery strings) is relatively large, the brightness value of each pixel point in the transition area between the first edge line and the second edge line can also be counted, and the brightness average of the brightness values ​​of these pixel points is calculated. If the brightness average is not greater than the first brightness threshold, it is determined that the image to be detected does not have a string head feature and a string tail feature.

[0144] If the brightness mean is greater than the first brightness threshold, considering that the brightness mean of the cell region (i.e., the region where the cell is located) is relatively small, the brightness value of each pixel in the cell region in the image to be detected can also be counted, and the brightness mean of the brightness values ​​of these pixels can be calculated. If the brightness mean of the cell region is not less than the second brightness threshold, it is determined that the image to be detected does not have a string head feature or a string tail feature.

[0145] If the average brightness of the battery cell area is less than the second brightness threshold, it is finally determined that the image to be detected has a string head feature, and it is determined that the image to be detected has a string tail feature, that is, the first edge straight line is the string tail straight line of the previous battery string, and the second edge straight line is the string head straight line of the next battery string.

[0146] In a possible implementation, the default value of the string tail signal is 0, indicating that the string tail signal is not received. In the case of a string tail feature, the battery string detection device determines whether the string tail signal is received, and if the string tail signal is received, the value of the string tail signal is modified to 1, indicating that the string tail signal is received. After the complete battery string image is output, the value of the string tail signal is modified to 0, indicating that the string tail signal is not received.

[0147] In a possible implementation, the battery string detection device can maintain a small picture detection image queue and a whole string image queue. When the image to be detected is obtained, if the number of spliced ​​images in the small picture detection image queue is 0, the image to be detected is stored in the small picture detection image queue. When the image to be detected is obtained, if the number of spliced ​​images in the small picture detection image queue is 1, the image to be detected is stored in the small picture detection image queue. When the image to be detected is obtained, if the number of spliced ​​images in the small picture detection image queue is 2, it is determined whether the current image in the small picture detection image queue is a valid image. If so, the second image to be detected in the small picture detection image queue is stored in the whole string image queue, and the small picture detection image queue is cleared at the same time. If not, the first image to be detected in the small picture detection image queue is removed. When the image to be detected is obtained, if the number of spliced ​​images is greater than or equal to 3, the image to be detected is stored in the whole string image queue, and so on, until the number of spliced ​​images is 0.

[0148] By maintaining the small image detection image queue and the whole string image queue, image stitching and defect detection are not performed for the images to be detected in the small image detection image queue. Image stitching and defect detection are performed for the images to be detected in the whole string image queue. The image stitching and defect detection process can refer to the above embodiment.

[0149] In one possible implementation, see Figure 7 As shown, it is a schematic diagram of the process of battery string splicing. In the battery string detection, the battery string splicing method plays a key role in ensuring the integrity of the battery string.

[0150] Step 701, determine whether the string head flag signal is true (True). If not, that is, the string head flag signal is false (False), then step 702 can be executed; if so, then step 705 can be executed.

[0151] Step 702: Determine whether the number of images in the thumbnail detection image queue is equal to 2.

[0152] If yes, then step 703 may be executed; if no, then step 704 may be executed.

[0153] Step 703: Enter the thumbnail detection process based on the thumbnail detection image queue.

[0154] In the thumbnail detection process, determine whether the image to be detected in the thumbnail detection image queue is a valid image. If so, add the currently acquired image to be detected to the entire string of image queues. When there is only a string head feature in the thumbnail splicing image, clear the entire string of image queues, add two images in the thumbnail detection image queue, clear the thumbnail detection image queue, set the string head flag signal to true, and end the current thumbnail splicing process. Or, when there is no string head and string tail feature in the thumbnail splicing image, remove the first image in the thumbnail detection image queue and end the current thumbnail splicing process. Or, when there is a string tail feature in the thumbnail splicing image, clear the thumbnail detection image queue, retain the last image in the entire string of image queues, and end the current thumbnail splicing process.

[0155] If not, the first image is removed from the small image detection image queue, and the current small image stitching process ends.

[0156] Step 704: Add the currently acquired image to be detected to the small image detection image queue.

[0157] Step 705: Add the currently acquired image to be detected to the entire image queue.

[0158] Step 706: Determine whether the number of images in the entire image queue is less than a number threshold. For example, the number threshold may be the difference between the maximum splicing number and 2, where the maximum splicing number represents the total number of cells in the target battery string. If yes, execute step 707; if no, terminate the process.

[0159] Step 707, set the string head flag signal to false (False), that is, the string head flag signal is not true, add the currently collected image to be detected to the small image detection image queue, and end the processing flow.

[0160] It can be seen from the above technical scheme that in the embodiment of the present application, the defects of the battery string are detected by image, and the fast and accurate defect detection of the super-long battery string and the conventional length battery string is realized, the subjectivity and error of manual visual inspection are avoided, the detection quality and efficiency are effectively improved, and the labor cost is reduced. The detection process does not need to contact the battery string, and no secondary damage will be caused to the battery string, so as to realize non-destructive detection. The "sliding window" dynamic area detection and the non-destructive detection method based on machine vision are adopted, and the sliding window detection method of splicing and inspection is adopted, which can be applied to the appearance detection of battery strings of various sizes. While the battery string is collecting data, the appearance detection of the battery string is carried out, which effectively shortens the detection time and improves the detection efficiency and real-time performance. The appearance defects of the battery string can be automatically identified and judged according to the battery string standard set by the user, so as to realize fast and accurate defect detection. The battery string picture is clearly displayed and the detection results are statistically analyzed and processed, which helps to determine the source and cause of the defective process, reduce production costs, increase production capacity, and promote the progress of the process. By identifying the splicing method of the head and tail of the battery string, the stable splicing of the battery string can be realized without mis-splicing or missing splicing.

[0161] Based on the same application concept as the above method, a battery string detection device is proposed in the embodiment of the present application. The target battery string includes at least one battery cell. Figure 8 As shown, it is a structural schematic diagram of the device, and the device may include: an acquisition module 81, used to acquire the image to be detected taken during the transmission of the target battery string; a splicing module 82, used to splice the image to be detected with the battery string image of the target battery string in response to acquiring the image to be detected to obtain a spliced ​​image, wherein the battery string image is spliced ​​by the images to be detected taken during the transmission of the target battery string; acquiring a battery cell sub-image from the spliced ​​image, wherein the battery cell sub-image includes a target battery cell, and the target battery cell is a battery cell that has not been inspected; a detection module 83, used to perform defect detection on the target battery cell based on the battery cell sub-image.

[0162] Exemplarily, the image to be detected includes an image to be detected of the upper surface of the target battery string acquired by a front camera and / or an image to be detected of the lower surface of the target battery string acquired by a back camera; the front camera is located above the target battery string, and the back camera is located below the target battery string;

[0163] The image to be inspected on the upper surface is used to perform defect inspection on the upper surface of the target battery string; and the image to be inspected on the lower surface is used to perform defect inspection on the lower surface of the target battery string.

[0164] Exemplarily, the acquisition module 81 is further used to acquire front side inspection data of the upper surface of the target battery string, the front side inspection data including information on whether there are defects on the upper surface of the target battery string, and if there are defects on the upper surface, the front side inspection data also includes the defect type and defect area of ​​the upper surface of the target battery string; acquire back side inspection data of the lower surface of the target battery string, the back side inspection data including information on whether there are defects on the lower surface of the target battery string, and if there are defects on the lower surface, the back side inspection data also includes the defect type and defect area of ​​the lower surface of the target battery string; summarize the front side inspection data and the back side inspection data to obtain appearance inspection data of the target battery string, so that a user can manually inspect whether there are defects on the target battery string based on the appearance inspection data, and obtain the manual inspection result of the target battery string.

[0165] Exemplarily, the acquisition module 81 is also used to obtain a defect detection result of the target battery string after performing defect detection on the target battery cell based on the battery cell image, and the defect detection result includes at least one of the following: a result obtained when performing defect detection on the target battery cell based on the battery cell image, front detection data, back detection data, appearance detection data, and manual detection result; and generate an operation control instruction based on the defect detection result, and the operation control instruction is used to control the flow of the target battery string in a direction corresponding to the defect detection result.

[0166] Exemplarily, during the transportation of multiple battery strings, each battery string passes through the field of view of the camera in turn. The target battery string is any battery string. When the target battery string passes through the field of view of the camera, the camera acquires an image to be inspected for the target battery string.

[0167] Exemplarily, when the stitching module 82 stitches the image to be detected with the battery string image of the target battery string to obtain a stitched image, it is specifically used to: determine whether the number of stitched images of the battery string image is 0; if so, and the image to be detected has the string head feature of the target battery string, then update the image to be detected to the battery string image of the target battery string, and update the number of stitched images to 1; if not, and the number of stitched images does not reach the number threshold, then stitch the image to be detected with the battery string image of the target battery string to obtain a stitched image, and update the stitched image to the battery string image of the target battery string, and add 1 to the number of stitched images; wherein the number threshold is determined based on the configured maximum stitching number.

[0168] Exemplarily, the stitching module 82 is further used to, after determining whether the number of stitched images of the battery string image is 0, if not and the number of stitched images reaches the number threshold, determine whether the image to be detected has the end-of-string feature of the target battery string; if the end-of-string feature does not exist, stitch the image to be detected with the battery string image of the target battery string to obtain a stitched image, update the stitched image to the battery string image of the target battery string, and increase the number of stitched images by 1; if the end-of-string feature exists, stitch the image to be detected with the battery string image of the target battery string to obtain a stitched image, update the stitched image to the complete battery string image of the target battery string, output the complete battery string image, and update the number of stitched images to 0.

[0169] Exemplarily, the stitching module 82 is further used to, before updating the stitched image to the complete battery string image of the target battery string and outputting the complete battery string image, determine whether a string end signal is received in the presence of a string end feature, the string end signal indicating that the last battery cell of the target battery string is transported to the camera's field of view; if so, update the stitched image to the complete battery string image of the target battery string and output the complete battery string image; if not, output abnormal information; after updating the stitched image to the complete battery string image of the target battery string and outputting the complete battery string image, determine a sub-image of each battery cell from the complete battery string image, and add a defect detection result of the battery cell to the sub-image of the battery cell.

[0170] Exemplarily, the splicing module 82 is further used to determine whether the image to be detected is a valid image before splicing the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image; if the image to be detected is a valid image, then perform an operation of splicing the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image; if the image to be detected is an invalid image, then discard the image to be detected; wherein, if the brightness mean of the battery cell area in the image to be detected is less than the brightness threshold, at least one of the image to be detected having a string head feature and the image to be detected having a string tail feature is established, then the image to be detected is a valid image; otherwise, the image to be detected is an invalid image.

[0171] Exemplarily, the stitching module 82 is also used to determine edge pixel points of the target battery string from the image to be detected; if a first edge straight line and a second edge straight line are fitted based on the edge pixel points, the distance between the first edge straight line and the second edge straight line is greater than a distance threshold, the average brightness of the transition area between the first edge straight line and the second edge straight line is greater than a first brightness threshold, and the average brightness of the battery cell area in the image to be detected is less than a second brightness threshold, then it is determined that the image to be detected has a string head feature and a string tail feature; wherein the second brightness threshold is less than the first brightness threshold.

[0172] Based on the same application concept as the above method, an electronic device is proposed in the embodiment of the present application, see Fig. 9 As shown, the electronic device includes: a processor 91 and a machine-readable storage medium 92, the machine-readable storage medium 92 stores machine-executable instructions that can be executed by the processor 91; the processor 91 is used to execute the machine-executable instructions to implement the battery string detection method disclosed in the above example of this application.

[0173] Based on the same application concept as the above method, an embodiment of the present application also provides a battery string detection system, including: an image acquisition device, a battery string detection device, a control device, an MES device and a centralized control device; wherein: the image acquisition device is used to acquire the image to be detected during the transmission process of the target battery string; the battery string detection device is used to execute the battery string detection method of the above example based on the image to be detected, and obtain the defect detection result corresponding to the target battery string; the control device is used to control the transmission device and the image acquisition device; the MES device is used to obtain the defect detection result corresponding to the target battery string, and process the target battery string based on the defect detection result; the centralized control device is used to obtain the defect detection result corresponding to the target battery string, so that the user can manually detect whether the target battery string has defects based on the defect detection result, obtain the manual detection result of the target battery string, and return the manual detection result to the battery string detection device.

[0174] Based on the same application concept as the above method, an embodiment of the present application further provides a machine-readable storage medium, on which a number of computer instructions are stored. When the computer instructions are executed by a processor, the battery string detection method disclosed in the above example of the present application can be implemented.

[0175] The above-mentioned machine-readable storage medium may be any electronic, magnetic, optical or other physical storage device, which may contain or store information, such as executable instructions, data, etc. For example, the machine-readable storage medium may be: RAM (Radom Access Memory), volatile memory, non-volatile memory, flash memory, storage drive (such as hard disk drive), solid state drive, any type of storage disk (such as CD, DVD, etc.), or similar storage medium, or a combination thereof.

[0176] Based on the same application concept as the above method, an embodiment of the present application further provides a computer program product, which may include a computer program. When the computer program is executed by a processor, the battery string detection method disclosed in the above example of the present application is implemented.

[0177] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the embodiments of the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0178] The above is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included in the scope of the claims of the present application.

Claims

1. A method for detecting a battery string, characterized in that: The target battery string includes at least one battery cell, and the method includes: Acquire an image to be detected taken during the transmission process of the target battery string; In response to acquiring the image to be detected, splicing the image to be detected with a battery string image of the target battery string to obtain a spliced ​​image, wherein the battery string image is spliced ​​by the images to be detected taken during the transmission of the target battery string; Acquire a cell sub-image from the stitched image, wherein the cell sub-image includes a target cell, and the target cell is a cell that has not been inspected; Defect detection is performed on the target cell based on the cell sub-image.

2. The method according to claim 1, characterized in that The image to be detected includes an image to be detected of the upper surface of the target battery string acquired by a front camera and / or an image to be detected of the lower surface of the target battery string acquired by a back camera; the front camera is located above the target battery string, and the back camera is located below the target battery string; The image to be inspected on the upper surface is used to perform defect inspection on the upper surface of the target battery string; and the image to be inspected on the lower surface is used to perform defect inspection on the lower surface of the target battery string.

3. The method according to claim 2, characterized in that The method further comprises: Acquire front detection data of the upper surface of the target battery string, the front detection data including information on whether there is a defect on the upper surface of the target battery string, and if there is a defect on the upper surface, the front detection data also includes the defect type and defect area on the upper surface of the target battery string; Acquire back surface inspection data of the lower surface of the target battery string, the back surface inspection data including information on whether there is a defect on the lower surface of the target battery string, and if there is a defect on the lower surface, the back surface inspection data also includes the defect type and defect area of ​​the lower surface of the target battery string; The front detection data and the back detection data are summarized to obtain appearance detection data of the target battery string, so that a user can manually detect whether the target battery string has defects based on the appearance detection data and obtain a manual detection result of the target battery string.

4. The method according to any one of claims 1 to 3, characterized in that: After performing defect detection on the target cell based on the cell sub-image, the method further includes: Obtaining a defect detection result of the target battery string, wherein the defect detection result includes at least one of the following: a result obtained when defect detection is performed on the target battery cell based on the battery cell sub-image, front detection data, back detection data, appearance detection data, and manual detection results; An operation control instruction is generated according to the defect detection result, and the operation control instruction is used to control the target battery string to flow in a direction corresponding to the defect detection result.

5. The method according to claim 1, characterized in that During the transportation of multiple battery strings, each battery string passes through the field of view of the camera in turn. The target battery string is any battery string. When the target battery string passes through the field of view of the camera, the camera acquires an image to be inspected for the target battery string. The step of splicing the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image includes: Determine whether the number of stitched images of the battery string image is 0; If yes, and the image to be detected has the string head feature of the target battery string, then the image to be detected is updated to the battery string image of the target battery string, and the number of stitched images is updated to 1; If not, and the number of the stitched images does not reach the number threshold, stitching the image to be detected with the battery string image of the target battery string to obtain a stitched image, and updating the stitched image to the battery string image of the target battery string, and adding 1 to the number of the stitched images; The quantity threshold is determined based on the configured maximum splicing quantity.

6. The method according to claim 5, characterized in that After determining whether the number of spliced ​​images of the battery string image is 0, the method further includes: If not, and the number of the stitched images reaches the number threshold, determining whether the image to be detected has the end-of-string feature of the target battery string; if the end-of-string feature does not exist, stitching the image to be detected with the battery string image of the target battery string to obtain a stitched image, updating the stitched image to the battery string image of the target battery string, and adding 1 to the number of stitched images; If there is a string tail feature, the image to be detected is spliced ​​with the battery string image of the target battery string to obtain a spliced ​​image, the spliced ​​image is updated to a complete battery string image of the target battery string, the complete battery string image is output, and the number of spliced ​​images is updated to 0.

7. The method according to claim 6, characterized in that Before the stitched image is updated to the complete battery string image of the target battery string and the complete battery string image is output, the method further comprises: in the case where there is a string tail feature, determining whether a string tail signal is received, the string tail signal indicating that the last battery cell of the target battery string is transported to the field of view of the camera; if so, updating the stitched image to the complete battery string image of the target battery string and outputting the complete battery string image; if not, outputting abnormal information; After updating the stitched image to a complete battery string image of the target battery string and outputting the complete battery string image, the method further comprises: determining a sub-image of each battery cell from the complete battery string image, and adding a defect detection result of the battery cell to the sub-image of the battery cell.

8. The method according to claim 1, characterized in that Before splicing the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image, the method further includes: Determine whether the image to be detected is a valid image; if the image to be detected is a valid image, perform an operation of splicing the image to be detected with the battery string image of the target battery string to obtain a spliced ​​image; if the image to be detected is an invalid image, discard the image to be detected; If the brightness mean of the cell area in the image to be detected is less than the brightness threshold, and at least one of the string head feature and the string tail feature of the image to be detected is true, then the image to be detected is a valid image; otherwise, the image to be detected is an invalid image; The method further includes: determining edge pixel points of a target battery string from the image to be detected; if a first edge straight line and a second edge straight line are fitted based on the edge pixel points, the spacing between the first edge straight line and the second edge straight line is greater than a distance threshold, the average brightness of a transition area between the first edge straight line and the second edge straight line is greater than a first brightness threshold, and the average brightness of a battery cell area in the image to be detected is less than a second brightness threshold, then it is determined that the image to be detected has a string head feature and a string tail feature; wherein the second brightness threshold is less than the first brightness threshold.

9. A battery string detection device, characterized in that: The target battery string includes at least one battery slice, and the device includes: An acquisition module, used for acquiring an image to be detected taken during the transmission process of the target battery string; a splicing module, configured to, in response to acquiring the image to be detected, splice the image to be detected with a battery string image of the target battery string to obtain a spliced ​​image, wherein the battery string image is spliced ​​by images to be detected taken during the transmission of the target battery string; and to obtain a battery cell sub-image from the spliced ​​image, wherein the battery cell sub-image includes a target battery cell, and the target battery cell is a battery cell that has not been detected; A detection module is used to perform defect detection on the target battery cell based on the battery cell sub-image.

10. A battery string detection system, characterized in that: include: Image acquisition equipment, battery string detection equipment, control equipment, MES equipment and centralized control equipment; among which: The image acquisition device is used to acquire the image to be detected during the transmission process of the target battery string; The battery string detection device is used to perform the method according to any one of claims 1 to 7 based on the image to be detected to obtain a defect detection result corresponding to the target battery string; The control device is used to control the transmission device and the image acquisition device; The MES device is used to obtain a defect detection result corresponding to the target battery string, and process the target battery string based on the defect detection result; The centralized control device is used to obtain the defect detection result corresponding to the target battery string, so that the user can manually detect whether the target battery string has defects based on the defect detection result, obtain the manual detection result of the target battery string, and return the manual detection result to the battery string detection device.

Images