Dust removal detection method and device of battery module, storage medium and terminal equipment

Abstract

The invention discloses a dust removal detection method and device for a battery module, a storage medium and terminal equipment. The method comprises the following steps: acquiring video data entering a target detection area; when it is determined that at least one target battery module is included according to the video data, action detection is carried out on the area where the target battery module is located; and marking the state of the battery module according to the action detection result. According to the dust removal detection method, detection is started when the battery module enters the detection area, action detection is performed on the battery module needing dust removal in the video data, and the state of the battery module is marked according to the action detection result, so that the problem of excessive killing caused by the existence of the module without dust collection and manual module carrying is solved; and multi-path parallel detection is realized.

Description

Technical Field

[0001] The present application relates to the field of image recognition technology, and in particular to a dust removal detection method, device, storage medium and terminal device for a battery module. Background Art

[0002] At present, the post-welding cleaning of battery modules requires the staff to vacuum voluntarily, so metal welding slag residue often occurs, causing quality incidents. The main way to detect violations of post-welding vacuuming of the busbar of the battery module is for the auditors to monitor whether the on-site staff have performed vacuuming in accordance with regulations by watching the surveillance video or on-site inspections. The auditors cannot monitor in real time whether the employees pulling the wires have operated correctly, which is inefficient and detection is easily delayed. Summary of the invention

[0003] In view of the above problems, the present application provides a dust removal detection method, device, storage medium and terminal device for a battery module. The detection begins when the battery module enters the detection area, and motion detection is performed on the battery modules that need dust removal in the video data. The status of the battery module is marked according to the motion detection result, which solves the problem of over-detection caused by modules that do not need dust removal and manual handling of modules, and realizes multi-channel parallel detection.

[0004] In a first aspect, the present application provides a dust removal detection method for a battery module, the method comprising: acquiring video data of entering a target detection area; when it is determined based on the video data that at least one target battery module is included, performing motion detection on the area where the target battery module is located; and marking the state of the target battery module based on the motion detection result.

[0005] In the technical solution of the embodiment of the present application, after the battery module enters the target detection area, video data of the battery module is acquired and the video data is processed, such as feature extraction, to determine whether there is a target battery module in the video data. The target battery module may include a battery module that needs to be vacuumed. When the video data only includes an empty vehicle, a master sample, and a battery module without a bar, and does not include a battery module that needs to be vacuumed after Busbar welding, no action detection is performed on this area. By filtering out the components that do not need to be vacuumed, the detection efficiency is improved. When it is determined that there is a battery module that needs to be dusted in the video data, action detection is performed on the area where the target battery module is located. The action detection includes a vacuuming action and a cutting pliers action. If, when the target battery module leaves the target detection area, the detection result is a vacuuming action and there is no cutting pliers action, it is determined that the detection result is a vacuuming action, and the status of the battery module is marked as OK; if, when the target battery module leaves the target detection area, the detection result shows a cutting pliers action, it is determined that the detection result is a cutting pliers action, and the status of the battery module is marked as NG. Among them, the cutting pliers action means that the operator judges that if there are solder bumps generated by welding in the welding area, trimming needs to be performed with pliers. After the operator performs the cutting pliers action, secondary vacuuming is also required to reduce the quality risk of uncleaned welding slag.

[0006] In some embodiments, marking the status of the target battery module according to the action detection result includes: when the action detection result is a vacuuming action, marking the status of the target battery module as the first status; when the action detection result is a cutting pliers action, marking the status of the target battery module as the second status. By marking the status of each target battery module in the detection area, it is convenient to subsequently determine whether the video data needs to be saved according to the status of the target battery module, so as to view this section of video later and assist in management.

[0007] In some embodiments, the above dust removal detection method for the battery module further includes: when it is determined that the target battery module leaves the target detection area, if there is a target battery module with the status of the second status, save the video data and push a reminder message; if there is no target battery module with the status of the second status, delete the video data. When it is detected that the target battery module leaves the target detection area, the status of each target battery module is detected. If the status of all target battery modules is that there is no cutting pliers action (second status), it is determined that the battery modules in the acquired video data are vacuumed in accordance with the specifications, and the current video data does not need to be saved, that is, the video data is deleted to free up memory. If there is a target battery module with the status of a cutting pliers action (second status), it is determined that there is a situation where the battery modules in the acquired video data are not vacuumed in accordance with the specifications, save the video data and push a reminder message to view the video data and assist in management.

[0008] In some embodiments, the target detection area includes a start area, a detection area, and a leaving area. The target battery module is marked in the start area, and when the number of frames including the target battery module in the detection area exceeds a first preset number of frames, it is determined that the target battery module enters the target detection area, and a detection action is performed. Thus, it is possible to perform action detection when it is determined that the battery module to be detected enters the start area, preventing false detection.

[0009] In some embodiments, when one of the following conditions is met, it is determined that the target battery module leaves the target detection area: the target battery module enters the leaving area and the number of frames including the target battery module exceeds a second preset number of frames; the time when the target battery module enters the target detection area exceeds a first preset time; the target battery module is not detected in the detection area and lasts for a second preset time. Thus, it is possible to accurately determine whether the battery module leaves the target detection area, reduce the probability of misoperation, and avoid the situation where the video is continuously saved when the operator leaves without protection, resulting in excessive memory occupation of the video, or the situation where the battery module is protected after the operator leaves and the module is not detected for a long time, affecting the subsequent detection process.

[0010] In some embodiments, determining that the target battery module is included according to the video data includes: processing the video data to obtain image data; classifying the image data using a preset image classification model; and determining that the target battery module is included when it is determined that there is a battery module according to the classification result. Thus, the components in the detection area can be divided into two categories, one category includes the modules that need to be vacuumed, and the other category includes the modules that do not need to be vacuumed. For the empty vehicle, the detection table base, and the battery module sample, it is not necessary to perform area marking and judgment according to the action, solving the overkill caused by the products that do not need to be vacuumed on site and improving the detection efficiency.

[0011] In some embodiments, before performing action detection on the area where the target battery module is located, the above method further includes: predicting the area where the battery module is located based on the target detection model to obtain a prediction result; and setting a plurality of prediction frames with different confidence levels based on non-maximum suppression to filter the prediction result. For the vacuuming action, the cutting pliers action, the scanning code action, and the tapping bar action, due to reasons such as the tool being blocked, similar gestures, different habits of each operator using the same tool, and the differences of the tools, a plurality of different confidence levels are set to filter out the scanning code action and the tapping bar action similar to the vacuuming action or the cutting pliers action, improving the detection efficiency and detection accuracy.

[0012] In a second aspect, the present application provides a dust removal detection device for a battery module, including: an acquisition module configured to acquire video data entering a target detection area; a detection module configured to perform motion detection on the area where a target battery module is located when it is determined according to the video data that at least one target battery module is included; and a marking module configured to mark the status of the target battery module according to the motion detection result.

[0013] In the technical solution of the embodiment of the present application, after the battery module enters the target detection area, the acquisition module acquires the video data of the battery module and processes the video data, such as performing feature extraction, to determine whether there is a target battery module in the video data. The target battery module may include a battery module that needs dust suction. When the video data only includes an empty carrier, a master sample, and a battery module without a bus bar, and does not include a battery module that needs dust suction after Busbar welding, motion detection is not performed on this area, and by filtering out components that do not require dust suction, the detection efficiency is improved. When the detection module determines that there is a battery module that needs dust removal in the video data, it performs motion detection on the area where the target battery module is located. The motion detection includes a dust suction motion and a cutting pliers motion. If all the motions in this area are dust suction motions and there is no cutting pliers motion, it is determined that the detection result is a dust suction motion, and the marking module marks the status of the battery module as OK; if there is a cutting pliers motion in this area, it is determined that the detection result is a cutting pliers motion, and the marking module marks the status of the battery module as NG. Among them, the cutting pliers motion means that if there are raised parts after Busbar welding, they need to be clipped off with pliers, and after the operator performs the cutting pliers motion, secondary dust suction is also required to reduce the quality risk of uncleaned welding slag.

[0014] In a third aspect, the present application provides a computer-readable storage medium, on which a dust removal detection program for a battery module is stored. When the dust removal detection program for the battery module is executed by a processor, the above-mentioned dust removal detection method for the battery module is implemented.

[0015] In a fourth aspect, the present application provides a terminal device, which is characterized by including a memory, a processor, and a dust removal detection program for a battery module stored on the memory and executable on the processor. When the processor executes the dust removal detection program for the battery module, the above-mentioned dust removal detection method for the battery module is implemented.

[0016] The above description is only an overview of the technical solution of the present application. In order to be able to understand the technical means of the present application more clearly, it can be implemented according to the content of the specification. And in order to make the above and other purposes, features, and advantages of the present application more obvious and understandable, the following specifically gives the specific implementation manners of the present application. Description of the Drawings

[0017] Upon reading the following detailed description of the preferred embodiments, various other advantages and benefits will become clear to those of ordinary skill in the art. The drawings are only for the purpose of showing the preferred embodiments and are not considered to be a limitation of the present application. Moreover, in all the drawings, the same reference numerals are used to denote the same components. In the drawings:

[0018] Figure 1 is a flowchart of a dust removal detection method for a battery module according to some embodiments of the present application;

[0019] Figure 2 is an architecture diagram of a dust removal detection system for a battery module according to some embodiments of the present application;

[0020] Figure 3 is a schematic diagram of a dust removal detection method for a battery module according to some embodiments of the present application;

[0021] Figure 4 is a block schematic diagram of a dust removal detection device for a battery module according to some embodiments of the present application;

[0022] Figure 5 is a block schematic diagram of a terminal device according to some embodiments of the present application. Detailed Embodiments

[0023] The embodiments of the technical solutions of the present application will be described in detail below with reference to the drawings. The following embodiments are only used to illustrate the technical solutions of the present application more clearly, and thus are only examples and cannot be used to limit the protection scope of the present application.

[0024] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which this application belongs; the terms used herein are only for the purpose of describing specific embodiments and are not intended to limit this application; the terms "including" and "having" and any variations thereof in the specification and claims of this application and the above drawings are intended to cover non-exclusive inclusion.

[0025] In the description of the embodiments of the present application, technical terms such as "first" and "second" are only used to distinguish different objects and cannot be understood as indicating or implying relative importance or implicitly indicating the quantity, specific order or primary-secondary relationship of the indicated technical features. In the description of the embodiments of the present application, the meaning of "a plurality" is more than two, unless otherwise specifically defined.

[0026] References to "embodiments" in this specification mean that the particular features, structures, or characteristics described in connection with the embodiments can be included in at least one embodiment of the present application. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive of other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0027] In the description of the embodiments of the present application, the term "and / or" is merely a description of the association relationship of associated objects, indicating that there can be three relationships, for example, A and / or B, which can represent: A exists alone, A and B exist simultaneously, and B exists alone. In addition, the character " / " in this text generally represents an "or" relationship between the associated objects before and after.

[0028] In the description of the embodiments of the present application, the term "plurality" refers to two or more (including two). Similarly, "multiple groups" refers to two or more groups (including two groups), and "multiple pieces" refers to two or more pieces (including two pieces).

[0029] In the description of the embodiments of the present application, the orientation or positional relationship indicated by technical terms such as "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", "clockwise", "counterclockwise", "axial", "radial", "circumferential", etc. is based on the orientation or positional relationship shown in the drawings. It is only for the convenience of describing the embodiments of the present application and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be construed as a limitation on the embodiments of the present application.

[0030] In the description of the embodiments of the present application, unless otherwise clearly specified and limited, technical terms such as "installation", "connection", "connection", "fixation", etc. should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or integrated; it can also be a mechanical connection or an electrical connection; it can be directly connected or indirectly connected through an intermediate medium, and it can be the communication inside two elements or the interaction relationship between two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the embodiments of the present application can be understood according to specific circumstances.

[0031] At present, the post-welding cleaning of battery modules requires workers to vacuum dust consciously, so there are often problems of residual metal welding slag, resulting in quality incidents. The method for detecting violations of vacuuming after Busbar welding of battery modules mainly relies on inspectors to monitor whether on-site workers vacuum dust properly by checking surveillance videos or conducting on-site inspections. Inspectors cannot monitor in real time whether each worker on each production line is operating correctly, resulting in low efficiency and extremely easy detection delays. Therefore, AI (Artificial Intelligence) detection means are needed to identify whether the visual inspection post after welding vacuums dust properly (each welded battery module needs to ensure one vacuuming action. If the operator trims the metal welding slag, the operator is required to perform a second vacuuming), warn against improper vacuuming behavior, assist in management to prevent Particle insulation failure, and reduce the quality risk of uncleaned welding slag.

[0032] Combined with the actual scenario, the problems faced include: the materials coming out of the machine at some points are those that need to be vacuumed after welding, and those manually placed in the detection area do not need to judge whether they have been vacuumed; the starting time point of module vacuuming is not fixed, and there may be multiple modules in the same area undergoing vacuuming detection; at the same detection point, there may be an empty carrier, a master sample, a module without a busbar piece, and a module that needs to be vacuumed after Busbar welding appearing simultaneously. The first three situations do not require vacuuming and need to be filtered; for NG behavior, it is necessary to save the video of the entire module working process to assist inspectors in judging whether the employee has not vacuumed or has not vacuumed after trimming. However, problems such as manual handling of the module during the process, long-term absence of personnel from the module without protection while still at the visual inspection post without moving will affect the judgment at the end of the process; due to reasons such as the use of tools being blocked, similar gestures, different habits of each employee using the same tool, and tool differences for vacuuming and trimming actions, there are significant differences in the confidence levels of identifying vacuuming and trimming actions at different points.

[0033] In this application, the process is judged to start only when the battery module enters the detection area through the machine transfer entrance, solving the overkill caused by manually handling materials or battery modules that do not require vacuuming; after starting the detection, it is judged whether there are battery modules that need to be vacuumed. If there are battery modules that need to be vacuumed but no vacuuming action is detected, the entire process is marked as NG, solving the overkill caused by changes in the number of battery modules and realizing the simultaneous detection of multiple target battery modules in the detection area.

[0034] For the convenience of explanation in the following embodiments, combined with Figure 1 the dust removal detection method of the battery module of this application will be described.

[0035] Refer to Figure 1 , the dust removal detection method of the battery module in this application may include the following steps:

[0036] S101, Obtain the video data entering the target detection area.

[0037] S102, When it is determined according to the video data that there is at least one target battery module, perform motion detection on the area where the target battery module is located respectively.

[0038] S103, Mark the status of the target battery module according to the motion detection result.

[0039] Specifically, the battery module assembly process includes: pole column addressing -> CCS installation -> busbar welding -> post-weld cleaning -> weld seam detection. This application is used for post-weld cleaning inspection of the busbar. After the battery module enters the target detection area, obtain the video data of the battery module and process the video data, such as feature extraction, to determine whether there is a target battery module in the video data. The target battery module can include the battery module that needs to be vacuumed. When the video data only includes the empty fixture, the master sample, and the battery module without the busbar piece, and does not include the battery module that needs to be vacuumed after the busbar welding, do not perform motion detection on this area, and improve the detection efficiency by filtering out the components that do not need to be vacuumed. When it is determined that there is a battery module that needs to be dusted in the video data, perform motion detection on the area where the target battery module is located. The motion detection includes the vacuuming action and the pliers action. If when the target battery module leaves the target detection area, the detection result is the vacuuming action and there is no pliers action, then determine that the detection result is the vacuuming action and mark the status of the battery module as OK; if when the target battery module leaves the target detection area, the detection result has the pliers action, then determine that the detection result is the pliers action and mark the status of the battery module as NG. Among them, the pliers action means that the operator judges that if there are solder bumps generated by welding in the welding area, it is necessary to trim with pliers. After the operator performs the pliers action, secondary vacuuming is also required to reduce the quality risk of uncleaned welding slag.

[0040] In some embodiments, marking the status of the target battery module according to the motion detection result includes: when the motion detection result is the vacuuming action, marking the status of the target battery module as the first status; when the motion detection result is the pliers action, marking the status of the target battery module as the second status.

[0041] Specifically, after the battery module enters the detection area, each target battery module is marked with a label, and each target battery module is subjected to motion detection. If it is detected that the operator performs a dust suction operation on the target battery module, the first state is determined to be OK; if it is detected that the operator performs a cutting pliers action on the target battery module, the second state is determined to be NG. Among them, the state of the target battery module is generally determined according to the last action of the target battery module leaving the target detection area; for components in the detection area that are not target battery modules, the state of the component is marked as OK, and the state of the component is not changed when it is in the detection area, improving the detection efficiency. By marking the state of each target battery module in the detection area, it is convenient to determine whether video data needs to be saved according to the state of the target battery module for subsequent viewing of the video segment to assist management.

[0042] In some embodiments, the above dust removal detection method for the battery module further includes: when it is determined that the target battery module leaves the target detection area, if there is a target battery module with the second state, save the video data and push a reminder message; if there is no target battery module with the second state, delete the video data.

[0043] Specifically, when it is detected that the target battery module leaves the target detection area, the state of each target battery module is detected. If the states of all target battery modules are dust suction actions (the first state), it is determined that the battery modules are dusted in accordance with the specifications in the acquired video data, and the current video data does not need to be saved, that is, the video data is deleted to free up memory. If there is a target battery module with the state of cutting pliers action (the second state), it is determined that there is a situation where the battery modules are not dusted in accordance with the specifications in the acquired video data, and the video data is saved and a reminder message is pushed to facilitate viewing of the video data to assist management. Moreover, the function of asynchronous video saving is added, so that the video data saved during detection and the saved NG video data do not affect each other, improving the output rate of NG video data and the real-time performance of video stream reading.

[0044] Exemplarily, the current video data includes two target battery modules that need to be vacuumed, denoted as target battery module A and target battery module B. Motion detection is performed on target battery module A and target battery module B respectively. Among them, target battery module A has gone through a vacuuming motion within the target detection area, and target battery module B has gone through a vacuuming motion and a cutting pliers motion within the target detection area. At this time, there is a cutting pliers motion in the detection result. Target battery module A is marked as OK, target battery module B is marked as NG, and the current video data is saved. Exemplarily, the current video data includes a target battery module that needs to be vacuumed, denoted as target battery module C. Motion detection is performed on target battery module C. When target battery module C has gone through a vacuuming motion, a cutting pliers motion, and a second vacuuming motion within the target detection area, the detection result at this time is a vacuuming motion. Target battery module C is marked as OK, and the current video is deleted to release memory.

[0045] It should be noted that in the above embodiments, the motion detection results of the target battery modules are all determined by the last motion when the target battery modules leave the target detection area. For example, if the last motion is a vacuuming motion, the motion detection result is a vacuuming motion; if the last motion is a cutting pliers motion, the motion detection result is a cutting pliers motion.

[0046] In some embodiments, the target detection area includes a start area, a detection area, and a leaving area. The target battery module is marked in the start area, and when the number of frames containing the target battery module in the detection area exceeds a first preset number of frames, it is determined that the target battery module enters the target detection area, and the detection action is executed. Among them, the first preset number of frames can be calibrated according to the actual situation. For example, the first preset number of frames can be 5 frames.

[0047] Specifically, the target detection area can be divided into three sub-areas, in the order of the start area, the detection area, and the leaving area. That is, the battery module first reaches the start area, queues in the start area to enter the detection area, and the operator completes the vacuuming motion and the cutting pliers motion of the target battery module in the detection area, and then enters the leaving area to enter the next process. When the battery module enters the start area, the battery module is identified as the target battery module and marked. When the target battery module enters the detection area, it is judged whether the number of frames of the target battery module in the detection area exceeds 5 frames. If it exceeds, the motion detection is started; if it does not exceed, the motion detection is not performed, so that the motion detection can be performed when it is determined that the battery module to be detected enters the start area, preventing false detection.

[0048] In some embodiments, when one of the following conditions is met, it is determined that the target battery module leaves the target detection area: the target battery module enters the leaving area and the number of frames containing the target battery module exceeds the second preset number of frames; the time for the target battery module to enter the target detection area exceeds the first preset time; the target battery module is not detected in the detection area and lasts for the second preset time. Among them, the second preset number of frames, the first preset time, and the second preset time can be calibrated according to the actual situation. For example, the second preset number of frames can be 5 frames, the second preset time can be 2 min, and the first preset time can be 30 min.

[0049] That is to say, when the detection area and the leaving area are relatively close, it is easy to cause instability in the detection of the size of the module target due to the addition and increase of human movements, resulting in part or all of the target battery module in the detection area entering the leaving area. Therefore, when the number of frames for the target battery module to enter the leaving area exceeds 5 frames, it is considered that the target battery module leaves the target detection area. Or, when the operator leaves the work station without protecting the target battery module, the video data is continuously saved, resulting in excessive memory occupation of the video. Therefore, when the time for the target battery module to enter the target detection area exceeds the first preset time, it is considered that the target battery module leaves the target detection area. Or, when the operator leaves the work station, protects the target battery module, and the target battery module is not detected for a long time, it is considered that the target battery module leaves the target detection area. Thus, it is possible to accurately determine whether the battery module leaves the target detection area, reduce the probability of misoperation, and avoid excessive memory occupation of the video caused by continuous saving of the video when the operator leaves without protection, or the subsequent detection process is affected because the battery module is not detected for a long time after the operator adds protection to the battery module.

[0050] In some embodiments, determining that the video data contains the target battery module includes: processing the video data to obtain image data; classifying the image data using a preset image classification model; and determining that the target battery module is included when it is determined that there is a battery module according to the classification result.

[0051] Specifically, the video data is processed in the target detection area to obtain image data, and the image data is input into a pre-trained classification model. This classification model can divide the image data into two categories. One category is the presence of a battery module (including the target battery module), and the other category is the absence of a battery module (including an empty carrier, a detection table base, and a module that does not require dust collection, such as a battery module sample). When the classification result is the category of the presence of a battery module, it is determined that the video data includes the battery module that needs dust collection, that is, the target battery module. Thus, the components in the detection area can be divided into two categories. For the empty carrier, the detection table base, and the battery module sample, no action detection is required, which improves the detection efficiency.

[0052] It should be noted that since the color of the battery module sample is different from the appearance of the battery module after welding (including color, labels pasted on the module or tapes pasted on special positions, etc.), and the battery module sample does not contain busbar pieces, it is also possible to determine whether the video data contains the target battery module according to the color of the battery module, the presence or absence of busbar pieces, etc.

[0053] In some embodiments, before performing action detection on the area where the target battery module is located, the above method further includes: predicting the area where the battery module is located based on a target detection model to obtain a prediction result; and setting multiple different confidence prediction boxes based on non-maximum suppression to filter the prediction result.

[0054] Specifically, the target detection model is obtained by labeling and training the targets of actions such as module cutting pliers, vacuuming, tapping busbar pieces, and screwdriver scribing using the labelme annotation software. For example, using the target detection YOLO-v5, the non_max_suppression part is modified to support different confidence filtering for multiple targets. After entering the target detection area, the video data is screened, different confidences are set, the video data of the code scanning action and the tapping busbar piece action are filtered out, and the videos of the vacuuming action and the cutting pliers action are retained, and then action detection is performed. For the vacuuming action, cutting pliers action, code scanning action, and tapping busbar piece action, due to reasons such as the tool being blocked, similar gestures, different habits of each operator using the same tool, and tool differences, multiple different confidences are set to filter out the code scanning action and the tapping busbar piece action similar to the vacuuming action or the cutting pliers action, etc., to improve the detection efficiency and detection accuracy.

[0055] As a specific example, as Figure 2 and Figure 3 shown, by calling the monitoring platform API (Application Programming Interface), the real-time video stream is obtained, the RTSP (Real Time Streaming Protocol) is used to fetch the stream, the target detection YOLO-v5 and the image classification model LeNet are used to perform real-time detection of the illegal operations of the personnel vacuuming after Busbar welding on the AI computing inference server, and the result data is summarized and written into the database. At the same time, the abnormal result display data is written into the shared path through CIFS (Common Internet File System) mounting and pushed to the corresponding warning platform to complete the business closed-loop.

[0056] Retrieve the video data of on-site surveillance cameras through RTSP, collect the dataset of normal operation behaviors of the Busbar post-welding visual inspection position, and filter and screen the data. Use the labelme annotation software to annotate and train the targets of actions such as module, cutting pliers, dust suction, tapping the busbar, and screw marking to obtain an object detection model (YOLO-v5 object detection model). Classify the battery modules, empty carriers, and the base of the inspection table into two categories: the presence and absence of modules, and use the LeNet algorithm to train the classification model to obtain a classification model for judging whether there is a module that needs dust suction in each module area.

[0057] Obtain the monitoring video stream information and the object detection area of each point to be inspected, including the start area, detection area, and departure area. Detect each battery module area. If the battery module has a sign of passing through the start area and is recognized in the detection area for more than 5 frames, it is determined that the battery module starts detection, and the result video is initialized and newly saved. If the battery module starts detection and there is a battery module label in the current frame, use the image classification model to classify and identify the areas where each battery module of multiple battery modules is located to check for the situation of empty modules. Detect each area where the battery module is located separately. If the current area is an empty module situation, skip it and do not make dust suction and cutting pliers judgments; if the module in the current area is not empty, perform dust suction and cutting pliers action detection. If the current wire-pulling module starts detection and the dust suction action exists in the battery module detection area, set the current module status to OK. If the previous wire-pulling module starts detection and the cutting pliers action exists in the module detection area, set the current battery module status to NG. If any of the following three conditions are met: the battery module has started detection and is recognized as having passed through the departure area for more than 5 frames, or the detection duration of the battery module exceeds the specified time, or no module has been detected for more than 2 minutes, it is determined that the battery module has left the object detection area. If the current frame is the frame when the battery module leaves, check whether there is an NG situation in each battery module area. If there is an NG situation in one area, save the NG video; otherwise, empty the currently saved queue and delete the video.

[0058] When a dust suction violation operation is detected at the Busbar post-welding visual inspection position, push the message to the abnormal result feedback platform and the group where the relevant person in charge is located, and save the entire violation process to the specified path for archiving to achieve real-time alarm for violation events.

[0059] In summary, in the technical solution of the embodiment of the present application, after the battery module enters the target detection area, video data of the battery module is acquired and processed, for example, feature extraction is performed to determine whether there is a target battery module in the video data. The target battery module may include a battery module that needs to be vacuumed. When the video data only includes an empty tool, a master sample, and a battery module without a bar, and does not include a battery module that needs to be vacuumed after Busbar welding, action detection is not performed on this area. By filtering out components that do not need to be vacuumed, the detection efficiency is improved. When it is determined that there is a battery module that needs to be dusted in the video data, action detection is performed on the area where the target battery module is located. The action detection includes a vacuuming action and a cutting pliers action. If, when the target battery module leaves the target detection area, the detection result is a vacuuming action and there is no cutting pliers action, it is determined that the detection result is a vacuuming action, and the status of the battery module is marked as OK; if, when the target battery module leaves the target detection area, the detection result shows a cutting pliers action, it is determined that the detection result is a cutting pliers action, and the status of the battery module is marked as NG. Among them, the cutting pliers action means that if there are raised areas after Busbar welding, they need to be clipped off with pliers. After the operator performs the cutting pliers action, secondary vacuuming is also required to reduce the quality risk of uncleaned welding slag.

[0060] Corresponding to the above embodiment, the present application also proposes a dust removal detection device for a battery module.

[0061] As Figure 4 shown, the dust removal detection device 100 for a battery module according to an embodiment of the present application may include: an acquisition module 110, a detection module 120, and a marking module 130.

[0062] Among them, the acquisition module 110 is used to acquire video data entering the target detection area. The detection module 120 is used to perform action detection on the area where the target battery module is located respectively when it is determined according to the video data that at least one target battery module is included. The marking module 130 is used to mark the status of the target battery module according to the action detection result.

[0063] Specifically, the battery module assembly process includes: pole column addressing -> CCS installation -> busbar welding -> post-weld cleaning -> weld seam detection. This application is used for post-weld cleaning inspection of the busbar. After the battery module enters the target detection area, the acquisition module 110 acquires video data of the battery module and processes the video data, such as performing feature extraction, to determine whether there is a target battery module in the video data. The target battery module may include a battery module that needs to be vacuumed. When the video data only includes an empty fixture, a master sample, and a battery module without a busbar, and does not include a battery module that needs to be vacuumed after busbar welding, no action detection is performed on this area, and by filtering out components that do not need to be vacuumed, the detection efficiency is improved. When the detection module 120 determines that there is a battery module that needs to be dusted in the video data, it performs action detection on the area where the target battery module is located. The action detection includes a vacuuming action and a snipping action. If when the target battery module leaves the target detection area, the detection result is a vacuuming action and there is no snipping action, it is determined that the detection result is a vacuuming action, and the marking module 130 marks the status of the battery module as OK; if when the target battery module leaves the target detection area, the detection result has a snipping action, it is determined that the detection result is a snipping action, and the marking module 130 marks the status of the battery module as NG. Among them, the snipping action means that the operator judges that if there are solder bumps generated by welding in the welding area, it is necessary to trim them with pliers. After the operator performs the snipping action, secondary vacuuming is also required to reduce the quality risk of uncleaned welding slag.

[0064] In some embodiments, the marking module 130 marks the status of the target battery module according to the action detection result, specifically for: in the case where the action detection result is a vacuuming action, marking the status of the target battery module as the first status; in the case where the action detection result is a snipping action, marking the status of the target battery module as the second status.

[0065] Specifically, after the battery module enters the detection area, a label is marked for each target battery module, and action detection is performed on each target battery module. If it is detected that the operator performs a vacuuming operation on the target battery module, the marking module 130 determines that the first status is OK; if it is detected that the operator performs a snipping action on the target battery module, the marking module 130 determines that the second status is NG. Among them, the status of the target battery module is generally determined according to the last action of the target battery module leaving the target detection area; for components in the detection area that are not target battery modules, the status of the components is marked as OK, and when in the detection area, the status of this component is not changed, improving the detection efficiency. By marking the status of each target battery module in the detection area, it is convenient to subsequently determine whether it is necessary to save the video data according to the status of the target battery module, so as to view this section of the video later and assist in management.

[0066] In some embodiments, the dust removal detection device for the battery module further includes: a storage module, configured to store video data and push a reminder message if there is a target battery module in the second state when it is determined that the target battery module leaves the target detection area; a deletion module, configured to delete the video data if there is no target battery module in the second state.

[0067] Specifically, when the detection module 120 detects that the target battery module leaves the target detection area, it detects the state of each target battery module. If the states of all target battery modules are dust suction operations (the first state), it is determined that the dust suction operations on the battery modules in the acquired video data are all in accordance with the specifications, and there is no need to store the current video data. That is, the video data is deleted by the deletion module to release the memory. If there is a target battery module in the state of pliers operation (the second state), it is determined that there is a situation where the dust suction operation on the battery module in the acquired video data does not conform to the specifications. The video data is stored by the storage module and a reminder message is pushed to facilitate viewing the video data for auxiliary management. Moreover, the function of asynchronous video storage is added, so that the video data stored during detection does not affect the storage of NG video data, improving the output rate of NG video data and the real-time performance of video stream reading.

[0068] Exemplarily, the current video data includes two target battery modules that need dust suction, denoted as target battery module A and target battery module B. The action detection is performed on target battery module A and target battery module B respectively. Among them, target battery module A has gone through a dust suction operation in the target detection area, and target battery module B has gone through a dust suction operation and a pliers operation in the target detection area. At this time, there is a pliers operation in the detection result. Target battery module A is marked as OK, and target battery module B is marked as NG, and the current video data is stored. Exemplarily, the current video data includes one target battery module that needs dust suction, denoted as target battery module C. The action detection is performed on target battery module C. When target battery module C has gone through a dust suction operation, a pliers operation, and a second dust suction operation in the target detection area, the detection result at this time is a dust suction operation. Target battery module C is marked as OK, and the current video is deleted to release the memory.

[0069] It should be noted that in the above embodiments, the action detection results of the target battery module are all determined by the last action of the target battery module when it leaves the target detection area. For example, if the last action is a dust suction operation, the action detection result is a dust suction operation; if the last action is a pliers operation, the action detection result is a pliers operation.

[0070] In some embodiments, the target detection area includes a starting area, a detection area, and a leaving area. The target battery module is marked in the starting area, and when the number of frames containing the target battery module in the detection area exceeds a first preset number of frames, it is determined that the target battery module enters the target detection area, and a detection action is executed. The first preset number of frames can be calibrated according to the actual situation. For example, the first preset number of frames can be 5 frames.

[0071] Specifically, the target detection area can be divided into three sub-areas in the order of the starting area, the detection area, and the leaving area. That is, the battery module first reaches the starting area, queues in the starting area to enter the detection area, and the operator completes the dust suction action and the cutting pliers action on the target battery module in the detection area, and then enters the leaving area to enter the next process. When the battery module enters the starting area, the battery module is identified as the target battery module and the target battery module is marked. When the target battery module enters the detection area, it is judged whether the number of frames of the target battery module in the detection area exceeds 5 frames. If it exceeds, the action detection is started; if it does not exceed, the action detection is not performed, so that the action detection can be performed when it is determined that the battery module to be detected enters the starting area, preventing misdetection.

[0072] In some embodiments, the detection module 120 is further configured to determine that the target battery module leaves the target detection area when one of the following conditions is met: the target battery module enters the leaving area and the number of frames containing the target battery module exceeds a second preset number of frames; the time when the target battery module enters the target detection area exceeds a first preset time; the target battery module is not detected in the detection area and lasts for a second preset time. The second preset number of frames, the first preset time, and the second preset time can be calibrated according to the actual situation. For example, the second preset number of frames can be 5 frames, the second preset time can be 2 min, and the first preset time can be 30 min.

[0073] That is to say, when the detection area and the leaving area are relatively close, it is easy to have instability in the detection of the size of the module target due to the addition and increase in the amplitude of human actions, causing part or all of the target battery module in the detection area to enter the leaving area. Therefore, when the number of frames in which the target battery module enters the leaving area exceeds 5 frames, it is considered that the target battery module leaves the target detection area. Or, when the operator leaves the work station without protecting the target battery module, the video data is continuously saved, resulting in excessive memory occupation by the video. Therefore, when the time for the target battery module to enter the target detection area exceeds the first preset time, it is considered that the target battery module leaves the target detection area. Or, when the operator leaves the work station and protects the target battery module, and the target battery module is not detected for a long time, it is considered that the target battery module leaves the target detection area. Thus, it is possible to accurately determine whether the battery module leaves the target detection area, reduce the probability of misoperation, and avoid the continuous saving of the video when the operator leaves without protection, resulting in excessive memory occupation by the video, or when the operator leaves and protects the battery module, and the module is not detected for a long time, affecting the subsequent detection process.

[0074] In some embodiments, the detection module 120 determines the inclusion of the target battery module according to the video data, specifically for: processing the video data to obtain image data; classifying the image data using a preset image classification model; and determining the inclusion of the target battery module when it is determined that there is a battery module according to the classification result.

[0075] Specifically, the video data is processed in the target detection area to obtain image data, and the image data is input into a pre-trained classification model. This classification model can divide the image data into two categories. One category is the presence of a battery module (including the target battery module), and the other category is the absence of a battery module (including an empty carrier, a detection table base, and a module that does not require dust suction, such as a battery module sample). When the classification result is the category of the presence of a battery module, it is determined that the video data includes the battery module that needs dust suction, that is, the target battery module. Thus, the components in the detection area can be divided into two categories. For the empty carrier, the detection table base, and the battery module sample, no action detection is required, improving the detection efficiency.

[0076] It should be noted that since the color of the battery module sample is different from the appearance of the welded battery module (including color, labels pasted on the module, or tapes pasted on special positions), and the battery module sample does not contain a bar piece, it is also possible to determine whether the video data includes the target battery module according to the color of the battery module, the presence or absence of a bar piece, etc.

[0077] In some embodiments, before performing motion detection on the area where the target battery module is located, the above-mentioned device further includes: a filtering module, configured to predict the area where the battery module is located based on a target detection model to obtain a prediction result; and set a plurality of different confidence prediction boxes based on non-maximum suppression to filter the prediction result.

[0078] Specifically, the target detection model is obtained by annotating and training the targets of actions such as pliers, dust suction, tapping the bar piece, and screw scribing of the module through the labelme annotation software. For example, the target detection YOLO-v5 is used, and the non_max_suppression part is modified to support different confidence filtering for multiple targets. After entering the target detection area, the video data is screened, different confidences are set, the video data of the code scanning action and the tapping bar piece action are filtered out, and the videos of the dust suction action and the pliers action are retained, and then motion detection is performed. For the dust suction action, the pliers action, the code scanning action, and the tapping bar piece action, due to reasons such as the tool being blocked, similar gestures, different habits of each operator using the same tool, and tool differences, a plurality of different confidences are set to filter out the code scanning action and the tapping bar piece action similar to the dust suction action or the pliers action, etc., to improve the detection efficiency and detection accuracy.

[0079] It should be noted that for the details not disclosed in the dust removal detection device of the battery module in the embodiments of the present application, please refer to the details disclosed in the dust removal detection method of the battery module in the embodiments of the present application, and will not be elaborated here specifically.

[0080] Corresponding to the above embodiments, the present application also proposes a computer-readable storage medium.

[0081] The computer-readable storage medium of the present application stores a dust removal detection program for a battery module, and when the dust removal detection program for the battery module is executed by a processor, the above-mentioned dust removal detection method for the battery module is implemented.

[0082] Corresponding to the above embodiments, the present application also proposes a terminal device.

[0083] As Figure 5 shown, the terminal device 200 of the present application includes a memory 210, a processor 220, and a dust removal detection program for a battery module stored in the memory 210 and executable on the processor 220. When the processor executes the dust removal detection program for the battery module, the above-mentioned dust removal detection method for the battery module is implemented.

[0084] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features; and these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the present application, and they should all be covered within the scope of the claims and the description of the present application. In particular, as long as there is no structural conflict, the various technical features mentioned in each embodiment can be combined in any way. The present application is not limited to the specific embodiments disclosed in the text, but includes all technical solutions that fall within the scope of the claims.

Claims

1. A dust removal detection method for a battery module, characterized in that, the method includes: Obtaining video data entering the target detection area; When it is determined according to the video data that at least one target battery module is included, performing motion detection on the area where the target battery module is located respectively; Marking the status of the target battery module according to the motion detection result.

2. The method according to claim 1, characterized in that, marking the status of the target battery module according to the motion detection result includes: When the motion detection result is a dust suction motion, marking the status of the target battery module as the first status; When the motion detection result is a cutting pliers motion, marking the status of the target battery module as the second status.

3. The method according to claim 2, characterized in that, further includes: When it is determined that the target battery module leaves the target detection area, if there is a target battery module with the status of the second status, saving the video data and pushing a reminder message; If there is no target battery module with the status of the second status, deleting the video data.

4. The method according to claim 3, characterized in that, The target detection area includes a start area, a detection area and a departure area. Mark the target battery module in the start area, and when the number of frames including the target battery module in the detection area exceeds the first preset number of frames, determine that it enters the target detection area and perform detection actions.

5. The method according to claim 4, characterized in that, When one of the following conditions is met, it is determined that the target battery module leaves the target detection area: The target battery module enters the departure area and the number of frames including the target battery module exceeds the second preset number of frames; The time when the target battery module enters the target detection area exceeds the first preset time; The target battery module is not detected in the detection area and lasts for the second preset time.

6. The method according to any one of claims 1-5, characterized in that, determining that the target battery module is included according to the video data includes: Processing the video data to obtain image data; Classifying the image data by using a preset image classification model; When it is determined that there is a battery module according to the classification result, determining that the target battery module is included.

7. The method according to claim 1, characterized in that, Before performing motion detection on the area where the target battery module is located, the method further includes: Predicting the area where the battery module is located based on a target detection model to obtain a prediction result; Based on non-maximum suppression, setting multiple different confidence prediction frames to filter the prediction result.

8. A dust removal detection device for a battery module, characterized in that, includes: An acquisition module for acquiring video data entering the target detection area; A detection module for performing motion detection on the area where the target battery module is located respectively when it is determined according to the video data that at least one target battery module is included; A marking module, configured to mark the state of the target battery module according to the action detection result.

9. A computer-readable storage medium, characterized in that a dust removal detection program for a battery module is stored thereon, and when the dust removal detection program for the battery module is executed by a processor, the dust removal detection method for the battery module according to any one of claims 1-7 is implemented.

10. A terminal device, characterized in that it includes a memory, a processor, and a dust removal detection program for a battery module stored on the memory and executable on the processor. When the processor executes the dust removal detection program for the battery module, the dust removal detection method for the battery module according to any one of claims 1-7 is implemented.

Images