CN115156093A

CN115156093A - Battery shell defect detection method, system and device

Info

Publication number: CN115156093A
Application number: CN202210764899.9A
Authority: CN
Inventors: 杨凯; 林成龙; 崔磊
Original assignee: Shanghai Sensetime Intelligent Technology Co Ltd
Current assignee: Shanghai Sensetime Intelligent Technology Co Ltd
Priority date: 2022-06-29
Filing date: 2022-06-29
Publication date: 2022-10-11

Abstract

The embodiment of the application discloses a battery shell defect detection method, a system and a device, wherein the method is applied to a battery shell defect detection system, the battery shell defect detection system comprises an image acquisition module and an image processing module, and the method comprises the following steps: the image acquisition module acquires a battery image to be detected; the image processing module extracts a welding surface in the collected battery image to be detected to obtain a welding surface area of the battery; the image processing module determines a defect detection result of the battery shell based on the welding surface area of the battery.

Description

Battery shell defect detection method, system and device

Technical Field

The present application relates to the field of image detection technologies, and in particular, to a method, a system, and an apparatus for detecting defects of a battery case.

Background

The housing of a battery, such as a lithium battery, typically has some surface defects due to the manufacturing process and assembly. In the production line production process, the defects of the battery shell are mainly detected through helium detection, namely, the tightness of the battery shell is detected by filling helium into the battery shell, and the battery shell with the defects of holes, sand holes and the like can be detected through the helium detection.

Disclosure of Invention

In view of the above, the present disclosure provides a method, a system and a device for detecting defects of a battery case.

The technical scheme of the embodiment of the application is realized as follows:

in a first aspect, an embodiment of the present application provides a battery case defect detection method, which is applied to a battery case defect detection system, where the battery case defect detection system includes an image acquisition module and an image processing module, and the method includes: the image acquisition module acquires a battery image to be detected; the image processing module extracts a welding surface in the acquired battery image to be detected to obtain a welding surface area of the battery; the image processing module determines a defect detection result of the battery case based on the welding surface area of the battery.

In some embodiments, the battery case defect detection system further comprises a sensor and a controller, the method further comprising: the sensor detects whether a battery to be detected is located in a detection area; under the condition that the battery to be detected is located in the detection area, the sensor sends a first signal to the controller; the controller responds to the first signal and controls the image processing module to collect the battery image to be detected.

Therefore, the battery to be detected can be subjected to image acquisition only under the condition that the battery to be detected is located in the area to be detected, so that the number of acquired images is reduced, and the calculation amount of subsequent image processing is reduced.

In some embodiments, the battery casing defect detecting system further includes a robot arm, and after determining the defect detection result of the battery casing, the robot arm further includes: under the condition that the defect detection result represents that no defect exists, the image processing module converts the defect detection result into a second signal; the image processing module sends the second signal to the controller; and the controller responds to the second signal and controls the mechanical arm to place the battery to be detected in a good product area.

Therefore, under the condition that the defect detection result represents that no defect exists, the defect detection result is converted into a second signal, the mechanical arm is controlled to place the battery to be detected in a good product area, and the battery to be detected is distinguished.

In some embodiments, the battery casing defect detecting system further includes a robot arm, and after determining the defect detection result of the battery casing, the robot arm further includes: under the condition that the defect detection result represents that the defect exists, the image processing module converts the defect detection result into a third signal; the image processing module sends the third signal to the controller; and the controller responds to the third signal and controls the mechanical arm to place the battery to be detected in the defective product area.

Therefore, under the condition that the defect detection result represents that the defect exists, the defect detection result is converted into the third signal, the mechanical arm is controlled to place the battery to be detected in the defective product area, the battery to be detected is distinguished, and the defect reason of the battery with the defect can be conveniently analyzed by a production line engineer.

In some embodiments, the battery images to be detected include at least two batteries, and after determining the defect detection result of each battery case, the method further includes: the image processing module acquires the position information of each battery shell; converting the defect detection results of the at least two battery cases and the position information of each battery case into a set of identification signals, wherein the identification signals comprise the second signal or the third signal; sending the set of identification signals to the controller; the controller controls the robotic arm to grasp a battery of the at least two batteries in response to the set of identification signals; based on the group of identification signals, the controller controls the mechanical arm to place the battery in the at least two batteries in the defective product area or the good product area.

In this way, under the condition that the battery image to be detected comprises at least two batteries, the position information of each battery shell is obtained; then converting the defect detection results of at least two battery shells and the position information of each battery shell into a group of identification signals, and sending the identification signals to a controller; and then the controller responds to a group of identification signals and controls the mechanical arm to place the batteries in the at least two batteries in a defective product area or a good product area, so that the good products and the defective products are distinguished.

In some embodiments, the image processing module acquires position information of each battery case, including: the image processing module detects the welding surface of each battery in the battery image to be detected to obtain a detection frame of the welding surface of each battery; and determining the position information of each battery shell based on the position information of the detection frame of the welding surface of each battery in the battery image to be detected.

Like this, through the detection frame that acquires the face of weld of each battery, obtain the positional information of each battery case to conveniently fix a position the battery, and then can make the arm snatch target battery according to the position and distinguish to good products district or defective products district.

In some embodiments, the battery casing defect detecting system further includes a workbench and a driving component, wherein the workbench includes at least two regions to be placed, each region to be placed is used for placing the battery to be detected, and the image collecting module collects an image of the battery to be detected, and includes: the controller responds to a fourth signal, controls the driving component to drive the workbench to rotate, and places a target area to be placed in the area to be detected; the image acquisition module acquires a battery image to be detected, which is located in the detection area.

Like this, under the condition that battery case defect detecting system includes operation panel and driver part, it is rotatory to control driver part drive operation panel through the controller, treats the target and puts the district and arrange in treating the detection zone for the battery image that waits that the image acquisition module collection is located to treat the detection zone treats treating, and the district is treated with the change to the automatic rotation that realizes the operation panel.

In some embodiments, the extracting, by using the image processing module, the welding surface in the collected battery image to be detected to obtain the welding surface area of the battery includes: the image processing module carries out image detection on the collected battery image to be detected by using a target detection algorithm to obtain a welding surface area of the battery;

the determining, by the image processing module, a defect detection result of the battery case based on the welding surface area of the battery includes: and the image processing module performs image detection on the welding surface area of the battery to obtain a defect detection result of the battery shell.

Therefore, the welding surface in the battery image to be detected can be quickly extracted and the defects in the welding surface can be accurately detected, so that the defect detection method of the battery shell can meet the beat requirement and the detection requirement of a production line.

In some embodiments, the welding location of the welding surface includes at least one of: the positive and negative terminal outer edges, the liquid injection port outer edge, the vent hole outer edge and the top plate outer edge; the type of defect includes at least one of: the image processing module carries out image detection on the welding surface area of the battery to obtain the defect detection result of the battery shell, and the defect detection result comprises the following steps: and the image processing module performs image detection on at least one region of the outer edges of the positive and negative terminals, the outer edge of the liquid injection port, the outer edge of the vent hole and the outer edge of the top plate of the battery to obtain a defect detection result of the battery shell, wherein the defect detection result comprises at least one defect type of a cavity, a sand hole, an uneven edge, a gap, a bulge and a damage.

In this way, through image detection of the welding position of the welding surface, a defect detection result of the battery case including a common defect type can be obtained.

In a second aspect, an embodiment of the present application provides a battery case defect detection system, which includes: the image acquisition module is positioned above the battery to be detected and used for acquiring the battery image to be detected; the image processing module is used for extracting the welding surface in the collected battery image to be detected to obtain the welding surface area of the battery; determining a defect detection result of the battery case based on the welding surface area of the battery.

In a third aspect, an embodiment of the present application provides a battery case defect detection apparatus, where the apparatus includes: the acquisition module is used for acquiring a battery image to be detected by using the image acquisition module; the extraction module is used for extracting the welding surface in the collected battery image to be detected by using the image processing module to obtain the welding surface area of the battery; and the determining module is used for determining the defect detection result of the battery shell based on the welding surface area of the battery by utilizing the image processing module.

The embodiment of the application provides a battery shell defect detection method based on deep learning and a computer vision scheme. Therefore, the speed and the accuracy of quality inspection can be improved, and the labor cost can be reduced.

It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the technical aspects of the application.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the application.

Fig. 1A is a schematic structural diagram of a battery case defect detection system according to an embodiment of the present disclosure;

fig. 1B is a schematic flow chart illustrating an implementation of a method for detecting defects of a battery case according to an embodiment of the present disclosure;

fig. 2A is a schematic structural diagram of another battery case defect detection system according to an embodiment of the present disclosure;

fig. 2B is a schematic implementation flowchart of another method for detecting defects of a battery case according to an embodiment of the present disclosure;

fig. 3A is a schematic structural diagram of another battery case defect detection system according to an embodiment of the present disclosure;

fig. 3B and fig. 3C are schematic diagrams illustrating an implementation flow of another method for detecting defects of a battery case according to an embodiment of the present application;

fig. 4A is a schematic structural diagram of another battery case defect detection system according to an embodiment of the present disclosure;

fig. 4B is a schematic view of an implementation process for acquiring a battery image to be detected according to an embodiment of the present disclosure;

fig. 5A is a schematic structural diagram of another battery case defect detection system according to an embodiment of the present disclosure;

fig. 5B is a schematic structural diagram of a disposition of a light source and a camera provided in an embodiment of the present application;

fig. 6 is a schematic structural diagram of a battery case defect detection apparatus according to an embodiment of the present disclosure.

Detailed Description

In order to make the purpose, technical solutions and advantages of the present application clearer, the technical solutions of the present application are further described in detail with reference to the drawings and the embodiments, the described embodiments should not be considered as limiting the present application, and all other embodiments obtained by a person of ordinary skill in the art without creative efforts belong to the protection scope of the present application.

In the following description, reference is made to "some embodiments" which describe a subset of all possible embodiments, but it is understood that "some embodiments" may be the same subset or different subsets of all possible embodiments, and may be combined with each other without conflict.

Reference to the terms "first/second/third" merely distinguishes similar objects and does not denote a particular ordering for the objects, and it is to be understood that "first/second/third" may, where permissible, be interchanged in a particular order or sequence so that embodiments of the present application described herein can be practiced otherwise than as specifically illustrated or described herein.

Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the present application.

Before further detailed description of the embodiments of the present application, terms and expressions referred to in the embodiments of the present application will be described, and the terms and expressions referred to in the embodiments of the present application will be used for the following explanation.

Artificial Intelligence (AI) is a theory, method, technique and application system that uses a digital computer or a machine controlled by a digital computer to simulate, extend and expand human Intelligence, perceive the environment, acquire knowledge and use the knowledge to obtain the best results. In other words, artificial intelligence is a comprehensive technique of computer science that attempts to understand the essence of intelligence and produce a new intelligent machine that can react in a manner similar to human intelligence. Artificial intelligence is the research of the design principle and the realization method of various intelligent machines, so that the machines have the functions of perception, reasoning and decision making.

The artificial intelligence technology is a comprehensive subject and relates to the field of extensive technology, namely the technology of a hardware level and the technology of a software level. The artificial intelligence infrastructure generally includes technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technologies, operation/interaction systems, mechatronics, and the like. The artificial intelligence software technology mainly comprises a computer vision technology, a voice processing technology, a natural language processing technology, machine learning/deep learning and the like.

Computer Vision technology (CV) is a science for researching how to make a machine "see", and further refers to that a camera and a Computer are used to replace human eyes to perform machine Vision such as identification, tracking and measurement on a target, and further image processing is performed, so that the Computer processing becomes an image more suitable for human eyes to observe or is transmitted to an instrument to detect. As a scientific discipline, computer vision research-related theories and techniques attempt to build artificial intelligence systems that can capture information from images or multidimensional data. Computer vision technologies generally include image processing, image Recognition, image semantic understanding, image retrieval, optical Character Recognition (OCR), video processing, video semantic understanding, video content/behavior Recognition, three-Dimensional object reconstruction, three-Dimensional graphics (3D) technology, virtual reality, augmented reality, synchronous positioning, map construction, and other technologies, and also include common biometric technologies such as face Recognition and fingerprint Recognition.

The target detection algorithm is intended to detect whether a target object exists in an image or a video and position information of the target object, and is generally output in the form of a detection box.

In brief, image semantic division refers to division performed by a computer according to image semantics, and in an image field, semantics refers to content of an image, and division means that different objects in a picture are divided from the perspective of pixels to identify the pixels in an original image.

Image instance segmentation, namely the combination of target detection and semantic segmentation, wherein the instance segmentation can be accurate to the edge of an object relative to a bounding box of the target detection; with respect to semantic segmentation, instance segmentation can label different individuals of the same class of objects on the graph.

Neural Architecture Search (NAS) is a technology for automatically designing a Neural network, and a high-performance network structure can be automatically designed according to a sample set through an algorithm.

Machine Learning (ML) is a multi-domain cross subject, and relates to multi-domain subjects such as probability theory, statistics, approximation theory, convex analysis, algorithm complexity theory and the like. The method specially studies how a computer simulates or realizes the learning behavior of human beings so as to acquire new knowledge or skills and reorganize the existing knowledge structure to continuously improve the performance of the computer. Machine learning is the core of artificial intelligence, is the fundamental approach to make computers have intelligence, and is applied in various fields of artificial intelligence. Machine learning and deep learning generally include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and teaching learning.

With the research and progress of artificial intelligence technology, the artificial intelligence technology is developed and applied in a plurality of fields, such as common smart homes, smart wearable devices, virtual assistants, smart speakers, smart marketing, unmanned driving, automatic driving, unmanned aerial vehicles, robots, smart medical care, smart customer service, and the like.

The scheme provided by the embodiment of the invention relates to the technologies of machine learning, computer vision technology and the like of artificial intelligence, and is specifically explained by the following embodiments:

the embodiment of the application provides a battery shell defect detection method, which is applied to a battery shell defect detection system 100 shown in fig. 1A, and the system includes an image acquisition module 101 and an image processing module 102. Fig. 1B is a schematic view of an implementation flow of a battery case defect detection method provided in an embodiment of the present application, and as shown in fig. 1B, the method includes steps S101 to S103:

s101, the image acquisition module acquires a battery image to be detected;

here, the image capturing module may be a terminal device such as a camera or a video camera capable of capturing images, for example, a monocular camera, a linear scanning camera, or the like, and the type of the image capturing module is not limited in the embodiments of the present application. In some embodiments, the image capturing module may be located above the battery to be detected, and is configured to capture an image of the battery to be detected. In some embodiments, the image capturing module can be in a height-adjustable form, the height of the image capturing module can be adjusted higher for a battery with a larger model, and the height of the image capturing module can be adjusted lower for a battery with a smaller model, so that the captured image of the battery to be detected comprises a part of the battery shell which needs to be detected. When the device is implemented, the image acquisition module can be arranged on a bracket with a sliding rail, and the height of the image acquisition module can be continuously adjusted by changing the position of the image acquisition module on the sliding rail; the image acquisition module can be arranged on the support with different height gears, and the non-continuous adjustment of the height of the image acquisition module is realized by changing the gears of the image acquisition module on the support. The embodiment of the application does not limit the mode of realizing the height adjustable of the image acquisition module.

The battery may include, but is not limited to, a lithium battery, a sodium battery, a hydrogen battery, and the like.

The battery image to be detected may include a weld face region of the battery, and the battery image to be detected may be: a Two-dimensional (2D) image or a Three-dimensional (3D) image, wherein the 2D image may include: red Green Blue (RGB) images, depth images, etc. collected by a monocular or monocular camera, the 3D images may include: a linear scanning camera acquires a 3D image.

In some embodiments, the acquired battery image to be detected may include one battery, two batteries, or multiple batteries. Under the condition that the battery image to be detected comprises at least two batteries, the subsequent defect detection of the battery shell can be carried out on the at least two batteries at one time, and the detection efficiency is improved.

In some embodiments, the implementation of step S101 may set the period in which the image acquisition module acquires the image according to the time period in which the battery to be detected is located in the detection area, for example, every 5 seconds the battery to be detected is transmitted to the detection area through the transmission belt, and then the period in which the image acquisition module acquires the image may be set to 5 seconds, so as to reduce the frequency of acquiring the image by the image acquisition module and reduce the amount of calculation of the post-image processing. Wherein, treat that the detection zone can gather the battery that waits to detect the battery image for the image acquisition module and place the region. For example, the detection area to be detected can be the area under the image acquisition module, also can be the area in other shooting ranges of the image acquisition module, and the detection area is treated not to be restricted in this application embodiment.

In some embodiments, the battery case defect detection system may further include a light source, which may be located above the battery to be detected and used for polishing the battery to be detected when the image acquisition module performs image acquisition on the battery to be detected, so as to improve the resolution of the battery image to be detected and facilitate detection of subsequent battery case defects. In some embodiments, the light source may be a point light source, a line light source, a coaxial light source, and the like, and the type of the light source is not limited in the embodiments of the present application. Under the condition that the light source is coaxial light under a dark field, the welding seam in the welding surface can be more obvious and clear, and the accuracy of detecting the defects of the battery shell is further improved. In some embodiments, the light source can also be in a height-adjustable form, the height of the light source can be adjusted higher for a battery with a large model, and the height of the light source can be adjusted lower for a battery with a small model, so that the collected battery image to be detected is clearer, wherein the light source can be adjusted in height by referring to the image collection module to realize the height-adjustable form.

In some embodiments, the implementation of step S101 may also set a period for turning on the light source according to a time period for the battery to be detected to be located in the to-be-detected region, so that when the image acquisition module performs image acquisition on the battery to be detected, the battery to be detected is polished. In some embodiments, the battery case defect detection system may further include a controller, and the controller controls the turn-on time of the light source and the image acquisition module.

In some embodiments, the implementation of step S101 may further set the image capturing module and the light source to be normally open during the process of starting the production line to detect the defect of the battery case, so as to capture the battery image to be detected.

In some embodiments, after the image capturing module collects the battery image to be detected in step S101, the noise reduction processing may be performed on the battery image to be detected to obtain a battery image to be detected after noise reduction, which includes: and carrying out bilateral filtering processing and Gaussian filtering processing on the battery image to be detected to obtain the battery image to be detected after noise reduction. For example, bilateral filtering processing and gaussian filtering processing may be sequentially performed on a battery image to be detected to obtain a battery image to be detected after noise reduction; for another example, the battery image to be detected may be subjected to gaussian filtering processing and bilateral filtering processing in sequence to obtain the battery image to be detected after noise reduction. The filter parameters of the bilateral filter and the Gaussian filter, such as the shape and the size of the filter kernel, can be adjusted according to the appearance of the battery to be detected, shooting noise and the like. The bilateral filtering processing and the Gaussian filtering processing are carried out on the battery image to be detected, so that the imaging noise and the interference of complex fine textures in the battery image to be detected can be removed, and the accuracy of subsequent defect detection is improved.

In some embodiments, the contrast of the battery image to be detected may also be enhanced to obtain the battery image to be detected with enhanced contrast, for example, the contrast of the battery image to be detected may be enhanced by adopting linear transformation, piecewise linear transformation, nonlinear transformation, histogram equalization, and the like. For example, the range of the gray scale value of the battery image to be detected is (40, 150), and the gray scale value of the battery image to be detected may be mapped to the interval (0, 255) to enhance the contrast of the battery image to be detected. By enhancing the contrast of the battery image to be detected, the gray scale difference of the battery image to be detected can be increased, so that the follow-up neural network can be facilitated to capture the defect characteristics in the battery image to be detected.

Step S102, the image processing module extracts a welding surface in the collected battery image to be detected to obtain a welding surface area of the battery;

here, the image Processing module may be a processor having an image Processing function, wherein the processor may be an image processor (GPU) or a Central Processing Unit (CPU). The image acquisition module and the image processing module can be connected in a wired mode such as a transmission line and can also be connected in a wireless mode such as a Bluetooth mode or a wireless network mode.

Since the sealability of the battery is critical to the life and safety performance of the battery, the sealability of the finished battery is guaranteed to meet standards. Factors influencing the sealing performance mostly come from poor welding between components on the battery shell and cracks caused by bulges of the battery shell, wherein the welding positions between the battery shell components at least comprise one of the following: the types of the defects at least comprise one of the following types: voids, blisters, edge misalignment, gaps, bulges, and breakage, which can be observed through the weld plane (i.e., the plane of the top plate). In addition, because the acquired battery image to be detected has low resolution and the background image of other non-welding surfaces can interfere the welding surface, so that the accuracy of defect detection is reduced, the welding surface in the battery image can be extracted through the image processing module in the step S102, so that the defect detection of the welding surface in the subsequent step is facilitated, and the defect detection result of the battery shell is obtained.

In some embodiments, the implementation of step S102 may detect the welding surface in the battery image by the image processing module using the network model of the target detection algorithm, and obtain the welding surface area of the battery, for example: retinaNet, YOLO, fast RCNN and the like, when the method is implemented, a pre-trained network model of a target detection algorithm can be adopted for image detection, and an output target detection frame is a welding surface area; the image processing module may also perform image segmentation on the welding surface in the battery image based on an image segmentation algorithm (e.g., semantic segmentation, instance segmentation, etc.) of the neural network to obtain a welding surface region of the battery, for example: mask RCNN, PSPNet, deep Labv3+, etc., when implementing, can adopt the network model of the image segmentation algorithm trained in advance to carry on the image segmentation, the connected domain outputted is the welding face area.

In some embodiments, the light-weight model may be obtained by pruning the network model of the target detection algorithm, i.e., pre-training the network model using the training image set, determining unimportant subnetworks, and then pruning the unimportant subnetworks. For example, in the pre-training stage, a neural structure search technique is used to randomly sample the sub-networks in the search space before each iteration, and the sampled sub-networks are subjected to forward and backward propagation and gradient updating. After the pre-training is finished, a plurality of subnets meeting the conditions of Floating-point operation per second (FLOPS) are randomly sampled in a search space and tested, so that a pareto optimal subnet structure set is calculated, and finally, a proper subnet structure is selected from the pareto optimal subnet structure set to obtain a final light-weight network.

In some embodiments, the implementation of step S102 may further acquire a mask image of the battery welding surface through the image processing module, and extract the welding surface area in the battery image to be detected based on the mask image. In implementation, different mask images can be preset for different battery models. The mask image of each model can comprise a mask area and a non-mask area, the mask area is used for representing position information of a welding surface in a battery graph of the battery model, the non-mask area is used for representing position information of an area except the welding surface in the battery graph, a battery image to be detected is aligned with the mask image, an image area belonging to the mask area in the aligned battery image to be detected is extracted and used as the welding surface area, and therefore the welding surface area in the battery image to be detected is extracted quickly.

In some embodiments, since the welding surface of the battery is a large component and has an obvious single characteristic, in order to increase the detection speed, the image processing module performs image detection on the acquired battery image to be detected by using a single-stage object detection algorithm to obtain the welding surface area of the battery, such as RetinaNet, YOLO, SSD, and the like, so that the welding surface area can be better applied to the production line production process to improve the production efficiency.

The method for extracting the welding surface in the battery image is not limited in the embodiment of the application.

In some embodiments, after the implementation of step S102, the method may further include: and aligning the welding surface area of the battery through an image processing module to obtain an area to be processed.

In practice, a target image may be preset, wherein the target image may be an image of a battery welding surface area having no included angle with the target direction. Under the condition that the position of the welding surface area of the battery obtained by extraction is not correct, namely a certain included angle exists between the welding surface area of the battery and the target direction, the welding surface area of the battery and the target image can be aligned through the image processing module to obtain a to-be-processed area, and then the defect detection result of the battery shell is determined based on the to-be-processed area, so that the accuracy of subsequent defect detection is improved.

In the case where the battery image to be detected includes at least two batteries, the implementation of step S102 includes: and the image processing module extracts the welding surface of each battery in the acquired battery image to be detected to obtain the welding surface area of each battery.

And step S103, the image processing module determines a defect detection result of the battery shell based on the welding surface area of the battery.

Here, the defect detection result represents the existence or non-existence of a defect, and when the defect detection result represents the existence of a defect, the defect detection result may further include the type of the defect, such as a void, a sand hole, an uneven edge, a gap, a bulge, a breakage, and the like; in some embodiments, in the case that the defect detection result indicates that there is a defect, the defect detection result may not include the type of the defect, and only the defect is output.

In some embodiments, the implementation of step S103 may perform image detection on the welding surface area of the battery through the image processing module based on an object detection algorithm of a neural network, so as to obtain a defect detection result of the battery case, for example: during implementation, a first network model of a pre-trained target detection algorithm can be adopted for image detection, detection frames of each defect type are output, and then a defect detection result of the battery shell is obtained by judging the confidence coefficient of each detection frame. Under the condition that the confidence coefficient is greater than a first preset threshold value, a defect detection result representing the detection frame is that a defect exists and the detection frame is a corresponding defect type; and when the confidence coefficient is less than or equal to a first preset threshold value, the defect detection result representing the detection frame is that no defect exists.

In some embodiments, the welding location of the welding face includes at least one of: the positive and negative terminal outer edges, the liquid injection port outer edge, the vent hole outer edge and the top plate outer edge; the type of defect includes at least one of: voids, blisters, misalignment of edges, gaps, bulges, and damage. Then, the implementation of step S103 may perform image detection on at least one area of the outer edges of the positive and negative terminals, the outer edge of the liquid injection port, the outer edge of the vent hole, and the outer edge of the top plate of the battery through an image processing module based on a target detection algorithm of a neural network, so as to obtain a defect detection result of the battery case, where the defect detection result includes a defect type of at least one of a cavity, a sand hole, an uneven edge, a gap, a bulge, and a breakage. In this way, through image detection of the welding position of the welding surface, a defect detection result of the battery case including a common defect type can be obtained.

In some embodiments, in the case where there are at least two detection frames, the defect detection result of the battery case is determined to be defective in response to the defect detection result of at least one detection frame being defective, and the defect detection result of the battery case is determined to be non-defective in response to the defect detection results of all the detection frames being non-defective.

In some embodiments, when the defect detection result of the battery case is that a defect exists, the position information of the defect may be further output, that is, the position information of the detection frame whose confidence is greater than the first preset threshold, for example, if the defect type corresponding to the detection frame is a sand hole, the output position information of the defect is the position information of the detection frame corresponding to the sand hole, where the position information may be composed of an upper left corner point coordinate and a lower right corner point coordinate of the detection frame, for example, (x 1, y1, x2, y 2), where, (x 1, y 1) is the upper left corner point coordinate, and (x 2, y 2) is the lower right corner point coordinate; the position information may also be composed of a center point coordinate of the detection frame and a width and a height of the detection frame, for example, (x, y, w, h), where (x, y) is the center point coordinate, w is the width of the detection frame, and h is the height of the detection frame, and the method for representing the position information is not limited in the embodiment of the present application.

In some embodiments, the first network model may be trained using a first set of training images, wherein the training images in the first set of training images include annotation data for defect regions. The label data of the defective area may be label data including position information of the defective area and a defect type. For example, the label data of the defect area may be represented by position information of a rectangular frame and a defect type corresponding to the rectangular frame. Of course, the defect area may have other shapes, and is not limited herein. In this example, the first network model is trained by employing a first training image set with labeling data for the defect region, thereby enabling the first network model to learn the ability to identify the type of defect in the weld region.

When the first network model is trained, the first network model may be trained by using positive and negative samples, where the positive sample may be a sample containing a certain type of defect, such as a void, and the negative sample may be a sample not containing a certain type of defect. Because the size and the angle of the same defect type are different, in order to improve the detection rate of the first network model to the defect types with different sizes and angles, when the first network model is trained, the positive and negative samples can be scaled by different proportions or rotated by different angles and then sent to the first network model for training.

In some embodiments, because the types and forms of the defects of the battery shell are more and the number of the samples is not uniform, when the first network model is trained, the detection rate of rare defect samples can be improved by using local loss, and then the detection of different types and forms of defects is realized.

In some embodiments, since the defect type, size, angle, shape, and other factors of the battery case are complex, in order to improve the recall rate and accuracy of the detection model, the step S103 may be implemented by using a multi-stage object detection algorithm, such as Fast RCNN, faster RCNN, cascade RCNN, and the like, to detect the welding surface region of the battery through the image processing module, so as to obtain the defect detection result of the battery case.

In some embodiments, the multiple stages may be two stages, where the first stage is used to detect the welding surface area of the battery to obtain a candidate frame, and the candidate frame is an area where a defect may exist, i.e. a potential defect area; and the second stage is used for detecting the candidate frame to obtain the defect detection result of the battery shell. Under the condition of adopting a two-stage target detection algorithm, the second stage is detection based on the candidate frame obtained in the first stage, so that the interference of the background in the welding surface area on the detection result is removed, and equivalently, the candidate frame is further accurately detected, so that the obtained defect detection result of the battery shell is more accurate. Therefore, for the condition that factors such as defect types and forms are complex, the detection result can be more accurate by adopting a multi-stage target detection algorithm.

In some embodiments, the step S103 may also be implemented by the image processing module performing image segmentation on the welding surface in the battery image based on an image segmentation algorithm (semantic segmentation, instance segmentation, etc.) of a neural network, so as to obtain a defect detection result of the battery case, for example: mask RCNN, PSPNet, deepLabv3+, etc. During implementation, a pre-trained second network model of an image segmentation algorithm can be adopted for image segmentation to obtain potential defect regions of the welding surface region, the potential defect regions correspond to different defect types, then a connected domain of the potential defect regions is determined, and finally the defect detection result of the battery shell is obtained by judging the size of the area of the connected domain (the area of the connected domain is the number of pixels in the connected domain). Determining that the defect detection result of the connected domain is the existence of a defect and the connected domain is the corresponding defect type under the condition that the area of the connected domain is larger than or equal to a second preset threshold; and determining that the defect detection result of the connected domain is that no defect exists when the area of the connected domain is smaller than a second preset threshold. The detection result is screened through the area of the connected domain, so that the influence of noise in the battery image to be detected on the defect detection is reduced, and the accuracy of the defect detection result is improved.

In some embodiments, the second network model may be trained using a second set of training images, wherein the training images in the second set of training images include annotation data for defect regions. The label data of the defective area may be label data including position information (pixel level) of the defective area and a defect type. For example, the labeling data of the defect area can be represented by the position information of the pixel and the defect type corresponding to the pixel. In this example, the second network model is trained by employing a second training image set with labeling data for the defect regions, thereby enabling the second network model to learn the ability to identify potential defect regions in the weld face region. Wherein the training images in the second set of training images may be the same or different from the training images in the first set of training images. Because the image segmentation algorithm can be accurate to the pixel level, the defect detection result can be more accurate.

In some embodiments, the second network model may be pruned to obtain a lightweight network model, and then the lightweight network model is trained by using the second training image set, so that defect detection of the battery case by using the lightweight network model is achieved, the defect detection speed is increased, and the method is better applied to a fast-paced production process of a production line.

In the case where the battery image to be detected includes at least two batteries, the implementation of step S103 includes: the image processing module determines a defect detection result for each battery case based on the weld face area of each battery.

The embodiment of the application provides a battery shell defect detection method based on deep learning and computer vision schemes. Therefore, the speed and the accuracy of quality inspection can be improved, and the labor cost can be reduced.

In some embodiments, as shown in fig. 2A, the battery case defect detecting system 100 may further include a sensor 202 and a controller 201, and correspondingly, the present embodiment also provides a battery case defect detecting method, as shown in fig. 2B, the method includes the following steps S201 to S206:

step S201, a sensor detects whether the battery to be detected is located in a detection area.

The sensor is a detection device which can sense whether a battery to be detected is positioned in a region to be detected, and can convert the sensed information into an electric signal or other information in a required form according to a certain rule for outputting, so as to meet the requirements of information transmission, processing, storage, display, recording, control and the like. In some embodiments, the sensor may be, for example, an infrared laser trigger, a light sensitive sensor, a pressure sensor, or the like. When the battery to be detected is located in the area to be detected, the sensing element in the sensor is triggered to change and is converted into an electric signal to be sent to the controller, and the controller makes an instruction based on the signal. Because the battery that awaits measuring is lieing in waiting to detect the district and not lieing in the condition of waiting to detect the district, the signal of telecommunication through the sensor conversion can be different, consequently, the controller can judge through the signal of telecommunication received whether the battery that awaits measuring is located waits to detect the district.

The sensor is taken as an infrared laser trigger for example, the infrared laser trigger generally includes an optical system, a sensing element and a conversion circuit, where the sensing element may be a thermosensitive element or a photosensitive element, for example, the sensing element is a thermosensitive element, because the battery to be detected is located in the detection area, infrared rays radiated by the battery to be detected irradiate on the thermosensitive element after passing through the optical system, which may cause a change in resistance of the thermosensitive element, and the infrared rays are converted into electrical signals through the conversion circuit and output to the controller, and compared with a case where the battery to be detected is not located in the detection area, the resistance of the thermosensitive element is different, therefore, in two cases where the battery to be detected is located in the detection area and not located in the detection area, two different electrical signals may be output, and the controller may determine whether the battery to be detected is located in the detection area according to the magnitude of the electrical signals.

Step S202, under the condition that the battery to be detected is located in the detection area, the sensor sends a first signal to the controller.

Here, the controller may be a central processing Unit, a Digital Signal Processor (DSP), a Microprocessor Unit (MPU), a Field-Programmable Gate Array (FPGA), a Programmable Logic Array (PLA), or the like. The controller and the image processing module are logical concepts, and the controller and the image processing module can be physically separated when the controller and the image processing module are realized, for example, the controller is a central processing unit, and the image processing module is a graphic processor; of course, the controller and the image processing module may be physically integrated, for example, the controller and the image processing module are implemented by using a central processing unit. The first signal is a signal which is triggered by the change of the sensing information of the sensing element in the sensor and sent to the controller after the battery to be detected is located in the area to be detected, and the first signal is different from a signal which is sent to the controller by the sensor when the battery to be detected is not located in the area to be detected. For example, the first signal may be a high level, and the signal sent by the sensor to the controller may be a low level when the battery to be detected is not located in the region to be detected; in some embodiments, the first signal may also be a high level, and the signal sent by the sensor to the controller is a low level when the battery to be detected is not located in the region to be detected.

Step S203, the controller responds to the first signal and controls the image processing module to collect the battery image to be detected.

Here, since the signals sent to the controller by the sensor are different between the battery to be detected and the battery not located in the area to be detected, for example, when the battery to be detected is located in the area to be detected, the first signal sent to the controller by the sensor is at a high level; when the battery that waits to detect is not located and waits to detect the zone time, the signal that the sensor sent for the controller is the low level, consequently, the controller can judge through the height of level whether the battery that waits to detect is located treats the detection zone to treat under the condition that the battery that waits to detect is located treats the detection zone in the affirmation, control the image processing module and open, realize treating the collection of the battery image that detects.

And step S204, the image acquisition module acquires a battery image to be detected.

S205, an image processing module extracts a welding surface in the collected battery image to be detected to obtain a welding surface area of the battery;

step S206, the image processing module determines the defect detection result of the battery shell based on the welding surface area of the battery

Here, the steps S204 to S206 correspond to the steps S101 to S103, respectively, and specific embodiments of the steps S101 to S103 may be referred to when the steps are performed.

In the embodiment of the application, whether the battery to be detected is located the detection area of treating through the sensor detection to under the condition that the battery to be detected is located the detection area of treating, send first signal to the controller in the affirmation, then the controller responds to first signal, controls the collection of image processing module and waits to detect the battery image. Therefore, the battery to be detected can be subjected to image acquisition only under the condition that the battery to be detected is located in the area to be detected, so that the number of acquired images is reduced, and the calculation amount of subsequent image processing is reduced.

In some embodiments, as shown in fig. 3A, the battery case defect detecting system 100 may further include a robot arm 301, and correspondingly, as shown in fig. 3B, after step S103, further include:

step S104a: under the condition that the defect detection result represents that no defect exists, the image processing module converts the defect detection result into a second signal;

here, the second signal is a signal indicating that the battery to be detected has no defect, and the form of the second signal may be preset in implementation. For example, "1" may be preset to represent that the battery to be detected has a defect, and "0" represents that the battery to be detected has no defect, then the second signal may be "0"; the "OK" may also be preset to indicate that the battery to be detected has a defect, and the "NG" indicates that the battery to be detected does not have a defect, so that the second signal may be "NG", and the form of the second signal is not limited in the embodiment of the present application.

Step S105a: the image processing module sends the second signal to the controller;

here, the image processing module and the controller may be connected in a wired manner such as a transmission line, or may be connected in a wireless mode such as a bluetooth mode or a wireless network, so that the image processing module sends the second signal to the controller.

Step S106a: and the controller responds to the second signal and controls the mechanical arm to place the battery to be detected in a good product area.

The good product area is an area for placing batteries without defects, wherein if the batteries to be detected do not need to perform other procedures after the detection of the defects of the battery shell is finished, the good product area can be an area for specially placing the batteries without defects; if the battery to be detected needs to be subjected to other processes after the defect of the battery shell is detected, the good product area can refer to stations of other processes. The controller may cause the robot arm to place the battery to be detected in the good area by changing parameters of the robot arm, such as power, moving distance, and the like, based on the second signal.

In the embodiment of the application, under the condition that the defect detection result representation does not have the defect, the defect detection result is converted into the second signal, the mechanical arm is controlled to place the battery to be detected in the good product area, and the battery to be detected is distinguished.

In some embodiments, the battery case defect detecting system may further include a robot arm, and correspondingly, as shown in fig. 3C, after step S103, further include:

step S104b: under the condition that the defect detection result represents that the defect exists, the image processing module converts the defect detection result into a third signal;

here, the third signal is a signal indicating that the battery to be detected has a defect, and the form of the second signal may be preset in implementation. For example, a preset value of "1" may be used to represent that the battery to be detected has a defect, and a preset value of "0" represents that the battery to be detected has no defect, so that the third signal may be "1"; it may also be preset to "OK" to indicate that the battery to be detected has a defect, and "NG" to indicate that the battery to be detected has no defect, then the third signal may be "OK", and the form of the third signal is not limited in the embodiment of the present application.

Step S105b: the image processing module sends the third signal to the controller;

step S106b: and the controller responds to the third signal and controls the mechanical arm to place the battery to be detected in the defective product area.

Here, the defective area is an area where a defective battery is placed, and steps S105b and S106b can refer to steps S105a and S106a.

In the embodiment of the application, under the defect detection result characterization has the condition of defect, through turning into the third signal with the defect detection result, the control arm will wait to detect the battery and place in the defective products district, realize treating the differentiation of detecting the battery to make things convenient for production line engineer to carry out the analysis of bad reason to the battery that has the defect.

In some embodiments, the battery image to be detected may include at least two batteries, and after determining the defect detection result of each battery case in step S103, the method further includes: :

step S104: the image processing module acquires the position information of each battery shell;

here, the position information may be coordinate information of each battery case in the battery image to be detected, and may be, for example, a center point coordinate (x) of each battery case in the battery image to be detected _a ，y _a ) (ii) a For example, the position serial number of each battery case in the battery image to be detected is labeled according to a sequence from left to right, and the position serial number of a certain battery case in the battery image to be detected is the label of the battery in the battery image to be detected, for example, four batteries in total are used, and the label of the battery in the battery image to be detected may be "4".

In some embodiments, the implementation of step S104 may include:

step S1041: the image processing module detects the welding surface of each battery in the battery image to be detected to obtain a detection frame of the welding surface of each battery;

here, the step S1041 may be performed by using an object detection algorithm, for example, retinaNet, YOLO, fast RCNN, etc., to detect the welding surface of each battery, and the output of the detection result is a detection frame of the welding surface of each battery.

Step S1042: and determining the position information of each battery shell based on the position information of the detection frame of the welding surface of each battery in the battery image to be detected.

Here, the position information of the detection frame of the soldering surface of each battery in the image of the battery to be detected may be the coordinates (x) of the center point of the detection frame of the soldering surface of each battery _a ，y _a ) To show that, correspondingly, the center point coordinate (x) of each battery welding surface detection frame can be detected _a ，y _a ) As the position information corresponding to the battery case, the position information corresponding to the battery case is (x) _a ，y _a )。

In some embodiments, the step S1042 may also be implemented to determine a position number of each battery case in the battery image to be detected according to the center point coordinate of each battery welding surface detection frame, for example, if the batteries in the battery image to be detected are sequentially placed from left to right, the minimum x in the center point coordinate of the battery welding surface detection frame is the "1" battery, the next "2" battery, then the "3" battery, and finally the "4" battery.

This application embodiment, through the detection frame that acquires the face of weld of each battery, obtains the positional information of each battery case to the convenience is fixed a position the battery, and then can make the arm snatch target battery according to the position and distinguish to yields district or defective products district.

Step S105: converting the defect detection results of the at least two battery cases and the position information of each battery case into a set of identification signals, wherein the identification signals comprise the second signal or the third signal;

in the embodiment of the application, the position information of each battery shell is obtained by detecting the welding surface of each battery; and the defect detection result of each battery shell is obtained by identifying the defects of the welding surface area in the welding surface detection result, so that the corresponding relation exists between the position information and the defect detection result, and the defect detection result and the position information can be combined and converted into an identification signal. Wherein the identification signal includes both the position information of the battery and the defect detection result of the battery (i.e. the second signal or the third signal). The group of identification signals are identification signals of all batteries in the battery image to be detected, for example, if the battery image to be detected includes 4 batteries, the group of identification signals includes a defect detection result (for example, whether a defect exists or not) of each battery and position information of each battery.

Step S106: sending the set of identification signals to the controller;

here, step S106 can be seen in step S105a.

Step S107: the controller controls the robotic arm to grasp a battery of the at least two batteries in response to the set of identification signals;

here, since the defect detection result of each battery and the position information of each battery are included in the set of identification signals, the robot arm grasps each of the at least two batteries according to the position information of each battery under the control of the controller when step S107 is performed.

Step S108: based on the set of identification signals, the controller controls the mechanical arm to place the battery of the at least two batteries in the defective product area or the good product area.

Here, also since the set of identification signals includes the defect detection result of each battery and the position information of each battery, when step S108 is performed, the robot arm may be controlled to place the battery without the defect in the good product area and the battery with the defect in the defective product area according to the defect detection result of each battery and the position information of each battery. For example, the battery image to be detected includes two batteries, the battery with the position information of (x 3, y 3) has no defect as the defect detection result, and the battery with the position information of (x 4, y 4) has a defect as the defect detection result, the robot arm places the battery with the position information of (x 3, y 3) in the good product area, and places the battery with the position information of (x 4, y 4) in the defective product area.

In the embodiment of the application, under the condition that the battery image to be detected comprises at least two batteries, the position information of each battery shell is obtained; then converting the defect detection results of at least two battery shells and the position information of each battery shell into a group of identification signals, and sending the identification signals to a controller; and then the controller responds to a group of identification signals and controls the mechanical arm to place the batteries in the at least two batteries in a defective product area or a good product area, so that the good products and the defective products are distinguished.

In some embodiments, as shown in fig. 4A, the battery casing defect detecting system 100 may further include a workbench 401 and a driving component 402, where the workbench includes at least two regions to be placed, each region to be placed is used for placing a battery to be detected, and correspondingly, as shown in fig. 4B, the step S101 "the image capturing module captures an image of the battery to be detected" may be implemented by:

step S1011: the controller responds to a fourth signal, controls the driving component to drive the workbench to rotate, and places a target area to be placed in the area to be detected;

here, the table may have a circular shape, an elliptical shape, a square shape, or the like, and the shape of the table is not limited in the embodiments of the present application. The operation platform comprises at least two areas to be placed, wherein the intervals between the areas to be placed can be uniformly arranged or non-uniformly arranged. One battery can be placed in each region to be placed, and at least two batteries can also be placed in each region to be placed. The target area to be placed is an area to be placed where the battery to be detected is located, for example, the operation platform comprises four areas to be placed, the numbers of which are respectively 1,2,3 and 4, the battery placed in the area to be placed 1 needs to be detected, and the target area to be placed is the area to be placed 1. The driving component is used for driving the workbench to rotate, such as a motor and the like.

In some embodiments, the fourth signal may be a power-on signal of the controller, for example, in a case that a defect of the battery case needs to be tested, after the controller is turned on, a signal, that is, the fourth signal, is given to the controller, and the controller controls the driving component to drive the workbench to rotate in response to the fourth signal, so as to place the target region to be detected in the region to be detected.

In some embodiments, the fourth signal may also be the second signal or the third signal, that is, after the defect detection of the battery to be detected is completed and the defect detection result is obtained, the controller may control the driving component to drive the workbench to rotate, so as to replace the to-be-placed area located in the to-be-detected area, and implement the defect detection of the battery in the next to-be-placed area.

Step S1012: the image acquisition module acquires a battery image to be detected, which is located in the detection area.

Here, step S1012 may be performed in step S101.

In the embodiment of the application, under the condition that battery case defect detecting system includes operation panel and driver part, through making the controller respond to the fourth signal, control driver part drive operation panel rotation, treat the target and put the district in and treat the detection zone for the battery image that waits to detect that the image acquisition module collection is located treats the detection zone, realize that the automation of operation panel rotates and changes and treat and put the district.

In some embodiments, the battery casing defect detecting system 100 may further include a display screen, and the image processing module 102 may send the defect detection result of the battery casing to the display screen, and the display screen displays the defect detection result of the battery casing, so as to realize visualization of the result.

Wherein, the display content of the defect detection result may include the existence of defect or the nonexistence of defect; in the case of a defect, the display content of the defect detection result may further include a defect type and a photo corresponding to the defect type, where the photo corresponding to the defect type is an image corresponding to the detection frame determined as a defect in step S103 or an image corresponding to a potential defect area; in the case that at least two batteries are included in the battery image to be detected, if the defect detection result indicates that a defect exists, the display content of the defect detection result may further include position information of the defective battery, for example, "4".

In the embodiment of the application, under the condition that the battery shell defect detection system comprises the display screen, the visualization of the defect detection result of the battery shell can be realized through the display screen, and the production line staff can conveniently know the battery adverse condition in time and respond.

The embodiment of the present application provides a battery case defect detecting system 100 as shown in fig. 1A, including:

the image acquisition module 101 is positioned above the battery to be detected and used for acquiring the battery image to be detected;

the image processing module 102 is configured to extract a welding surface in the acquired battery image to be detected, so as to obtain a welding surface area of the battery; determining a defect detection result of the battery case based on the welding surface area of the battery.

In some embodiments, as shown in fig. 2A, the battery case defect detection system 100 further includes: the sensor is used for detecting whether the battery to be detected is positioned in a detection area to be detected or not, and sending a first signal to the controller under the condition that the battery to be detected is positioned in the detection area to be detected; the controller is used for responding to the first signal and controlling the image processing module to collect the battery image to be detected.

In some embodiments, as shown in fig. 3A, the battery case defect detection system 100 further includes: the mechanical arm 301 is used for placing the battery to be detected in a good product area or a defective product area; the image processing module 102 is further configured to convert the defect detection result into a second signal and send the second signal to the controller 201 when the defect detection result indicates that no defect exists; under the condition that the defect detection result represents that the defect exists, converting the defect detection result into a third signal, and sending the third signal to the controller 201; the controller 201 is further configured to control the mechanical arm 301 to place the battery to be detected in a good product area in response to the second signal; in response to the third signal, the robot arm 301 is controlled to place the battery to be detected in the defective product region.

In some embodiments, as shown in fig. 4A, the battery case defect detection system 100 further includes: the operation platform 401 comprises at least two regions to be placed, and each region to be placed is used for placing the battery to be detected; a driving unit 402 for driving the work table 401 to rotate; the controller 201 is further configured to control the driving component 402 to drive the work table 401 to rotate in response to a fourth signal, so as to place the target area to be detected in the area to be detected.

The following describes an application of the battery case defect detection method provided in this embodiment in an actual scene, where the method is applied to a battery case defect detection system shown in fig. 5A, and the system is composed of a light source 502, a camera 501 (i.e., the image acquisition module), an edge quality inspection platform 505 (including the image processing module and the Controller), a production line control device (PLC) 503 (i.e., the Controller), a mechanical arm 504, and a display screen 507. The light source 502 and the camera 501 are used as an erection equipment rack to be placed on a production line, and the light source is responsible for supplementing light and irradiating detection points (namely, welding surfaces); the camera is responsible for carrying out image acquisition on the detection point. Light source and camera deployment the light source and camera are mounted on a support 509 and their heights can be adjusted by sliding rails on the support as shown in figure 5B. When the detection device is implemented, the positions of the light source and the camera can be set to be relatively high for the battery with the large model, and the positions of the light source and the camera can be set to be relatively low for the battery with the small model, so that the acquired complete and clear image of the battery to be detected can be acquired.

The method comprises the following steps:

a first part: when the battery shell reaches a specified position, the PLC sends a signal (namely the first signal) to the edge industrial quality inspection platform;

a second part: after receiving the in-place signal of the battery shell, the edge industrial quality inspection platform calls a camera and a light source to shoot a picture (namely the battery image to be detected);

and a third part: the camera returns the shot pictures to the edge industrial quality inspection platform;

the fourth part: the edge industrial quality inspection platform analyzes the photo, analyzes the photo through a network model based on a deep learning visual algorithm, judges whether defects exist or not, and displays the photo and an analysis result on a display screen;

in some embodiments, the edge industrial quality inspection platform may adopt an Nvdia NX platform capable of providing sufficient image processor computing resources, and simultaneously support the network model to perform sensor RT deployment acceleration, thereby improving the operation speed of the network model.

In some embodiments, the fourth part may also optimize a network model based on a deep learning vision algorithm, wherein for detection of a welding surface, a single-stage Retina network model may be used for image detection, and meanwhile, a neural structure search technology is used for pruning a backbone in the Retina network model to obtain a lightweight model, so as to improve the speed of extracting the welding surface. For the defect detection of the welding surface, a network model of a pixel-level segmentation algorithm, such as HRNet, can be adopted to segment the image of the welding surface, and meanwhile, pruning processing is carried out on the basis of an original network model, such as HRNet, to obtain a corresponding lightweight model (such as HRNet tiny), so that the defect detection speed is improved.

In some embodiments, under the condition that pruning processing and TensorRT deployment acceleration are performed on the network model, the operation speed of the network model can be increased, and the beat requirement of a production line can be better met.

A fifth part: the edge industrial quality inspection platform converts the analysis result into corresponding signals (namely the group of identification signals) and transmits the corresponding signals to the PLC;

a sixth part: and the PLC rotates the turntable to send the battery to the next station (namely the target area to be placed) according to the received signal, and controls the mechanical arm to process the battery case according to the detection result.

Based on the foregoing embodiments, the present application provides a battery case defect detection apparatus, which includes units and modules included in the units, and can be implemented by a processor in a computer device; of course, the implementation can also be realized through a specific logic circuit; in the implementation process, the processor may be a central processing unit, a microprocessor, a digital signal processor, a Field Programmable Gate Array (FPGA), or the like.

Fig. 6 is a schematic structural diagram of a battery case defect detecting apparatus according to an embodiment of the present application, and as shown in fig. 6, the battery case defect detecting apparatus 600 includes: an acquisition module 610, an extraction module 620, and a determination module 630, wherein:

the acquisition module 610 is used for acquiring a battery image to be detected by using an image acquisition module;

the extracting module 620 is configured to extract a welding surface in the acquired battery image to be detected by using the image processing module to obtain a welding surface area of the battery;

a determining module 630, configured to determine a defect detection result of the battery case based on the welding surface area of the battery by using the image processing module.

In some embodiments, the battery case defect detection system further comprises a sensor and a controller, the apparatus further comprising: the detection module is used for detecting whether the battery to be detected is positioned in a detection area by using the sensor; the first sending module is used for sending a first signal to the controller by using the sensor under the condition that the battery to be detected is located in the detection area to be detected; and the first control module is used for utilizing the controller to respond to the first signal and controlling the image processing module to acquire the battery image to be detected.

In some embodiments, the battery case defect detection system further comprises a robotic arm, the apparatus further comprising: the first conversion module is used for converting the defect detection result into a second signal by using the image processing module under the condition that the defect detection result represents that no defect exists; the second sending module is used for sending the second signal to the controller by utilizing the image processing module; and the second control module is used for responding to the second signal by utilizing the controller and controlling the mechanical arm to place the battery to be detected in a good product area.

In some embodiments, the battery case defect detection system further comprises a robotic arm, the apparatus further comprising: the second conversion module is used for converting the defect detection result into a third signal by using the image processing module under the condition that the defect detection result represents that the defect exists; the third sending module is used for sending the third signal to the controller by utilizing the image processing module; and the third control module is used for responding to the third signal by utilizing the controller and controlling the mechanical arm to place the battery to be detected in the defective product area.

In some embodiments, the battery images to be detected include at least two batteries, and after determining the defect detection result of each battery case, the apparatus further includes: the acquisition module is used for acquiring the position information of each battery shell by using the image processing module; a third converting module, configured to convert the defect detection results of the at least two battery cases and the position information of each battery case into a set of identification signals, where the identification signals include the second signal or the third signal; a fourth sending module, configured to send the group of identification signals to the controller; a fourth control module for controlling the robotic arm to grasp a battery of the at least two batteries in response to the set of identification signals with the controller; a fifth control module, configured to control the mechanical arm to place the battery of the at least two batteries in the defective product area or the defective product area based on the set of identification signals.

In some embodiments, the obtaining module comprises: the detection submodule is used for detecting the welding surface of each battery in the battery image to be detected by using the image processing module to obtain a detection frame of the welding surface of each battery; and the determining submodule is used for determining the position information of each battery shell based on the position information of the detection frame of the welding surface of each battery in the battery image to be detected.

In some embodiments, the battery casing defect detecting system further includes a workbench and a driving component, wherein the workbench includes at least two regions to be placed, each region to be placed is used for placing the battery to be detected, and the collecting module 610 includes: the control sub-module is used for controlling the driving component to drive the workbench to rotate and place the target area to be placed in the area to be detected by utilizing the controller to respond to a fourth signal; and the acquisition submodule is used for acquiring the battery image to be detected positioned in the detection area by using the image acquisition module.

In some embodiments, the extracting module 620 is further configured to perform image detection on the acquired battery image to be detected by using the image processing module and using a target detection algorithm, so as to obtain a welding surface area of the battery; the determining module 630 is further configured to perform image detection on the welding surface area of the battery by using the image processing module, so as to obtain a defect detection result of the battery case.

In some embodiments, the welding location of the welding surface includes at least one of: the positive and negative terminal outer edges, the liquid injection port outer edge, the vent hole outer edge and the top plate outer edge; the type of defect includes at least one of: the determining module 630 is further configured to perform image detection on at least one region of the outer edges of the positive and negative terminals, the outer edge of the liquid injection port, the outer edge of the vent hole, and the outer edge of the top plate of the battery by using the image processing module to obtain a defect detection result of the battery case, where the defect detection result includes a defect type of at least one of the hole, the gap, the bulge, and the damage.

The above description of the apparatus embodiments, similar to the above description of the method embodiments, has similar beneficial effects as the method embodiments. In some embodiments, functions of or modules included in the apparatus provided in the embodiments of the present disclosure may be used to perform the methods described in the above method embodiments, and for technical details not disclosed in the embodiments of the apparatus of the present disclosure, please refer to the description of the embodiments of the method of the present disclosure for understanding.

If the technical scheme of the application relates to personal information, a product applying the technical scheme of the application clearly informs personal information processing rules before processing the personal information, and obtains personal independent consent. If the technical scheme of the application relates to sensitive personal information, before the sensitive personal information is processed, a product applying the technical scheme of the application obtains individual consent and simultaneously meets the requirement of 'explicit consent'. For example, at a personal information collection device such as a camera, a clear and significant identifier is set to inform that the personal information collection range is entered, the personal information is collected, and if the person voluntarily enters the collection range, the person is regarded as agreeing to collect the personal information; or on the device for processing the personal information, under the condition of informing the personal information processing rule by using obvious identification/information, obtaining personal authorization by modes of popping window information or asking a person to upload personal information of the person by himself, and the like; the personal information processing rule may include information such as a personal information processor, a personal information processing purpose, a processing method, and a type of personal information to be processed.

It is to be noted here that: the foregoing description of the various embodiments is intended to highlight differences between various embodiments that are the same or similar and that which may be referenced otherwise. The above description of the system and apparatus embodiments is similar to the above description of the method embodiments, with similar beneficial effects. For technical details not disclosed in the embodiments of the system and apparatus of the present application, reference is made to the description of the embodiments of the method of the present application for understanding.

It should be appreciated that reference throughout this specification to "one embodiment" or "an embodiment" means that a particular feature, structure or characteristic described in connection with the embodiment is included in at least one embodiment of the present application. Thus, the appearances of the phrases "in one embodiment" or "in an embodiment" in various places throughout this specification are not necessarily all referring to the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments. It should be understood that, in the various embodiments of the present application, the sequence numbers of the above steps/processes do not mean the execution sequence, and the execution sequence of the steps/processes should be determined by their functions and inherent logic, and should not constitute any limitation to the implementation process of the embodiments of the present application. The above-mentioned serial numbers of the embodiments of the present application are merely for description and do not represent the merits of the embodiments.

It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrases "comprising a component of' 8230; \8230;" does not exclude the presence of another like element in a process, method, article, or apparatus that comprises the element.

In the several embodiments provided in the present application, it should be understood that the disclosed apparatus and method may be implemented in other manners. The above-described device embodiments are merely illustrative, for example, the division of the unit is only a logical functional division, and there may be other division ways in actual implementation, such as: multiple units or modules may be combined, or may be integrated into another system, or some features may be omitted, or not implemented. In addition, the coupling, direct coupling or communication connection between the components shown or discussed may be through some interfaces, and the indirect coupling or communication connection between the devices or units may be electrical, mechanical or other forms.

The units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units; can be located in one place or distributed on a plurality of network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiment.

In addition, all functional units in the embodiments of the present application may be integrated into one processing unit, or each unit may be separately regarded as one unit, or two or more units may be integrated into one unit; the integrated unit can be realized in a form of hardware, or in a form of hardware plus a software functional unit.

Those of ordinary skill in the art will understand that: all or part of the steps for realizing the method embodiments can be completed by hardware related to program instructions, the program can be stored in a computer readable storage medium, and the program executes the steps comprising the method embodiments when executed; and the aforementioned storage medium includes: various media that can store program codes, such as a removable Memory device, a Read Only Memory (ROM), a magnetic disk, or an optical disk.

Alternatively, the integrated units described above in the present application may be stored in a computer-readable storage medium if they are implemented in the form of software functional modules and sold or used as independent products. Based on such understanding, the technical solutions of the present application or portions thereof contributing to the related art may be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a removable storage device, a ROM, a magnetic or optical disk, or other various media that can store program code.

The above description is only an embodiment of the present application, but the scope of the present application is not limited thereto, and any person skilled in the art can easily think of the changes or substitutions within the technical scope of the present application, and shall cover the scope of the present application.

Claims

1. A battery shell defect detection method is applied to a battery shell defect detection system, the battery shell defect detection system comprises an image acquisition module and an image processing module, and the method comprises the following steps:

the image acquisition module acquires a battery image to be detected;

the image processing module extracts a welding surface in the collected battery image to be detected to obtain a welding surface area of the battery;

the image processing module determines a defect detection result of the battery shell based on the welding surface area of the battery.

2. The method of claim 1, wherein the battery case defect detection system further comprises a sensor and a controller, the method further comprising:

the sensor detects whether a battery to be detected is located in a detection area;

under the condition that the battery to be detected is located in the detection area, the sensor sends a first signal to the controller;

the controller responds to the first signal and controls the image processing module to collect the battery image to be detected.

3. The method of claim 2, wherein the battery housing defect detection system further comprises a robotic arm, after determining the defect detection result for the battery housing, further comprising:

under the condition that the defect detection result represents that no defect exists, the image processing module converts the defect detection result into a second signal;

the image processing module sends the second signal to the controller;

and the controller responds to the second signal and controls the mechanical arm to place the battery to be detected in a good product area.

4. The method of claim 3, wherein the battery case defect detection system further comprises a robotic arm, after determining the defect detection result for the battery case, further comprising:

under the condition that the defect detection result represents that the defect exists, the image processing module converts the defect detection result into a third signal;

the image processing module sends the third signal to the controller;

and the controller responds to the third signal and controls the mechanical arm to place the battery to be detected in the defective product area.

5. The method according to claim 4, wherein the battery images to be detected comprise at least two batteries, and after determining the defect detection result of each battery shell, the method further comprises:

the image processing module acquires the position information of each battery shell; converting the defect detection results of the at least two battery cases and the position information of each battery case into a set of identification signals, wherein the identification signals comprise the second signal or the third signal; sending the set of identification signals to the controller;

the controller controls the robotic arm to grasp a battery of the at least two batteries in response to the set of identification signals;

based on the group of identification signals, the controller controls the mechanical arm to place the battery in the at least two batteries in the defective product area or the good product area.

6. The method of claim 5, wherein the image processing module obtains the position information of each battery shell, comprising:

the image processing module detects the welding surface of each battery in the battery image to be detected to obtain a detection frame of the welding surface of each battery;

and determining the position information of each battery shell based on the position information of the detection frame of the welding surface of each battery in the battery image to be detected.

7. The method according to any one of claims 2 to 6, wherein the battery housing defect detecting system further comprises a workbench and a driving part, wherein the workbench comprises at least two regions to be placed, each region to be placed is used for placing the battery to be detected, and the image collecting module collects the image of the battery to be detected, and comprises:

the controller responds to a fourth signal, controls the driving component to drive the workbench to rotate, and places a target area to be placed in the area to be detected;

the image acquisition module acquires a battery image to be detected positioned in the detection area.

8. The method according to any one of claims 1 to 7, wherein the image processing module extracts the welding surface in the collected image of the battery to be detected to obtain the welding surface area of the battery, and the method comprises the following steps:

the image processing module carries out image detection on the collected battery image to be detected by using a target detection algorithm to obtain a welding surface area of the battery;

the image processing module determines a defect detection result of the battery case based on the welding surface area of the battery, and comprises:

and the image processing module performs image detection on the welding surface area of the battery to obtain a defect detection result of the battery shell.

9. The method of claim 8, wherein the weld location of the weld face comprises at least one of: the positive and negative terminal outer edges, the liquid injection port outer edge, the vent hole outer edge and the top plate outer edge; the type of defect includes at least one of: the image processing module carries out image detection on the welding surface area of the battery to obtain the defect detection result of the battery shell, and the defect detection result comprises the following steps:

the image processing module performs image detection on at least one region of the outer edges of the positive and negative terminals, the outer edge of the liquid injection port, the outer edge of the vent hole and the outer edge of the top plate of the battery to obtain a defect detection result of the battery shell, wherein the defect detection result comprises defect types of at least one of cavities, sand holes, irregular edges, gaps, bulges and damages.

10. A battery case defect detection system, comprising:

the image acquisition module is positioned above the battery to be detected and used for acquiring the battery image to be detected;

the image processing module is used for extracting the welding surface in the collected battery image to be detected to obtain the welding surface area of the battery; determining a defect detection result of the battery case based on the welding surface area of the battery.

11. The system of claim 10, further comprising:

the sensor is used for detecting whether the battery to be detected is positioned in a region to be detected or not, and sending a first signal to the controller under the condition that the battery to be detected is positioned in the region to be detected;

the controller is used for responding to the first signal and controlling the image processing module to collect the battery image to be detected.

12. The system of claim 11, further comprising:

the mechanical arm is used for placing the battery to be detected in a good product area or a defective product area;

the image processing module is further configured to convert the defect detection result into a second signal and send the second signal to the controller when the defect detection result indicates that no defect exists; under the condition that the defect detection result represents that the defect exists, converting the defect detection result into a third signal, and sending the third signal to the controller;

the controller is further used for responding to the second signal and controlling the mechanical arm to place the battery to be detected in a good product area; and responding to the third signal, controlling the mechanical arm to place the battery to be detected in a defective product area.

13. The system of claim 11 or 12, further comprising:

the operation platform comprises at least two regions to be placed, and each region to be placed is used for placing the battery to be detected;

the driving component is used for driving the workbench to rotate;

the controller is further used for responding to a fourth signal, controlling the driving component to drive the operation platform to rotate, and placing the target area to be placed in the area to be detected.

14. A battery case defect detection apparatus, comprising:

the acquisition module is used for acquiring a battery image to be detected by using the image acquisition module;

the extraction module is used for extracting the welding surface in the acquired battery image to be detected by using the image processing module to obtain the welding surface area of the battery;

and the determining module is used for determining the defect detection result of the battery shell based on the welding surface area of the battery by utilizing the image processing module.