Surface flaw detection device for battery shell processing

Abstract

The invention belongs to the field of new energy battery production, and particularly relates to a surface flaw detection device for battery shell processing, which comprises a bearing mechanism which comprises a first bearing table, a second bearing table and a plurality of groups of battery carriers, a control system is arranged at one end of the bearing mechanism, and a guiding, displacement, multi-channel transmission, grabbing detection, visual detection and pivot domain intelligent detection mechanism is arranged on the bearing mechanism; the battery carrier is conveyed to the guide mechanism through the first transmission mechanism, and the displacement mechanism conveys a battery to the multi-channel transmission mechanism; after the grabbing detection mechanism grabs a battery, the visual detection module is moved to detect, the visual detection module comprises a polarized light source unit, a multi-channel polarization imaging unit and the like, oxidation artifacts and physical deformation characteristics are decoupled through a deep convolutional neural network, and the pivot domain intelligent detection mechanism comprises a rotating table, a tab detection part, an airtight detection part and the like. And the grabbing detection mechanism puts the defective battery into a defective product placing table, and the qualified battery enters a discharging station after being subjected to tab and air tightness detection, so that full-process automatic defect detection of the battery shell is realized.

Description

Technical Field

[0001] This invention belongs to the field of new energy battery production, specifically a surface defect detection device for battery casing processing. Background Technology

[0002] During battery production, the surface quality of the battery casing directly affects the battery's performance and safety.

[0003] For example, publication number CN118483223A, entitled "A Battery Case Testing Machine," describes a battery case testing machine comprising a workbench. One end of the workbench is equipped with a feeding mechanism for conveying battery cases. The workbench has an internal cavity, and a support base is fixedly connected to the center of the cavity. A rotating assembly is mounted at the bottom of the support base, and a rotating platform is fixedly connected to the top of the rotating assembly, passing through the support base. A clamping module for holding battery cases is mounted at the top of the rotating platform. Several sets of clamping modules are arranged and distributed around a center on the surface of the rotating platform. One end of the support base is fixedly connected to a feeding mechanism for loading materials, and the other end of the support base is equipped with a line scan camera for scanning battery cases. A collection assembly for collecting the tested battery cases is mounted at the bottom of the workbench cavity.

[0004] Currently, battery casing inspection typically involves multiple steps, including surface defect detection, tab detection, and airtightness testing. However, these steps are often performed on different independent devices. This decentralized inspection method has significant drawbacks: First, the process is severely fragmented, with a lack of effective connection between each inspection step, resulting in an incoherent inspection flow. Second, manual transfer of battery casings between multiple devices increases the frequency of human intervention and prolongs the inspection cycle. Third, improper handling during manual transfer can easily cause secondary damage to the battery casings, affecting product quality. Finally, this inspection method is inefficient overall and cannot meet the demands of modern mass production of batteries. Furthermore, an oxide layer easily forms on the surface of the battery casing over long-term use or under specific processes. This oxide layer exhibits spectral reflectance characteristics similar to real pit defects in the near-infrared band (e.g., 940nm), leading to significant artifact interference in the acquired optical images. Deep learning models struggle to effectively distinguish between artifacts caused by the oxide layer and real physical pits when processing this type of data, resulting in a high false positive rate and severely impacting the accuracy and reliability of the inspection system.

[0005] Therefore, a surface defect detection device for battery casing processing is proposed to address the above problems. Summary of the Invention

[0006] The purpose of this invention is to address the shortcomings of the prior art by providing a surface defect detection device for battery casing processing, thereby solving the technical problems mentioned in the background art.

[0007] To address the above technical problems, the following technical solution is adopted: A surface defect detection device for battery casing processing includes a carrying mechanism. The carrying mechanism is equipped with at least a guiding mechanism, a displacement mechanism, a multi-channel transmission mechanism, a gripping detection mechanism, a vision detection module, and a pivot-based intelligent detection mechanism. Multiple battery carriers are mounted on the carrying mechanism, and a control system is located at one end of the carrying mechanism. The guiding mechanism includes a guide shaft, and a clamping component is located on one side of the guide shaft. The guide mechanism moves the battery carriers below the displacement mechanism. The displacement mechanism is located on one side of the clamping component, and a multi-channel transmission mechanism is located on one side of the displacement mechanism. The displacement mechanism delivers the batteries within the battery carriers to the multi-channel transmission mechanism. The multi-channel transmission mechanism is located away from the positioning mechanism. A gripping and inspection mechanism capable of multi-angle displacement and horizontal rotation is provided on one side of the moving mechanism. A vision inspection module is provided below the gripping and inspection mechanism. When the gripping and inspection mechanism grips the battery at the tail end of the multi-channel transmission mechanism, it moves it closer to the vision inspection module for 360° rotational vision defect detection. A pivot-based intelligent inspection mechanism is provided on the side of the gripping and inspection mechanism opposite to the vision inspection module. The pivot-based intelligent inspection mechanism includes at least an electrode detection mechanism, an airtightness detection fixture mechanism, a loading station, and an unloading station. A defective product placement table is provided on one side of the pivot-based intelligent inspection mechanism. The gripping and inspection mechanism places batteries with surface defects into the defective product placement table and batteries without surface defects into the unloading station.

[0008] Preferably, the gripping and detection mechanism includes a base control unit, a first horizontally movable extension arm is provided above the base control unit, a second horizontally movable extension arm is provided above the first horizontally movable extension arm and on the side away from the base control unit, a longitudinal telescopic unit is provided on the side of the second horizontally movable extension arm away from the first horizontally movable extension arm, and a mechanical gripping unit that can rotate laterally is provided below the longitudinal telescopic unit.

[0009] Preferably, the gripping and testing mechanism places the qualified batteries into the loading station, and then rotates them sequentially to the tab testing mechanism and the airtightness testing fixture mechanism for testing. An airtightness testing unit is connected to one side of the airtightness testing fixture mechanism.

[0010] Preferably, the bearing mechanism includes a first bearing platform and a second bearing platform, and a multi-channel transmission mechanism is disposed between the first bearing platform and the second bearing platform.

[0011] Preferably, the pivot-based intelligent inspection mechanism includes a rotary table, with a position sensor facing the loading station located at the center of the top of the rotary table; a reinforcing structure is provided on one side of both the tab detection mechanism and the airtightness detection fixture mechanism, and the bottom of the reinforcing structure is connected to the rotary table; a rotating component is located at the center of the bottom of the rotary table, and a support component is located around the rotating component and below the rotary table, with a pulley at one end of the support component that contacts the rotary table, and the other end of the support component connected to the bearing mechanism; a power source is also provided below the rotary table, which provides power to the rotating component to rotate the pivot-based intelligent inspection mechanism.

[0012] Preferably, the first carrier platform is provided with a first transmission mechanism for transferring the battery carrier to the guide mechanism. The first transmission mechanism includes fixed frames on both sides above the first transmission mechanism, a connecting rod between the fixed frames, and a transmission component on the surface of the connecting rod. The clamping component can slide on the guide shaft, and the clamping height of the clamping component can be adjusted up and down.

[0013] Preferably, the displacement mechanism includes a support fixed on the first bearing platform, a transverse guide post on the support, a guide member and a longitudinal telescopic mechanism sequentially arranged on the side of the transverse guide post facing the battery carrier; the guide member allows the longitudinal telescopic mechanism to move on the transverse guide post; a clamping arm is arranged below the longitudinal telescopic mechanism.

[0014] Preferably, the multi-channel transmission mechanism includes a transmission platform, a transmission belt surrounding the transmission platform, and a multi-channel limiting member above the transmission platform. The channels of the multi-channel limiting member are used for transmitting batteries. A portal frame is provided on the upper part of the multi-channel limiting member, which is connected to the base of the transmission mechanism at the bottom and to the multi-channel limiting member at the top, so that the multi-channel limiting member is suspended above the transmission platform. A distance sensor is provided on the top of the portal frame provided on the upper part of the multi-channel limiting member for detecting the position of the battery in the channel.

[0015] The beneficial effects of this invention are:

[0016] By integrating a guiding mechanism, a displacement mechanism, a multi-channel transmission mechanism, a gripping and detection mechanism, and a pivot-based intelligent inspection mechanism, continuous automated inspection of battery casing surface defects, tabs, and airtightness is achieved. This solves the problems of low efficiency and excessive manual intervention in traditional decentralized inspection methods, and has the advantages of improving inspection efficiency, reducing manual intervention, avoiding secondary damage, and achieving multi-stage integrated inspection.

[0017] By constructing a composite probe optical field with multiple spectra and polarization states, and utilizing the inherent differences in the optical response characteristics of the oxide layer and the substrate material under different wavebands and polarization states, the dimensions of characterization information are greatly enriched from the data source. The spectral polarization fusion processing module provides a high signal-to-noise ratio input data foundation for subsequent feature decoupling through precise calculation of this multidimensional information, fundamentally improving the system's robustness against oxidation interference.

[0018] By employing a multi-branch deep network structure through an oxidation feature decoupling module, the system can actively learn and separate features sensitive to oxidation artifacts and features sensitive to physical deformation. In particular, by introducing orthogonal constraint loss on feature vectors, the decoupling of the two types of interference signals is enforced at the algorithm level, ensuring the purity of the physical deformation feature vector. This allows the subsequent defect classifier to make decisions based on more fundamental deformation information, significantly reducing the false positive rate.

[0019] By deeply integrating optical physics mechanisms with deep learning models, a feature decoupling architecture guided by a physical model is formed. This system does not rely entirely on a data-driven black-box model; instead, it incorporates prior knowledge of oxide layer spectral polarization into the design of the network structure and loss function. This allows the entire system to possess both physical interpretability and model generalization ability, maintaining stable and reliable detection performance even when faced with minute changes in the surface state of the battery casing or novel oxidation modes. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] In the attached diagram:

[0022] Figure 1 This is a perspective view of the present invention;

[0023] Figure 2 This is a top view of the present invention;

[0024] Figure 3 This is a front view of the present invention;

[0025] Figure 4 This is a schematic diagram of the structure of the guide mechanism and displacement mechanism cooperating on the first vehicle in this invention;

[0026] Figure 5 This is a schematic diagram of the multi-channel transmission mechanism in this invention;

[0027] Figure 6This is a schematic diagram of the pivot-domain intelligent inspection mechanism in this invention;

[0028] Figure 7 This is a schematic diagram of the structure of the rotating platform and the support component of the present invention.

[0029] Figure 8 This is a schematic diagram of the gripping and detection mechanism in this invention;

[0030] Figure 9 This is a schematic diagram of the overall technical architecture of the polarization decoupling detection system for suppressing battery casing oxidation interference proposed in this invention;

[0031] Figure 10 This is a schematic diagram of the core principle framework of the oxidation feature decoupling module in this invention.

[0032] Legend:

[0033] 100. Bearing mechanism; 101. First bearing platform; 102. Second bearing platform; 200. First transmission mechanism; 201. Fixed frame; 202. Connecting rod; 203. Transmission component; 301. Guide shaft; 302. Clamping component; 1. Battery carrier; 400. Displacement mechanism; 401. Lateral guide post; 402. Guide component; 404. Clamping arm; 403. Longitudinal telescopic mechanism; 500. Multi-channel transmission mechanism; 501. Transmission platform; 502. Multi-channel limiting component; 503. Distance sensor; 600. Rotary pivot domain intelligent detection mechanism; 601. 602. Rotary table; 603. Rotating component; 604. Support component; 605. Position sensor; 7. Loading station; 8. Unloading station; 4. Electrode detection mechanism; 3. Air tightness detection unit; 5. Air tightness detection fixture mechanism; 6. Reinforcing structure; 700. Gripping and detection mechanism; 701. Base control unit; 702. First horizontal extension arm; 703. Second horizontal extension arm; 704. Longitudinal telescopic unit; 705. Mechanical gripping unit; 706. Contact monitoring module; 9. Visual inspection module; 10. Defective product placement table; 11. Control system. Detailed Implementation

[0034] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0035] Specific implementation examples are given below.

[0036] Example 1

[0037] Please see Figures 1-10This invention provides a surface defect detection device for battery casing processing, including a support mechanism 100. The support mechanism 100 is equipped with a guide mechanism, a displacement mechanism 400, a multi-channel transmission mechanism 500, a gripping detection mechanism 700, a vision inspection module 9, and a pivot-based intelligent inspection mechanism 600. The support mechanism 100 is equipped with multiple battery carriers 1, and a control system 11 is installed at one end. The guide mechanism includes a guide shaft 301 and a clamping component 302, used to move the battery carriers 1 below the displacement mechanism 400. The displacement mechanism 400 delivers the batteries within the battery carriers 1 to the multi-channel transmission mechanism 500. At the end of the multi-channel transmission mechanism 500 is a gripping detection mechanism 700 capable of multi-angle displacement and horizontal rotation, with the vision inspection module 9 installed below it. A contact monitoring module 706 is installed inside the gripping detection mechanism 700. The pivot-type intelligent inspection mechanism 600 includes a tab inspection mechanism 4, an airtightness inspection fixture mechanism 5, a loading and unloading station 8, and a defective product placement platform 10, which realizes automatic sorting of batteries after inspection.

[0038] The visual inspection module 9 consists of a polarization light source module, a multi-channel polarization imaging module, a spectral polarization fusion processing module, an oxidation feature decoupling module, and a defect decision output module connected sequentially. The modules are seamlessly integrated physically and logically via a high-speed data bus and an optomechanical interface, ensuring complete acquisition of optical signals and low-latency transmission of electrical signals.

[0039] The polarization light source module projects an illumination field with a specific polarization state and spectral combination onto the surface of the battery casing under test. This module includes a programmable light-emitting unit and a polarization control unit. The programmable light-emitting unit consists of two sets of narrowband semiconductor lasers with different center wavelengths. The first set of lasers has a center wavelength of 660 nm, a typical output power of 100 mW, and a beam divergence angle of 10 milliradians. The second set of lasers has a center wavelength of 940 nm, a typical output power of 150 mW, and a beam divergence angle of 8 milliradians. Both sets of lasers integrate temperature control circuitry and a drive current feedback loop to ensure the stability of the output wavelength within ±0.1 nm. The polarization control unit is located in the light output path of the light-emitting unit and consists of two independently rotatable linear polarizers and one quarter-wave plate. The rotation angle of the first linear polarizer is precisely controlled by a stepper motor, achieving a positioning accuracy of 0.1 degrees. The second linear polarizer is spatially orthogonal to the first, and its rotation mechanism is fine-tuned via a worm gear drive. A quarter-wave plate is fixedly mounted on an electrically controllable rotating bracket, with its fast axis at a 45-degree angle to the polarization direction of the incident light. By coordinating the rotation angles of the two linear polarizers and the insertion state of the quarter-wave plate, this module can generate illumination light in four linearly polarized states (0°, 45°, 90°, and 135°) and two circularly polarized states (left-handed and right-handed). The uniformity of the illumination field is optimized using a pair of aspherical homogenizing lenses, resulting in a light intensity distribution non-uniformity of less than 5% within the illumination area 300 mm from the light outlet.

[0040] The multi-channel polarization imaging module comprises a beam-splitting prism assembly, a polarization camera, and a near-infrared intensifier camera. The beam-splitting prism assembly has a cubic structure with an internal beam-splitting film. This film has a reflectivity of 95% at 660 nm and a transmittance of 92% at 940 nm. Incident light is spectrally split into visible and near-infrared channels by the beam-splitting prism assembly. The visible light channel connects to the polarization camera, which incorporates a pixelated polarization filter array with polarization directions of 0°, 45°, 90°, and 135°. Each polarization direction corresponds to an independent image sensor region, with a pixel size of 2048 x 1536. The polarization camera operates at a frame rate of 30 frames per second and a dynamic range of 70 dB. The near-infrared channel connects to the near-infrared intensifier camera, whose spectral response range covers 850 nm to 1050 nm, achieving a quantum efficiency of 45% at 940 nm. This camera uses a cooled sensor with dark current noise below 0.1 electrons per second, enabling the acquisition of high signal-to-noise ratio images in the 940 nm band. The two cameras achieve exposure synchronization via a hardware trigger signal, with a synchronization error of less than 1 microsecond. Both cameras feature low-distortion lenses with a distortion coefficient of less than 0.1%, and are equipped with an autofocus mechanism to ensure that the imaging plane is conjugate to the battery casing surface.

[0041] The spectral polarization fusion processing module performs image preprocessing, multimodal data registration, and polarization parameter calculation. Image preprocessing performs dark current correction and flat-field correction on the original image. Dark current correction is achieved by acquiring a set of dark-field images with exposure time and gain settings consistent with normal acquisition conditions. Flat-field correction uses a standard diffuse white board as a reference target, acquiring its image under uniform illumination as a flat-field template. The corrected pixel grayscale values ​​are calculated using formulas. Multimodal data registration maps images acquired by different cameras and polarization directions to the same world coordinate system based on pre-calibrated camera intrinsic and extrinsic parameters. The calibration process uses a checkerboard calibration board, and the camera's intrinsic parameter matrix, distortion coefficients, and extrinsic parameter matrix are solved using the Zhang Zhengyou calibration method. The registration algorithm employs a perspective transformation model, and the transformation matrix is ​​solved based on at least 20 pairs of matching feature points. Polarization parameter calculation calculates the degree of polarization and polarization angle for each registered pixel using the Stokes vector method. The Stokes vector contains four components, calculated from the intensity values ​​of the four polarization directions. The degree of polarization ranges from 0 to 1, and the polarization angle ranges from 0 to 180 degrees. For each spectral band, a set of polarization parameters is calculated independently, and these parameters are combined to form the multispectral polarization parameter matrix for that pixel. This matrix has a dimension of 2 x 3, where the first row corresponds to the degree of polarization, polarization angle, and intensity of the 660 nm band, and the second row corresponds to the degree of polarization, polarization angle, and intensity of the 940 nm band.

[0042] The oxidation feature decoupling module embeds a trained deep convolutional neural network model. The network's input is a multispectral polarization parameter matrix, and its structure includes three feature extraction branches and one feature fusion layer. The first branch extracts features sensitive to surface physical deformation at the 660 nm wavelength. This branch consists of three convolutional layers and two pooling layers, with a 3x3 kernel size and a stride of 1. The first convolutional layer has 32 output channels, the second has 64, and the third has 128. Each convolutional layer is followed by a modified linear unit activation function and a batch normalization layer. The second branch extracts features sensitive to oxide layer thickness variations at the 940 nm wavelength. Its structure is symmetrical to the first branch, but the input data is the polarization parameters at the 940 nm wavelength. The third branch analyzes the differences in polarization state changes between the two wavelengths. The input to this branch is a concatenated tensor of the polarization parameters for the two wavelengths, and its convolutional kernel count is the sum of the first two branches. The feature fusion layer concatenates the outputs of the three branches and regresses the oxidation artifact feature vector and physical deformation feature vector through two fully connected layers, respectively. The number of neurons in the fully connected layers is 512 and 256, respectively. The dimension of both feature vectors is 128. The training process of this network is as follows: First, a sample database containing known oxide layer states and real pit defects is constructed. The total number of samples is 10,000, of which 70% is used for training, 15% for validation, and 15% for testing. Each sample is labeled with an oxide layer region mask and a real defect region mask.

[0043] During training, the network's total loss function consists of three weighted parts. The first part is the mean squared error loss between the physical deformation feature vector and the mask of the real defect region, with a weighting coefficient of 0.6. The second part is the mean squared error loss between the oxide artifact feature vector and the mask of the oxide layer region, with a weighting coefficient of 0.3. The third part is the orthogonality constraint loss of the two feature vectors in the latent space, with a weighting coefficient of 0.1. The orthogonality constraint loss is achieved by calculating the square of the inner product of the two feature vectors, used to enforce the independence of their representational information. Training uses the adaptive moment estimation algorithm, with an initial learning rate of 0.001, a batch size of 32, and a training epoch of 100.

[0044] The defect decision output module receives the physical deformation feature vector from the oxidation feature decoupling module. This module first normalizes the feature vector using a minimum-maximum scaling (MPS) method, linearly transforming the values ​​of each dimension to the range of 0 to 1. The normalized feature vector is then input into a support vector machine (SVM) classifier. This classifier is based on a radial basis function (RBF) kernel, whose parameters are determined using a grid search method, with an optimal value of 0.1. The classifier maps the feature vector to a high-dimensional space for linear segmentation and outputs the probability value of a pixel belonging to a pit defect. The probability value is calculated based on the signed distance to the classification hyperplane and converted to a probability between 0 and 1 using a logistic function.

[0045] Finally, the module compares the probability value with two preset decision thresholds. A probability value higher than 0.95 is considered a confirmed defect; a probability value between 0.7 and 0.95 is considered a suspected defect requiring re-inspection; and a probability value lower than 0.7 is considered defect-free. The decision result is correlated with the original image coordinates to generate a defect distribution map with location markers.

[0046] The visual inspection module 9 operates within a real-time processing framework, which specifies that the end-to-end latency from image acquisition to result output shall not exceed 500 milliseconds. During the initialization phase, the visual inspection module 9 requires a white balance and optical system calibration. White balance is achieved by acquiring a standard color chart image and calculating the gain compensation coefficients for each channel. Optical system calibration includes light source uniformity correction, camera focal length adjustment, and coordinate system calibration.

[0047] During continuous inspection, the vision inspection module 9 cyclically executes the data acquisition, fusion, decoupling, and decision-making process at a frame rate of 10 Hz. Within each cycle, the vision inspection module 9 first activates the polarization light source module and the multi-channel polarization imaging module via a trigger signal to acquire a multispectral polarization image. The acquired data is transmitted via gigabit Ethernet to the spectral polarization fusion processing module, which completes image preprocessing and polarization parameter calculation within 50 milliseconds. The oxidation feature decoupling module extracts and decouples feature vectors within 100 milliseconds. The defect decision output module completes classification and decision-making within 20 milliseconds. The final result is uploaded to the production line monitoring system via the industrial Ethernet protocol to ensure synchronization with the production line cycle.

[0048] The supporting mechanism 100 refers to the basic platform supporting each functional module. It can adopt a combination structure of aluminum alloy frame and steel plate, and modular assembly is achieved through bolt connection. It is used to coordinate the spatial layout and power transmission of each inspection station. The clamping component 302 in the guiding mechanism can slide along the guide shaft 301. It adopts a pneumatic gripper and servo motor drive to achieve precise positioning and transmission path control of the battery carrier 1. The multi-channel transmission mechanism 500 adopts a combination structure of belt conveyor and channel limiter. The channel width of the limiter is adjustable to adapt to the batch transportation needs of batteries of different specifications. The rotary pivot domain intelligent inspection mechanism 600's rotary table 601 is driven by a geared motor. The rotation angle is controlled by encoder feedback, so that each inspection station switches to the operating position according to a preset rhythm.

[0049] Specifically, after the battery carrier 1 is positioned by the guiding mechanism, the displacement mechanism 400 transfers the battery to the multi-channel transmission mechanism 500. When the transmission mechanism delivers the battery to the end, the gripping and inspection mechanism 700 acquires battery size and surface resistance data through the contact monitoring module 706, screening out products with obvious defects. Qualified batteries are clamped above the vision inspection module 9 for 360° rotation scanning; the image data is processed by an algorithm to identify minor scratches or dents. Batteries that pass inspection are placed by the gripping mechanism at the loading station 7 of the rotary intelligent inspection mechanism 600. The rotary table 601 sequentially delivers the batteries to the tab inspection station and the airtightness inspection station. Tab inspection uses laser ranging and infrared imaging technology to identify tab welding quality and dimensional deviations. The airtightness inspection fixture pressurizes the battery cavity, and a pressure sensor monitors the leakage rate. Finally, qualified products enter the unloading station 8, while defective products are transferred by the gripping mechanism to a designated placement table.

[0050] Compared to existing technologies, traditional decentralized inspection requires three manual loading / unloading operations and equipment switching. This solution achieves continuous operation throughout the entire process through multi-channel transmission and a rotary sorting mechanism. In existing technologies, visual inspection and contact inspection are at different workstations. This device simultaneously completes initial contact inspection and visual re-inspection during the gripping process, reducing the single operation time by 40%. The 600-type rotary pivot-based intelligent inspection mechanism replaces the original independent inspection equipment, reducing the floor space by more than 60%, and eliminates inspection errors caused by manual placement deviations through tooling fixture positioning.

[0051] Through the above technical solutions, this application achieves integrated detection of surface defects, tab defects, and airtightness indicators in battery casings, eliminating material transfer links between processes. The coordinated control of the multi-channel transmission mechanism 500 and the gripping detection mechanism 700 ensures that the detection cycle matches the production line speed, preventing the detection process from becoming a production bottleneck. The automated sorting mechanism reduces the frequency of manual intervention and minimizes the risk of secondary damage caused by handling. The multi-station integrated design of the rotary sorting mechanism improves space utilization, enabling a single device to complete the detection tasks of three traditionally independent production lines.

[0052] Furthermore, such as Figure 8 As shown, the gripping and detection mechanism 700 includes a base control unit 701. A first horizontally movable extension arm 702 is provided above the base control unit 701. A second horizontally movable extension arm 703 is provided above the first horizontally movable extension arm 702 and on the side away from the base control unit 701. A longitudinal telescopic unit 704 is provided on the side of the second horizontally movable extension arm 703 away from the first horizontally movable extension arm 702. A mechanical gripping unit 705 that can rotate laterally is provided below the longitudinal telescopic unit 704. A contact monitoring module 706 is provided inside the mechanical gripping unit 705.

[0053] The base control unit 701 provides basic support and motion control, and can be implemented using a translation platform driven by a servo motor to provide stable reference positioning for the overall mechanism. The first horizontal extension arm 702 and the second horizontal extension arm 703 are robotic arm structures with independent lateral movement capabilities, and can be implemented using a transmission system combining linear guides and ball screws, expanding the gripping range through superimposed dual lateral displacements. The longitudinal telescopic unit 704 is an actuator that adjusts vertical displacement, and can be implemented using a cylinder or electric push rod structure to precisely control the gripping height. The mechanical gripping unit 705 is a clamping device with rotational function, and can be implemented using a pneumatic gripper integrated with a rotary motor structure, adjusting the gripping angle through lateral rotation to adapt to different workstation postures. The contact monitoring module 706 is a detection sensor integrated at the gripping end, and can be implemented using a pressure sensor array or contact probes to simultaneously collect battery surface state data during gripping.

[0054] Specifically, the base control unit 701 achieves the baseline positioning of the entire mechanism in the horizontal plane through servo motor drive. The first horizontal extension arm 702 performs basic lateral displacement along the linear guide rail, and the second horizontal extension arm 703 performs secondary lateral expansion based on the displacement of the first horizontal extension arm 702, forming a composite movement path to cover the needs of multiple workstations. After lateral positioning, the longitudinal telescopic unit 704 achieves precise vertical downward movement through electric push rod drive. When contacting the battery, the mechanical gripping unit 705 adjusts the clamping angle through a rotary motor to match the battery placement posture. The contact monitoring module 706 collects real-time pressure distribution data on the battery surface during the clamping action and determines whether there are dents or protrusions through the signal processing module. The various motion units achieve multi-degree-of-freedom coordinated movement in three-dimensional space through a layered and superimposed construction method, avoiding repeated positioning or posture adjustment caused by insufficient movement in a single direction.

[0055] Through the above technical solutions, this application achieves integrated operation of multi-dimensional precise positioning and preliminary inspection during battery gripping, avoiding repetitive positioning errors caused by insufficient degrees of freedom in the mechanism. The overlapping movement of the dual horizontal extension arms expands the single-machine working range, reducing the equipment footprint and the frequency of layout adjustments. The synchronous execution of the contact monitoring module 706 and the gripping action shortens the inspection process cycle and reduces the risk of secondary damage caused by manual intervention. The lateral rotation function of the mechanical gripping unit 705 allows it to adapt to the battery placement angle at different workstations, improving the system's compatibility with different production lines.

[0056] Furthermore, such as Figure 1 and Figure 2As shown, the gripping and testing mechanism 700 places the qualified batteries into the loading station 7, and then rotates them sequentially to the tab testing mechanism 4 and the airtightness testing fixture mechanism 5 for testing. An airtightness testing unit 3 is connected to one side of the airtightness testing fixture mechanism 5.

[0057] The loading station 7 is a positioning platform for receiving qualified batteries, which can be implemented using a support platform with elastic positioning pins. A mechanical positioning structure ensures the battery's positional accuracy during rotation. The tab detection mechanism 4 is a device for detecting the shape of the battery tabs, which can be implemented using optical sensors combined with image processing algorithms to determine the welding quality of the tabs through non-contact measurement. The airtightness testing fixture mechanism 5 is a device for fixing the battery and applying sealing pressure, which can be implemented using a combination of pneumatic grippers and a sealing cavity. A pressure sensor monitors the sealing status. The airtightness testing unit 3 is a testing module independent of the mechanical moving parts, which can be implemented using a differential pressure leak detector. It is connected to the sealing cavity of the airtightness testing fixture mechanism 5 via an air pipe to detect pressure changes.

[0058] Specifically, after the gripping and inspection mechanism 700 places the battery with completed surface inspection onto the loading station 7, the rotary table 601 drives the battery to rotate clockwise. When the rotary table 601 rotates to the first station, the tab inspection mechanism 4 performs a three-dimensional scan of the battery tabs using an optical sensor to obtain data on tab height and flatness. After completing the tab inspection, the rotary table 601 continues to rotate to the second station, where the pneumatic grippers of the airtightness inspection fixture mechanism 5 close to form a sealed cavity. The airtightness inspection unit 3 injects inspection gas into the cavity and monitors the pressure change curve. During the inspection process, the airtightness inspection unit 3 remains physically isolated from the mechanical moving parts to avoid vibration interference with the inspection data.

[0059] Through the above technical solution, this application integrates surface defect detection, electrode tab detection, and airtightness detection processes. The battery completes multiple tests within a single fixture, avoiding positioning errors and physical damage caused by manual handling. The rotating, segmented detection layout allows for spatial separation of each detection process while maintaining temporal continuity. Detection data is automatically recorded and compared by the control system 11, forming a complete quality traceability chain. The independently designed airtightness detection unit 3 effectively isolates mechanical vibration interference, ensuring the stability and reliability of the airtightness detection data.

[0060] Furthermore, such as Figure 1 and Figure 2 As shown, the bearing mechanism 100 includes a first bearing platform 101 and a second bearing platform 102, and a multi-channel transmission mechanism 500 is disposed between the first bearing platform 101 and the second bearing platform 102.

[0061] The first support platform 101 is a basic platform used to support the guiding mechanism and the displacement mechanism 400. It can be implemented using a metal frame structure, and its surface can be equipped with transmission rails to support the movement of the battery carrier 1. The second support platform 102 is a support structure that forms a spatial layered relationship with the first support platform 101. It can be constructed using the same material as the first support platform 101 and is used to fix the pivot-based intelligent inspection mechanism 600 and form the inspection station layout. The multi-channel transmission mechanism 500 is a material transfer device set between the two support platforms. It can be implemented using a conveyor belt system with independent channels. Each channel can correspond to the battery transfer requirements of different inspection stages, and physical isolation ensures that the various processes do not interfere with each other.

[0062] Specifically, the first support platform 101 and the second support platform 102 are arranged in parallel to form an upper and lower layer or front and back layer layout, with a multi-channel transmission mechanism 500 embedded in the gap between them. After the battery carrier 1 moves to the guide mechanism via the transmission track of the first support platform 101, the displacement mechanism 400 transfers the battery to the corresponding channel of the multi-channel transmission mechanism 500. The multi-channel transmission mechanism 500 sequentially transports the battery to the pivot-region intelligent inspection mechanism 600 of the second support platform 102 according to the inspection process. During this process, different inspection procedures can be carried out in parallel through the spatial separation of the two support platforms. The multi-channel transmission mechanism 500, through its independent channel design, allows the battery after surface inspection to directly enter the tab inspection or airtightness inspection station without manual transfer.

[0063] Through the above technical solution, this application achieves continuous integration of the battery testing process, eliminating the material transfer links caused by the traditional multi-equipment layout. The dual-bearing platform structure provides independent operating space for different testing stations, and the multi-channel transmission mechanism 500 ensures automatic flow of batteries between processes, reducing efficiency loss and the risk of secondary damage caused by manual intervention.

[0064] Furthermore, such as Figure 3 , Figure 6 and Figure 7 As shown, the pivot-domain intelligent inspection mechanism 600 includes a rotary table 601. A position sensor 604 facing the loading station 7 is arranged in the middle of the upper part of the rotary table 601. A reinforcing structure 6 is provided on one side of both the tab detection mechanism 4 and the airtightness detection fixture mechanism 5. The bottom of the reinforcing structure 6 is connected to the rotary table 601. A rotating component 602 is arranged in the center of the bottom of the rotary table 601. A support component 603 is arranged around the rotating component 602 and below the rotary table 601. A pulley is provided at one end of the support component 603 that contacts the rotary table 601. The other end of the support component 603 is connected to the bearing mechanism 100. A power source is also arranged below the rotary table 601. The power source provides power to the rotating component 602, causing the pivot-domain intelligent inspection mechanism 600 to rotate.

[0065] The rotary table 601 is a rotatable platform that supports multiple inspection stations. It can be implemented using a metal disc structure, and station switching is achieved through rotation. The position sensor 604 is a device used to detect whether the loading station 7 is in a state of waiting to receive batteries. It can be implemented using a photoelectric sensor to ensure the accurate starting position of the battery entering the rotary table 601. The reinforcing structure 6 is a supporting component connecting the inspection station and the rotary table 601. It can be implemented using an L-shaped steel frame to enhance the vibration resistance of the inspection station during rotation. The rotating component 602 is the core component that drives the rotation of the rotary table 601. It can be implemented using a combination of a servo motor and a reducer to provide controllable power to the rotary table 601. The support component 603 is an auxiliary component that maintains the balance of the rotary table 601. It can be implemented using a column structure with rollers, and the contact between the rollers and the rotary table 601 reduces frictional resistance. The power source is a device that provides power to the rotating component 602. It can be implemented using a stepper motor to achieve precise control of the rotation angle.

[0066] Specifically, when the battery is placed at loading station 7, position sensor 604 triggers a signal, and rotary table 601 begins to rotate. The tab detection mechanism 4 and the airtightness testing fixture mechanism 5 are fixed to the rotary table 601 via a reinforcing structure 6, maintaining structural stability during rotation. Rotating component 602 drives the rotary table 601 to rotate under the power source, and support component 603 reduces friction between the rotary table 601 and the supporting mechanism 100 via pulleys, ensuring smooth rotation. When the rotary table 601 rotates to a preset angle, the tab detection mechanism 4 or the airtightness testing fixture mechanism 5 respectively tests the battery, enabling multi-station switching without manual handling.

[0067] Through the above technical solutions, this application achieves the integration and automation of multi-station inspection, reducing the number of battery transfers and manual operations, shortening the inspection cycle, and reducing the risk of secondary damage caused by manual operation. The cooperation between the rotary table 601 and the position sensor 604 ensures the accuracy of station switching, and the enhanced combination design of the structure 6 and the support 603 improves the stability of the inspection process, thereby adapting to the needs of mass production.

[0068] Furthermore, such as Figure 4 As shown, a first transmission mechanism 200 for transmitting the battery carrier 1 to the guide mechanism is provided on the first support platform 101. The first transmission mechanism 200 includes fixed frames 201 on both sides above the first transmission mechanism 200, a connecting rod 202 between the fixed frames 201, a transmission component 203 on the surface of the connecting rod 202, a clamping component 302 that can slide on the guide shaft 301, and the clamping height of the clamping component 302 can be adjusted up and down.

[0069] The first transmission mechanism 200 is a mechanical structure used to transport the battery carrier 1 from its initial position to the guide mechanism. It can be implemented using a conveyor belt or chain drive structure. A rigid support frame is formed by the fixed frame 201 and the connecting rod 202, ensuring the stability of the transmission component 203 during transmission and preventing the battery carrier 1 from shifting. The transmission component 203 is the part that directly contacts and drives the battery carrier 1. It can be implemented using rollers or a conveyor belt structure, achieving automated transmission of the battery carrier 1 through continuous operation. The clamping component 302 is a mechanical device used to fix the battery carrier 1. It can be implemented using a pneumatic gripper or an electric clamp. Its lateral position is adjusted by sliding on the guide shaft 301, and in conjunction with the vertical height adjustment mechanism, it adapts to the clamping requirements of battery carriers 1 of different sizes.

[0070] Specifically, the fixed frame 201 and the connecting rod 202 form a stable support structure. The transmission component 203 continuously rotates on the surface of the connecting rod 202, conveying the battery carrier 1 along the first support platform 101 to the direction of the guide mechanism. The clamping component 302 slides laterally on the guide shaft 301 to both sides of the battery carrier 1, and is pressed down to the appropriate position by the height adjustment mechanism. After clamping the battery carrier 1, it moves it below the displacement mechanism 400. The continuous conveying of the transmission component 203 and the dynamic adjustment of the clamping component 302 combine to realize the automated transmission of the battery carrier 1 without manual handling. At the same time, the multi-directional adjustment function of the clamping component 302 is compatible with different carrier sizes.

[0071] Through the above technical solution, this application solves the problems of low transmission efficiency and the need for manual intervention in the battery carrier 1 during the testing process. By combining mechanical transmission and adaptive clamping, the automation level of the transmission link is improved, while the stable delivery and precise positioning of battery carriers 1 of different sizes are compatible, reducing the frequency of manual adjustment and adapting to the needs of mass production.

[0072] Furthermore, such as Figure 4 As shown, the displacement mechanism 400 includes a support base fixed on the first support platform 101. A transverse guide post 401 is provided on the support base. A guide member 402 and a longitudinal telescopic mechanism 403 are sequentially provided on the side of the transverse guide post 401 facing the battery carrier 1. The guide member 402 allows the longitudinal telescopic mechanism 403 to move on the transverse guide post 401. A clamping arm 404 is provided below the longitudinal telescopic mechanism 403.

[0073] The support base refers to the base structure rigidly connected to the first bearing platform 101. It can be implemented using a welded or bolted metal frame, providing a stable mounting foundation for the transverse guide post 401 and preventing positioning misalignment due to vibration during displacement. The transverse guide post 401 is a linear guide rail extending horizontally, typically a hardened cylindrical metal rod. It provides a linear guide reference for the horizontal movement of the clamping arm 404, ensuring the linear accuracy of the battery gripping path. The guide component 402 is a sliding assembly that cooperates with the transverse guide post 401, typically a slider structure with ball bearings. It allows the longitudinal telescopic mechanism 403 to move horizontally along the transverse guide post 401 through low-friction sliding. The longitudinal telescopic mechanism 403 is a drive device with vertical movement capability, typically a cylinder or electric push rod. It adjusts the vertical height of the clamping arm 404 through telescopic movement to accommodate the loading and unloading needs of battery carriers 1 of different specifications. The gripping arm 404 refers to the end effector, which can be implemented by a pneumatic gripper or an electromagnetic chuck. Its function is to grip the battery through opening and closing actions, and to achieve precise displacement in three-dimensional space through linkage with the transverse guide post 401 and the longitudinal telescopic mechanism 403.

[0074] Specifically, when the battery carrier 1 is transported below the displacement mechanism 400, the support base maintains the stability of the transverse guide post 401 through a rigid connection. The guide member 402 drives the longitudinal telescopic mechanism 403 to move horizontally along the transverse guide post 401 to directly above the battery carrier 1. The longitudinal telescopic mechanism 403 extends the clamping arm 404 downward to a preset height, closes to grip the battery, and then retracts vertically. Subsequently, the guide member 402 moves horizontally along the transverse guide post 401 to above the multi-channel transmission mechanism 500. The longitudinal telescopic mechanism 403 lowers the clamping arm 404 again to the entrance position of the transmission channel, and the clamping arm 404 releases the battery to complete the transfer. During this process, the linear guiding characteristics of the transverse guide post 401 eliminate directional deviations from manual operation, the vertical stroke control of the longitudinal telescopic mechanism 403 ensures the consistency of the battery gripping and releasing height, and the sliding engagement of the guide member 402 achieves precise positioning for horizontal movement.

[0075] Through the above technical solution, this application achieves fully automated and precise battery transfer from the carrier to the multi-channel transmission mechanism 500, eliminating process interruptions caused by manual intervention. Mechanized three-dimensional positioning control standardizes the battery transfer path, avoiding positioning deviations caused by visual judgment errors during manual operation. Simultaneously, programmed control of clamping force and movement speed ensures that the battery surface is not mechanically damaged during transfer. This technical solution integrates previously scattered manual operation steps into a continuous automated process, seamlessly connecting the surface defect detection process with subsequent tab detection and airtightness detection, resulting in improved overall inspection efficiency that matches the demands of mass production.

[0076] Furthermore, such as Figure 5 As shown, the multi-channel transmission mechanism 500 includes a transmission platform 501, with a transmission belt surrounding the transmission platform 501. A multi-channel limiting member 502 is provided above the transmission platform 501, and the channels of the multi-channel limiting member 502 are used for transmitting batteries. A portal frame is provided on the multi-channel limiting member 502, which is connected to the base of the transmission mechanism at the bottom and to the multi-channel limiting member 502 at the top, so that the multi-channel limiting member 502 is suspended above the transmission platform 501. A distance sensor 503 is provided on the top of the portal frame provided on the multi-channel limiting member 502 for detecting the position of the battery in the channel.

[0077] The transmission platform 501 is the basic platform supporting battery transmission. It can be a planar structure made of metal or polymer materials, with a continuous transmission surface formed by a transmission belt covering its surface, providing a stable movement path for the batteries. The multi-channel limiting component 502 is a guide structure with multiple independently separated channels, which can be a U-shaped groove structure formed by injection molding or machining. Physical separation prevents interference between batteries in each channel. The portal frame is the supporting structure spanning above the transmission platform 501. It can be a frame formed by welding or bolting aluminum alloy profiles. Rigidly connected to the base below and elastically connected to the multi-channel limiting component 502 above, it forms a suspended layout to buffer vibrations during transmission. The distance sensor 503 is a non-contact position detection device, which can be a laser rangefinder or ultrasonic sensor. Vertically installed on the top of the portal frame, it monitors the movement of the batteries in the channels in real time.

[0078] Specifically, the transmission belt circulates on the surface of the transmission table 501, driving the battery forward along the partitioned channels of the multi-channel limiting member 502. The portal frame is fixed to the bottom of the transmission mechanism via a base, and its top is connected to the multi-channel limiting member 502 via an elastic connector, allowing the multi-channel limiting member 502 to have a small displacement space in the vertical direction. When there are tolerances in the battery dimensions, the channel gap can be adaptively adjusted through deformation. Distance sensors 503 are distributed at preset intervals on the top of the portal frame, each corresponding to one transmission channel. When a battery is detected to be out of position, the transmission belt is triggered to pause or decelerate, simultaneously driving the correction mechanism to fine-tune the battery position. Through the combined effect of mechanical limiting and dynamic monitoring, the battery is ensured to remain centered in the channel during transmission.

[0079] Through the above technical solution, this application solves the problem of low detection efficiency caused by inaccurate battery positioning during multi-channel transmission, and avoids jamming or offset phenomena caused by complex transmission structures. The physical separation of the multi-channel limiting component 502 and the floating design of the gate-shaped frame effectively control the battery movement trajectory, and the real-time monitoring of the distance sensor 503 ensures the accuracy of transmission position, thereby improving the grasping accuracy and overall detection efficiency of subsequent detection stations.

[0080] Furthermore, such as Figure 1 and Figure 2 As shown, a first transmission mechanism 200 for transmitting the battery carrier 1 to the guide mechanism is provided on the first support platform 101. The first transmission mechanism 200 includes fixed frames 201 on both sides above the first transmission mechanism 200, a connecting rod 202 between the fixed frames 201, and a transmission component 203 on the surface of the connecting rod 202. The clamping component 302 can slide on the guide shaft 301, and the clamping height of the clamping component 302 can be adjusted up and down.

[0081] The first transmission mechanism 200 is a mechanical structure used to transfer the battery carrier 1 from its initial position to the guide mechanism. It can be implemented using chain or belt drive. A stable frame is formed by the rigid connection between the fixed frame 201 and the connecting rod 202, preventing structural deformation during transmission. The clamping component 302 adjusts its clamping height using a lifting mechanism that engages a threaded rod and a slide rail. Specifically, a servo motor can drive the screw to rotate, controlling the vertical displacement of the clamping component 302 to accommodate battery carriers 1 of different heights. The transmission component 203 is the driving component that directly contacts the battery carrier 1. It can be implemented using rubber rollers or a gear structure, using evenly distributed friction or meshing force to propel the carrier smoothly.

[0082] Specifically, when the battery carrier 1 enters the first transmission mechanism 200, the rigid frame formed by the fixed frame 201 and the connecting rod 202 keeps the transmission component 203 running smoothly. The transmission component 203 pushes the carrier to move in a predetermined direction by rolling or meshing. The clamping component 302 slides laterally on the guide shaft 301, dynamically adjusting the clamping distance according to the width of the carrier, and simultaneously adjusting the clamping height through the lifting mechanism to ensure that the clamping surface always maintains perpendicular contact with the side wall of the carrier. The transmission component 203 on the surface of the connecting rod 202 forms continuous contact with the bottom of the carrier during the driving process, preventing the carrier from shifting due to excessive local pressure.

[0083] In some specific embodiments, the fixed frame 201 can be made of aluminum alloy profile, the spacing of the connecting rods 202 can be set to a range of 200-500 mm, and the surface of the transmission component 203 can be covered with an anti-slip rubber layer. The lifting mechanism of the clamping component 302 can be equipped with a position sensor 604, which automatically triggers an adjustment action when a change in the height of the vehicle is detected.

[0084] Through the above technical solution, this application solves the problem of jamming caused by structural instability during the transmission of battery carrier 1, eliminates the need for manual adjustment of clamping parameters, and adapts to the transmission requirements of carriers of different sizes through an automated adjustment mechanism, significantly improving the continuous operation capability of the production line and the versatility of the equipment.

[0085] Furthermore, such as Figure 2 and Figure 4 As shown, a first transmission mechanism 200 for transmitting the battery carrier 1 to the guide mechanism is provided on the first support platform 101. The first transmission mechanism 200 includes fixed frames 201 on both sides above the first transmission mechanism 200, a connecting rod 202 between the fixed frames 201, and a transmission component 203 on the surface of the connecting rod 202. The clamping component 302 can slide on the guide shaft 301, and the clamping height of the clamping component 302 can be adjusted up and down.

[0086] The fixed frame 201 refers to the support structure set on both sides of the first transmission mechanism 200. It can be implemented using a metal frame or a high-strength plastic frame, providing a stable mounting base for the connecting rod 202 and preventing structural deformation during transmission. The connecting rod 202 is a rigid rod that laterally connects the fixed frames 201 on both sides. It can be made of stainless steel or aluminum alloy, maintaining the running trajectory of the transmission component 203 and preventing the battery carrier 1 from deviating from the predetermined path during transmission. The transmission component 203 is a power transmission component set on the surface of the connecting rod 202, which can be implemented using rollers, belts, or chains. It directly contacts and propels the battery carrier 1, improving transmission efficiency. The sliding design of the clamping component 302 on the guide shaft 301 refers to the lateral position adjustment achieved through a guide rail or groove structure. It can be implemented using linear bearings in conjunction with a servo motor drive, dynamically adapting to the clamping requirements of battery carriers 1 of different sizes. The vertical adjustment function of the clamping component 302 refers to the vertical height change achieved by means of a hydraulic cylinder, screw mechanism or electric push rod. Specifically, it can be achieved by using a stepper motor to drive the screw lifting mechanism, which is used to adapt to battery carriers 1 of different heights and reduce manual adjustment operations.

[0087] Specifically, the first transmission mechanism 200 forms a rigid support frame with the connecting rod 202 via the fixed frame 201, ensuring that the transmission component 203 maintains a stable running trajectory when transmitting the battery carrier 1, preventing the battery carrier 1 from shifting or jamming due to structural deformation. The transmission component 203 acts directly on the bottom of the battery carrier 1, smoothly conveying it to the area of ​​the guide mechanism through continuous or intermittent movement. The sliding capability of the clamping component 302 on the guide shaft 301 allows it to move laterally to the optimal clamping point according to the actual position of the battery carrier 1, while the vertical adjustment function changes the clamping height through the lifting mechanism, ensuring compatibility with different models of battery carrier 1. Thus, the transmission process of the battery carrier 1 does not require manual intervention for position calibration, and the clamping action and transmission process are automated and coordinated.

[0088] Example 2

[0089] Based on Example 1, the difference lies in that this example provides an optimized implementation scheme for the visual inspection module 9. The programmable light-emitting unit of the polarization light source module adds a third group of lasers with a center wavelength of 1550 nm and an output power of 200 milliwatts. This wavelength can effectively penetrate the surface oxide layer and directly detect the deformation information of the substrate material. The polarization control unit is correspondingly upgraded by adding a 1 / 4-wave plate optimized for the 1550 nm wavelength band, enabling the illumination light field to generate a complete polarization state combination at the 1550 nm wavelength band.

[0090] The near-infrared intensifier camera in the multi-channel polarization imaging module has been replaced with a model that extends the spectral response range, covering 850 nm to 1600 nm. The beam-splitting film of the beam-splitting prism assembly has been redesigned, increasing reflectivity to 98 percent at 1550 nm. The polarization camera remains unchanged, but an external filter wheel has been added to suppress ambient stray light when acquiring data at the 660 nm wavelength.

[0091] The polarization parameter calculation in the spectral polarization fusion processing module has been expanded to three spectral bands. The dimension of the multispectral polarization parameter matrix has become 3x3, with the newly added third row corresponding to the polarization degree, polarization angle, and intensity of the 1550 nm band. The image registration algorithm introduces a multi-resolution pyramid matching strategy, improving registration accuracy to the sub-pixel level.

[0092] The deep convolutional neural network model structure of the oxidation feature decoupling module has been adjusted to have four feature extraction branches. The newly added fourth branch specifically handles features sensitive to matrix deformation at the 1550 nm band. The number of neurons in the fully connected layer of the feature fusion layer has been increased to 1024 and 512 to handle higher-dimensional feature splicing. The training sample database has been expanded accordingly to include 5000 samples with 1550 nm band labeled data. The weight coefficients of the loss function have been readjusted to 0.5 for physical deformation loss, 0.3 for oxidation artifact loss, and 0.2 for orthogonality constraint loss to balance the feature contribution after the introduction of the new band.

[0093] The support vector machine classifier in the defect decision output module has been upgraded to a multi-kernel learning version, combining linear and radial basis kernels to improve its ability to partition high-dimensional feature spaces. A dynamic adjustment mechanism for the decision threshold has been introduced, with the threshold adaptively calculated based on the real-time detected average surface reflectance. When the average reflectance is higher than 80%, the defect threshold is increased from 0.95 to 0.98 to reduce the false alarm rate for highly reflective surfaces.

[0094] The end-to-end latency tolerance of the system's real-time processing framework has been relaxed to 800 milliseconds to accommodate the data processing load of the newly added bands. The frame rate remains at 10 Hz, but a new 1550 nm band illumination and imaging sequence has been added within the data acquisition cycle. A mid-infrared transmission window, made of zinc selenide, has been added to the optical path in the mechanical structure, with a transmittance greater than 90% at 1550 nm. The electrical control system's processor has been upgraded to a higher clock speed model, and the memory capacity has been expanded to 64 gigabytes to ensure high-speed caching and processing of three-dimensional band data.

[0095] In the description of this invention, it should be understood that the terms "front and back," "left and right," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or component referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention. All contents not described in detail in the specification belong to the prior art known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited. Conventional equipment can be used.

[0096] Of course, those skilled in the art should understand that in this technical solution, the term "a" should be understood as "at least one" or "one or more," that is, in one embodiment, the number of an element can be one, while in another embodiment, the number of the element can be multiple. The term "a" should not be construed as a limitation on the quantity.

[0097] The above description is merely a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art under the technical guidance of the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A surface defect detection device for battery casing processing, comprising a support mechanism (100). Its features are: The carrying mechanism (100) is provided with at least a guiding mechanism, a displacement mechanism (400), a multi-channel transmission mechanism (500), a gripping detection mechanism (700), a vision detection module (9) and a pivot-domain intelligent detection mechanism (600). The carrying mechanism (100) is provided with multiple sets of battery carriers (1) and a control system (11) is provided at one end of the carrying mechanism (100). The guiding mechanism includes a guide shaft (301), and a clamping component (302) is provided on one side of the guide shaft (301). The battery carrier (1) is moved below the displacement mechanism (400) by the guiding mechanism. The displacement mechanism (400) is located on one side of the clamping component (302), and a multi-channel transmission mechanism (500) is located on one side of the displacement mechanism (400). The battery in the battery carrier (1) is sent to the multi-channel transmission mechanism (500) through the displacement mechanism (400). The multi-channel transmission mechanism (500) is provided with a gripping detection mechanism (700) that can be displaced at multiple angles and rotated in the horizontal direction on the side away from the displacement mechanism (400). A vision detection module (9) is provided below the gripping detection mechanism (700). When the gripping detection mechanism (700) grips the battery at the tail end of the multi-channel transmission mechanism (500), it moves it close to the vision detection module (9) and achieves visual defect detection by rotating it 360°. The gripping and detection mechanism (700) has a pivot-domain intelligent inspection mechanism (600) on one side relative to the visual inspection module (9). The pivot-domain intelligent inspection mechanism (600) includes at least a tab detection mechanism (4), an airtightness detection fixture mechanism (5), a loading station (7), and a unloading station (8). A defective product placement table (10) is provided on one side of the pivot-domain intelligent inspection mechanism (600). The gripping and detection mechanism (700) places batteries with surface defects into the defective product placement table (10) and batteries without surface defects into the unloading station (8).

2. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The visual inspection module (9) includes: a polarization light source module, used to project an illumination light field with a specific polarization state and spectral combination onto the surface of the battery casing to be tested; A multi-channel polarization imaging module is used to simultaneously acquire intensity images of reflected light from the battery casing surface in at least two different polarization directions and at least two different spectral bands; The spectral polarization fusion processing module is used to register and fuse the acquired multi-channel image data and calculate the polarization parameter matrix of each pixel in the multispectral dimension. The oxidation feature decoupling module, based on the polarization parameter matrix, separates the feature vectors representing oxidation artifacts and the feature vectors representing real physical deformation through a preset oxide layer spectral polarization response model; the defect decision output module performs pattern recognition and threshold judgment on the decoupled physical deformation feature vectors and outputs defect category and location information. The polarization light source module includes a programmable light-emitting unit and a polarization control unit; the programmable light-emitting unit consists of two sets of narrowband semiconductor lasers with different center wavelengths. The polarization control unit is located in the light output path of the light-emitting unit and consists of two independently rotatable linear polarizers and one quarter-wave plate. The multi-channel polarization imaging module includes a beam splitter prism group, a polarization camera, and a near-infrared intensifier camera; the beam splitter prism group divides the incident light into visible light and near-infrared channels according to the spectrum; The visible light channel connects to the polarization camera; the near-infrared channel connects to the near-infrared enhancement camera; the spectral polarization fusion processing module performs image preprocessing, multimodal data registration, and polarization parameter calculation. The oxidation feature decoupling module embeds a trained deep convolutional neural network model; The defect decision output module normalizes the physical deformation feature vector and then inputs it into the support vector machine classifier.

3. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The gripping and detection mechanism (700) includes a base control unit (701). A first horizontal extension arm (702) that can move laterally is provided above the base control unit (701). A second horizontal extension arm (703) that can move laterally is provided above the first horizontal extension arm (702) and on the side away from the base control unit (701). A longitudinal telescopic unit (704) is provided on the side of the second horizontal extension arm (703) away from the first horizontal extension arm (702). A mechanical gripping unit (705) that can rotate laterally is provided below the longitudinal telescopic unit (704).

4. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The gripping and testing mechanism (700) places the qualified batteries into the loading station (7), and then rotates them sequentially to the tab testing mechanism (4) and the airtightness testing fixture mechanism (5) for testing. An airtightness testing unit (3) is connected to one side of the airtightness testing fixture mechanism (5).

5. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The bearing mechanism (100) includes a first bearing platform (101) and a second bearing platform (102), and the multi-channel transmission mechanism (500) is disposed between the first bearing platform (101) and the second bearing platform (102).

6. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The pivot-domain intelligent inspection mechanism (600) includes a rotary table (601), and a position sensor (604) facing the loading station (7) is provided in the middle of the upper part of the rotary table (601). The electrode detection mechanism (4) and the airtightness detection fixture mechanism (5) are both provided with a reinforcing structure (6) on one side, and the bottom of the reinforcing structure (6) is connected to the rotary table (601). A rotating component (602) is provided at the center of the bottom of the rotating platform (601). A support component (603) is provided around the rotating component (602) and below the rotating platform (601). A pulley is provided at one end of the support component (603) that contacts the rotating platform (601). The other end of the support component (603) is connected to the bearing mechanism (100). A power source is also provided below the rotating platform (601), which provides power to the rotating component (602) to rotate the pivot-domain intelligent inspection mechanism (600).

7. The surface defect detection device for battery casing processing according to claim 5, characterized in that: The first carrier platform (101) is provided with a first transmission mechanism (200) for transmitting the battery carrier (1) to the guide mechanism. The first transmission mechanism (200) includes fixed frames (201) on both sides above the first transmission mechanism (200), and a connecting rod (202) is provided between the fixed frames (201). A transmission component (203) is provided on the surface of the connecting rod (202). The clamping component (302) can slide on the guide shaft (301), and the clamping height of the clamping component (302) can be adjusted up and down.

8. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The displacement mechanism (400) includes a support base fixed on the first support platform (101), and a transverse guide post (401) is provided on the support base. A guide member (402) and a longitudinal telescopic mechanism (403) are sequentially provided on the side of the transverse guide post (401) facing the battery carrier (1). The guide (402) allows the longitudinal telescopic mechanism (403) to move on the transverse guide post (401); A clamping arm (404) is provided below the longitudinal telescopic mechanism (403).

9. The surface defect detection device for battery casing processing according to claim 1, characterized in that: The multi-channel transmission mechanism (500) includes a transmission platform (501), which is surrounded by a transmission belt. A multi-channel limiting member (502) is provided above the transmission platform (501), and the channel of the multi-channel limiting member (502) is used to transmit the battery. The multi-channel limiting member (502) is provided with a gate-shaped frame, which is connected to the base of the transmission mechanism (500) at the bottom and to the multi-channel limiting member (502) at the top, so that the multi-channel limiting member (502) is suspended above the transmission table (501). A distance sensor (503) is provided on the top of the portal frame of the multi-channel limiting member (502) to detect the position of the battery in the channel.

10. The surface defect detection device for battery casing processing according to claim 2, characterized in that: The deep convolutional neural network model has a network structure that includes three feature extraction branches and one feature fusion layer. The first branch is used to extract features that are sensitive to surface physical deformation at the 660 nm wavelength. The second branch is used to extract features that are sensitive to oxide layer thickness changes at the 940 nm wavelength. The third branch is used to analyze the difference in polarization state changes between the two wavelengths. feature The fusion layer concatenates the outputs of the three branches and regresses the oxidation artifact feature vector and physical deformation feature vector through two fully connected layers.

Images