CN-121998911-A - Defect detection method and device for secondary battery, terminal equipment and storage medium

CN121998911ACN 121998911 ACN121998911 ACN 121998911ACN-121998911-A

Abstract

The invention discloses a defect detection method, a defect detection system and a storage medium for a secondary battery, and relates to the technical field of intelligent manufacturing of batteries. The method comprises the steps of accurately associating winding alignment (X-ray) and lug welding spot (visible light) data on a high-speed production line to a unique cell ID through a time axis synchronous triggering mechanism, and constructing a multi-mode feature sequence. The method focuses on the critical morphology feature of the physical quantity at the qualified boundary, inputs the critical morphology feature into a deep learning model containing physical mechanism constraint, analyzes the nonlinear coupling relation between morphology and electrochemical performance, and outputs failure probability distribution. Based on the risk identification result, physical interception is performed before the capacity-division formation, or an adaptive drift compensation signal is transmitted to the front-end manufacturing equipment. The invention solves the problem of high-speed data alignment, and realizes the spanning from the detection of microscopic morphology to the prediction of electrochemical performance and the process closed-loop self-healing.

Inventors

LIANG SHAOJIAN
LI XIANG
LIN SHIPING
CHEN MINGYUAN
ZHANG SHANGQING

Assignees

广东金晟新能源股份有限公司

Dates

Publication Date: 20260508
Application Date: 20251229

Claims (10)

1. A defect detection method of a secondary battery, comprising: Acquiring first imaging detection data of a winding alignment degree station and second imaging detection data of a tab welding spot detection station of a secondary battery cell to be detected on a high-speed production line through a time axis synchronous trigger mechanism, wherein the high-speed production line is a production line with the production rate being greater than a preset threshold value; Constructing a multi-mode physical morphological feature sequence of the secondary battery cell to be detected according to the first imaging detection data, the second imaging detection data and the unique digital identification of the secondary battery cell to be detected; Extracting critical morphology features representing performance fluctuation of the battery cell from the multi-mode physical morphology feature sequence, wherein the critical morphology features are morphology features in which the difference value of the boundary value between a physical magnitude and a preset qualified interval is smaller than or equal to a preset boundary threshold value and the physical magnitude is in a normal distribution preset interval; The critical morphology features are input into a pre-trained performance association model, the performance association model is constructed based on a nonlinear coupling relation between morphology features in historical production data and electrical performance truth values after a preset cycle life stage, and the performance association model comprises a deep learning framework of physical mechanism constraint; calculating electrical property prediction probability distribution of the secondary battery cell to be tested according to the performance correlation model, and identifying the electrochemical failure risk of the secondary battery cell to be tested after a preset cycle life stage based on the probability distribution; And outputting a sorting interception instruction or performing self-adaptive drift compensation on the execution parameters of equipment corresponding to the front-end manufacturing procedure by a production control system before the secondary battery cell to be tested enters the capacity-dividing procedure according to the risk identification result.
2. The method for detecting defects of a secondary battery according to claim 1, wherein the acquiring of the first imaging detection data of the winding alignment station and the second imaging detection data of the tab welding spot detection station on the high-speed production line by the time axis synchronous triggering mechanism specifically comprises: installing an incremental encoder on a main conveyor drive shaft of a high-speed production line; the real-time motion control card based on the FPGA is used for receiving the pulse of the incremental encoder and comprises a first trigger port and a second trigger port, wherein the first trigger port is connected with an X-ray detector of a winding alignment station, and the second trigger port is connected with an industrial camera of a welding station; When a feeding sensor detects that a secondary battery cell to be detected enters a conveyor belt, generating a unique digital identification of the secondary battery cell to be detected, and writing the unique digital identification into a first address of a register; Updating the corresponding logic coordinates of the secondary battery cell to be tested in the register in real time according to the increment of the pulse of the incremental encoder; When the logic coordinates are matched with preset shooting points of the X-ray detector, the real-time motion control card outputs a first hardware trigger signal through the first trigger port; acquiring perspective image data of the X-ray detector, marking a current time stamp and the unique digital mark, and taking the perspective image data as first imaging detection data; When the logic coordinates are matched with the preset shooting points of the industrial camera, the real-time motion control card outputs a second hardware trigger signal through the second trigger port, acquires the surface appearance image data of the industrial camera, marks the current timestamp and the unique digital mark, and takes the surface appearance image data as second imaging detection data.
3. The method for detecting defects of a secondary battery according to claim 1, wherein the constructing a multi-mode physical form feature sequence of the secondary battery cell to be detected according to the first imaging detection data, the second imaging detection data and the unique digital identification of the secondary battery cell to be detected specifically comprises: Calculating a rotation and translation matrix by taking the edge of a battery shell of the secondary battery cell as a datum point, and calibrating the first imaging detection data and the second imaging detection data to a standard coordinate system; Performing ROI clipping and normalization processing on the first imaging detection data and the second imaging detection data; Splicing the first imaging detection data and the second imaging detection data in a channel dimension to form a multi-channel tensor; And through a full-connection layer, encoding a winding tension value and a welding power value which are related to the unique digital identification into vectors, and splicing the vectors with the multi-channel tensor to obtain the multi-mode physical form characteristic sequence of the secondary battery cell to be detected.
4. The method for detecting defects in a secondary battery according to claim 3, wherein said ROI cutting and normalizing process is performed on said first imaging detection data and said second imaging detection data, specifically comprising: Scanning along the gray gradient direction of the first imaging detection data, positioning the boundary line between the positive plate and the negative plate, expanding preset pixel widths to two sides by taking the boundary line as a central axis, and cutting out a first rectangular image only comprising Overhang areas; Positioning a welding geometric center of the second imaging detection data by using template matching, and cutting out a second rectangular image covering the whole heat affected zone by taking the welding geometric center as a rectangular center; And scaling the first rectangular image and the second rectangular image to the same size by adopting a bilinear interpolation algorithm to obtain the processed first imaging detection data and the processed second imaging detection data.
5. The method for detecting defects of a secondary battery according to claim 1, wherein the physical mechanism constraint deep learning architecture specifically refers to: and introducing an electrochemical mechanism regularization term into the loss function of the deep learning network, wherein the electrochemical mechanism regularization term comprises potential constraint based on a Nernst equation, impedance constraint based on an equivalent circuit model and concentration distribution constraint based on ion diffusion dynamics.
6. The method for detecting defects of a secondary battery according to claim 1, wherein the identifying the electrochemical failure risk of the secondary battery cell to be detected after a predetermined cycle life period based on the electrical performance prediction probability distribution specifically comprises: And when the accumulated probability exceeding the failure threshold value in the electrical performance prediction probability distribution is larger than the preset risk tolerance, judging that the electrical performance prediction probability distribution has potential failure risk, wherein the electrical performance prediction probability distribution comprises a confidence interval of capacity retention rate and an increase probability density function of direct current internal resistance.
7. The method for detecting defects of a secondary battery according to claim 1, wherein the outputting a sorting interception command by a production control system or performing adaptive drift compensation on an execution parameter of equipment corresponding to a front end manufacturing procedure according to a risk identification result before the secondary battery cell to be detected enters a capacity-division process step, specifically comprises: Determining the risk level of the secondary battery cell according to the electrical performance prediction probability distribution, wherein the secondary battery cell is in a first-level risk if the accumulated probability of the failure threshold is greater than a first failure threshold, the secondary battery cell is in a second-level risk if the accumulated probability of the failure threshold is greater than or equal to a second failure threshold and less than or equal to the first failure threshold, and the secondary battery cell is in a no-risk level if the accumulated probability of the failure threshold is less than the second failure threshold; outputting a sorting interception instruction through a production control system if the secondary battery cell is at primary risk; If the secondary battery cells are in secondary risk, calculating the average offset of the winding alignment degree of the latest multiple battery cells in real time, generating a position compensation signal aiming at a deviation correction controller of a winding machine or a tension adjustment signal aiming at a tension controller, monitoring the shrinkage trend of the nugget area or the penetration of a welding spot in real time, and generating a laser power adjustment signal or a welding pulse width modulation signal aiming at a laser welding machine.
8. A defect detecting device for a secondary battery, comprising: The imaging module is used for acquiring first imaging detection data of a winding alignment degree station and second imaging detection data of a tab welding spot detection station of a secondary battery cell to be detected on a high-speed production line through a time axis synchronous trigger mechanism, wherein the high-speed production line is a production line with the production rate being greater than a preset threshold value; The sequence module is used for constructing a multi-mode physical morphological feature sequence of the secondary battery cell to be detected according to the first imaging detection data, the second imaging detection data and the unique digital identification of the secondary battery cell to be detected; The extraction module is used for extracting critical morphology features representing the performance fluctuation of the battery cell from the multi-mode physical morphology feature sequence, wherein the critical morphology features are morphology features in which the difference value of the boundary value between the physical magnitude and a preset qualified interval is smaller than or equal to a preset boundary threshold value and the physical magnitude is in a normal distribution preset interval; The system comprises an input module, a performance correlation model, a control module and a control module, wherein the input module is used for inputting the critical morphological characteristics into a pre-trained performance correlation model, the performance correlation model is constructed based on a nonlinear coupling relation between the morphological characteristics in historical production data and an electrical performance truth value after a preset cycle life stage, and the performance correlation model comprises a deep learning framework of physical mechanism constraint; the detection module is used for calculating the electrical property prediction probability distribution of the secondary battery cell to be detected according to the performance correlation model, and identifying the electrochemical failure risk of the secondary battery cell to be detected after a preset cycle life stage based on the probability distribution; And the adjusting module is used for outputting a sorting interception instruction or carrying out self-adaptive drift compensation on the execution parameters of equipment corresponding to the front-end manufacturing procedure through the production control system before the secondary battery cell to be tested enters the capacity-dividing procedure according to the risk identification result.
9. A terminal device comprising a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, the processor implementing a defect detection method of a secondary battery according to any one of claims 1 to 7 when the computer program is executed.
10. A computer-readable storage medium, characterized in that it stores a computer program which, when executed by a processor, implements a defect detection method of a secondary battery according to any one of claims 1 to 7.

Description

Defect detection method and device for secondary battery, terminal equipment and storage medium Technical Field The present invention relates to the field of battery detection, and in particular, to a method and apparatus for detecting defects of a secondary battery, a terminal device, and a storage medium. Background With the explosive growth of new energy automobiles and electrochemical energy storage industry, the manufacturing process of secondary batteries (especially lithium ion power batteries) is evolving towards extremely high efficiency (the productivity beat is more than or equal to 200 PPM) and extreme manufacturing precision. However, existing high-speed production line quality control techniques still present significant technical bottlenecks. First, the problem of "spatiotemporal misalignment" of data correlations is serious. Because processes such as winding, welding and the like are large in physical span and the running speed of a conveyor belt is extremely high, traditional detection equipment is often self-contained, and X-ray perspective data and visible light appearance data acquired by different stations are accurately bound to a digital Identification (ID) of the same cell in a millisecond time window, so that a data tracing chain is broken. Second, conventional AOI detection logic has "survivor bias". In the prior art, a hard threshold judgment (Pass/Fail) based on a fixed rule is mostly adopted, and only dominant waste with severely exceeding size can be removed. For "sub-healthy" cells (i.e., with critical topographical features) that have approached a critical boundary or exhibit an abnormal statistical distribution, although within acceptable ranges for those physical parameters, conventional algorithms are not identifiable. In the subsequent long-period charge-discharge cycle, the battery cells are very easy to evolve into electrochemical failures such as lithium precipitation, internal short circuit and the like. Disclosure of Invention The embodiment of the invention provides a defect detection method, a defect detection device, terminal equipment and a storage medium for a secondary battery, which realize accurate synchronization of multi-mode data on a high-speed production line, deeply pre-detect electrochemical performance risks based on a physical mechanism and can perform process self-adaptive closed-loop compensation. To achieve the above object, a first aspect of an embodiment of the present invention provides a defect detection method of a secondary battery, including: Acquiring first imaging detection data of a winding alignment degree station and second imaging detection data of a tab welding spot detection station of a secondary battery cell to be detected on a high-speed production line through a time axis synchronous trigger mechanism, wherein the high-speed production line is a production line with the production rate being greater than a preset threshold value; Constructing a multi-mode physical morphological feature sequence of the secondary battery cell to be detected according to the first imaging detection data, the second imaging detection data and the unique digital identification of the secondary battery cell to be detected; Extracting critical morphology features representing performance fluctuation of the battery cell from the multi-mode physical morphology feature sequence, wherein the critical morphology features are morphology features in which the difference value of the boundary value between a physical magnitude and a preset qualified interval is smaller than or equal to a preset boundary threshold value and the physical magnitude is in a normal distribution preset interval; The critical morphology features are input into a pre-trained performance association model, the performance association model is constructed based on a nonlinear coupling relation between morphology features in historical production data and electrical performance truth values after a preset cycle life stage, and the performance association model comprises a deep learning framework of physical mechanism constraint; calculating electrical property prediction probability distribution of the secondary battery cell to be tested according to the performance correlation model, and identifying the electrochemical failure risk of the secondary battery cell to be tested after a preset cycle life stage based on the probability distribution; And outputting a sorting interception instruction or performing self-adaptive drift compensation on the execution parameters of equipment corresponding to the front-end manufacturing procedure by a production control system before the secondary battery cell to be tested enters the capacity-dividing procedure according to the risk identification result. In a possible implementation manner of the first aspect, the acquiring, by using a time axis synchronization trigger mechanism, first imaging detection data of a winding alignment station and second