Bare cell defect detection method and device

Abstract

The invention belongs to the technical field of battery detection, and particularly relates to a naked battery cell defect detection method and device, and the method comprises the steps: 1, in a pole piece processing flow, in an interval from the completion of a baking process to the start of a winding process, obtaining a pole piece surface image after the completion of each process or a part of processes in the interval; step 2, applying a preset alternating current to the pole piece to obtain an infrared thermal distribution image of the pole piece; 3, applying a preset direct-current high voltage between the positive electrode and the negative electrode of the naked battery cell to obtain an infrared thermal distribution image of the naked battery cell; 4, acquiring an X-ray image of the naked battery cell; and 5, inputting the obtained surface image of the pole piece, the obtained infrared heat distribution image of the pole piece, the obtained infrared heat distribution image of the naked battery cell and the obtained X-ray image of the naked battery cell into a preset AI analysis model based on deep learning, and outputting a defect detection result of the naked battery cell through the defect analysis model. According to the invention, the defect detection and identification precision and capability are improved.

Description

Technical Field

[0001] This invention belongs to the field of battery testing technology, specifically relating to a method and apparatus for detecting defects in bare battery cells. Background Technology

[0002] Electric vehicles are powered by batteries composed of cells, and bare cells are the core component of cell production. During the production of bare cells, due to the influence of production processes and process control conditions, bare cells are prone to various defects such as wrinkles and foreign particles. These defects not only lead to a decrease in cell performance but may also induce cell safety risks. Therefore, it is necessary to inspect bare cells during the production process to eliminate defective products.

[0003] Currently, the industry commonly uses visual recognition and Hi-pot (withstand voltage test) insulation testing for defect detection. These technologies can only identify electrode sheets and bare cells with relatively obvious defect characteristics, leading to some products with inconspicuous defects entering subsequent stages. These products are prone to defects after long-term use, potentially causing safety issues. In summary, comprehensive inspection of bare cell defects helps to further improve cell product quality, reduce the probability of problems in the market, and is of great significance to the rapidly developing electric vehicle and power battery industry.

[0004] Existing technologies propose acquiring image data using visible light as a primary light source and infrared light as a secondary light source to achieve cell defect detection. This approach utilizes light sources with different penetrability to acquire cell image data, and in addition to detecting defects, it can further distinguish between on-film and under-film defects. Although this approach involves the detection of internal defects in bare cells, its core focus is on differentiating between on-film and under-film defects, without further improving the detection capability for internal defects in bare cells. It can only directly identify defects through the image itself and cannot detect some minute defects with inconspicuous features.

[0005] Existing technologies also propose obtaining process battery parameters through multiple rounds of cyclic testing and comparing them with ideal data curves to determine one or more defects in the power battery based on abnormal process parameters. This method collects parameters through cyclic testing and compares them with the cell process parameters to determine cell defects, enabling defect detection after the cells are produced, effectively supplementing non-destructive defect detection methods after the cells leave the factory. However, in the cell production process, this method is independent of the cell production process, which can seriously affect the production rhythm; at the same time, this method cannot effectively detect minute defects that require long-term cycling to be exposed.

[0006] In summary, existing defect detection methods in the bare cell production process have the following problems: The detection targets or methods are limited and cannot cover different types of defects. They can only perform defect detection on a single process in the bare cell production process where defect detection is feasible.

[0007] It cannot detect potential electrical defects, such as micro-short circuits and internal impurities, which may only be exposed during long-term use. These defects are difficult to detect through conventional testing methods at the bare cell stage. For example, Hi-pot leakage current testing can only detect obvious defects such as internal short circuits or foreign object punctures.

[0008] Therefore, it is necessary to develop a new method and device for detecting defects in bare battery cells. Summary of the Invention

[0009] The purpose of this invention is to provide a method and apparatus for detecting defects in bare battery cells, which can improve the accuracy and capability of defect detection and identification.

[0010] To achieve the above objectives, the technical solution adopted by the present invention is as follows: In a first aspect, the present invention provides a method for detecting defects in bare battery cells, comprising the following steps: Step 1. In the electrode processing flow, during the interval between the completion of the baking process and the start of the winding process, for each process or part of the process within the interval, an image of the electrode surface is obtained after the process is completed. Step 2. Apply a preset alternating current to the electrode and acquire an infrared thermal distribution image of the electrode; Step 3. Apply a preset DC high voltage between the positive and negative terminals of the bare cell to obtain an infrared thermal distribution image of the bare cell; Step 4. Obtain X-ray images of the bare battery cell; Step 5. Input the acquired electrode surface image, the infrared thermal distribution image of the electrode, the infrared thermal distribution image of the bare cell, and the X-ray image of the bare cell into a preset deep learning-based AI analysis model, and output the defect detection results of the bare cell through the defect analysis model.

[0011] In one possible implementation, step 1 includes a rolling process and a die-cutting process within the interval between the baking process and the start of the winding process, acquiring surface images of the electrode sheet after baking, rolling, and die-cutting, respectively. This method, through continuous detection after each process, can accurately locate the process steps where defects occur, facilitating timely parameter adjustments and preventing defects from being transmitted to the bare cell during the process.

[0012] In one possible implementation, step 1 involves acquiring an image of the electrode surface using a high-resolution linear scan camera. This image is used for subsequent identification of surface defects on the electrode, a crucial step in ensuring the quality of the subsequent bare cells.

[0013] In one possible implementation, in step 2, a preset alternating current is applied to the electrode before the electrode winding process; the preset alternating current is a micro-current excitation method. The abnormal heating caused by the micro-current excitation (i.e., the preset alternating current) can expose potential electrical defects in the electrode, such as localized high resistance.

[0014] One possible implementation involves using a micro-current excitation method for the preset AC current. The current value of the micro-current excitation satisfies the following conditions: it applies the maximum applicable current value without damaging the normal electrode structure and performance, and the infrared thermal distribution image of the electrode is acquired in real time using a thermal imager. This maximizes the current excitation effect while ensuring electrode safety, making the thermal effect at defects (such as localized conductivity abnormalities) more significant. Combined with real-time acquisition by the thermal imager, this improves the identification accuracy of defects in the infrared thermal distribution image of the electrode.

[0015] In one possible implementation, in step 3, after the bare cell is cold-pressed or hot-pressed, a preset DC high voltage is applied between the positive and negative terminals of the bare cell. The applied preset DC high voltage must meet the following requirements: it must be able to excite a recognizable thermal effect at the micro-short circuit or foreign object connection point between the electrodes inside the bare cell, without damaging a normal bare cell. This ensures that a recognizable thermal signal can be excited at the micro-short circuit or foreign object connection point inside the bare cell through high voltage, while avoiding damage to qualified cells, thus achieving effective detection of latent electrical defects inside the bare cell.

[0016] In one possible implementation, in step 4, after cold pressing or hot pressing of the bare battery cell, an X-ray image of the bare battery cell is acquired using a CT scanner. CT scans have three-dimensional imaging capabilities, can penetrate the internal structure of the bare battery cell, and clearly present internal defects such as electrode layer alignment and foreign object embedding depth. Compared with ordinary X-rays, it can provide more comprehensive internal structural information and improve the accuracy of internal defect detection.

[0017] In one possible implementation, in step 5, the defect analysis model is a deep learning-based AI analysis model, which is trained through the following steps: Dataset construction: Collect normal images and images of different types of defects of electrode sheets and bare cells after different processes, and record the manual judgment results of the corresponding images simultaneously; Model Training and Validation: The dataset is divided into training and validation sets to train the defect analysis model until it meets preset requirements. This enables the defect analysis model to learn the characteristic patterns of different processes and defects. The model trained on a large amount of labeled data can achieve automated defect identification, improving detection efficiency and consistency.

[0018] One possible implementation is to adjust model parameters based on manual review results during model training and validation to complete the validation process. Manual review can correct biases in the training of the defect analysis model, and adjusting parameters through feedback can improve the model's ability to identify complex defects (such as defects with fuzzy features), ensuring the reliability of the defect analysis model's output results.

[0019] In one possible implementation, step 5 further includes model iterative optimization: updating the parameters of the defect analysis model based on the differences between the judgment results of the defect analysis model in production and the results of manual review. This enables the defect analysis model to continuously learn new defect types or characteristic changes that occur in production, adapt to process fluctuations in actual production, maintain high detection accuracy over a long period, and extend the effective lifespan of the defect analysis model.

[0020] Secondly, the bare cell defect detection device of the present invention includes a memory and a controller. The memory stores a computer-readable program. When the computer-readable program is invoked by the controller, it can execute the bare cell defect detection method as described in the present invention.

[0021] The present invention has the following unexpected technical effects: (1) Based on conventional detection of large-scale electrical defects, this invention can detect smaller-scale electrical defects inside electrodes and bare cells by applying micro-current excitation and high-voltage direct current, thereby improving the accuracy and capability of defect detection and identification.

[0022] (2) Based on the detection of defects in bare cells, this invention conducts comprehensive detection of bare cells in the process through different processes, multiple states and multiple detection methods, covering electrode surface defects, electrode internal electrical defects, bare cell internal electrical defects and bare cell internal mechanical damage defects. Attached Figure Description

[0023] Figure 1 This is a flowchart of the bare cell defect detection method described in the embodiments of this application; Figure 2 This is a flowchart of the bare cell defect detection device described in the embodiments of this application; In the diagram: 1. Memory, 2. Controller. Detailed Implementation

[0024] The embodiments of the present invention will be described below with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention and not for limiting the scope of protection of the present invention.

[0025] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a specific manner to facilitate understanding.

[0026] like Figure 1 As shown, a method for detecting defects in bare battery cells includes the following steps: Step 1: In the electrode processing flow, during the interval between the completion of the baking process and the start of the winding process, for each process within the interval, an image of the electrode surface is obtained after the process is completed.

[0027] For example, after each process is completed, an image of the electrode surface is acquired using a high-resolution linear scan camera.

[0028] Step 2: After completing Step 1, in the electrode processing flow, before the electrode winding process, apply a preset AC current to the electrode to obtain an infrared thermal distribution image of the electrode.

[0029] For example, a safe alternating current is applied to the electrode using microcurrent excitation, and an infrared thermal distribution image of the electrode is acquired using a thermal imager.

[0030] Step 3: Apply a preset DC high voltage between the positive and negative terminals of the bare cell to obtain an infrared thermal distribution image of the bare cell.

[0031] For example, after cold pressing or hot pressing of the bare cell, a preset DC high voltage is applied between the positive and negative electrodes of the bare cell, and an infrared thermal distribution image of the bare cell is obtained by a thermal imager.

[0032] Step 4: Obtain X-ray images of the bare battery cells.

[0033] For example, after cold pressing or hot pressing of the bare battery cell, a high-resolution X-ray image of the bare battery cell is obtained by a CT scanner.

[0034] Step 5: Input the image data obtained in Steps 1 to 4 into the preset defect analysis model, and output the defect detection results through the defect analysis model.

[0035] In one possible embodiment, in step 1, before the electrode is finally used for bare cell assembly, it generally needs to undergo key processes such as rolling (to increase the density of active material) and die-cutting (to cut into specific sizes and shapes). Each process may introduce new surface defects or expose previously hidden defects. For example, the baking process may cause cracking, blistering (due to solvent residue or uneven drying), or peeling of the active material layer on the electrode surface. The rolling process may cause pressure marks, creases, or uneven active material layer thickness and material loss due to uneven pressure. The die-cutting process may produce burrs, tears, gaps, or surface scratches caused by mechanical contact during the die-cutting process. Therefore, in the bare cell production process, acquiring images of the electrode surface after baking, rolling, and die-cutting for subsequent identification of surface defects is an important step in ensuring the quality of the subsequent bare cells. This method, through continuous detection after each process, can accurately locate the process steps that cause defects, facilitating timely parameter adjustments and preventing defects from being transmitted to the bare cells along the process.

[0036] In one possible embodiment, in step 2, the electrode is in the state before winding; wherein, the applied preset AC current must be within a safe range, that is, it will not have an adverse effect on the normal electrode. The abnormal heating caused by microcurrent excitation (i.e. preset AC current) can expose potential electrical defects such as local large resistance of the electrode.

[0037] The current used for microcurrent excitation should meet the following requirements: to apply the maximum current to the electrode to achieve the best exposure effect without damaging the normal electrode structure and performance; the higher the current, the more pronounced the abnormal heating caused by current excitation. Infrared thermal distribution images of the electrode should be acquired in real time using a thermal imager.

[0038] In one possible embodiment, in step 3, after the bare cell has been cold-pressed or hot-pressed, abnormal heating caused by a preset DC high voltage exposes electrical defects such as micro-short circuits and foreign objects between the cell electrodes. The applied preset DC high voltage must meet the following requirements: it must be able to induce a identifiable thermal effect at the junction of micro-short circuits or foreign objects between the electrodes in the bare cell, without damaging the normal bare cell.

[0039] A preset DC high voltage is applied to the positive and negative terminals of the bare cell. If a micro short circuit occurs between the electrodes inside the bare cell or if there are foreign objects connecting or puncturing, a complete circuit will be formed. The thermal effect generated by the current will be significantly different at the defect location, and more heat will accumulate. Finally, an abnormal heating point is detected by a thermal imager, thereby confirming whether there are any tiny electrical defects that cannot be detected by Hi-pot.

[0040] In one possible embodiment, in step 4, the X-ray image obtained by the CT scanner is a tomographic image, and the scanning frequency and speed are determined according to the actual production cycle. The tomographic image can detect whether there are mechanical damage defects such as electrode breakage inside the bare cell after cold pressing or hot pressing.

[0041] In one possible embodiment, in step 5, the defect analysis model employs an existing deep learning AI analysis model, and the training process of the defect analysis model is as follows: Construct a training dataset: Collect normal images and images of different types of defects obtained in steps 1-4, and record the manual judgment results of the corresponding images. Model training and validation: The dataset is divided into training and validation sets. The defect analysis model is trained using the area and depth of the electrode defects as the core quantifiable features. The model parameters are adjusted based on the results of manual review and confirmation until the defect analysis model meets the preset requirements (e.g., until the defect identification accuracy of the defect analysis model is not less than 99% and the error rate is not higher than 0.5%), or until the consistency between the judgment result of the defect analysis model on the validation set and the manual judgment result reaches 98%).

[0042] Model Production Application: During the electrode production process, the surface image of the electrode in step 1, the infrared thermal distribution image of the electrode in step 2, the infrared thermal distribution image of the bare cell in step 3, and the X-ray image of the bare cell in step 4 are collected in real time and input into the trained and validated defect analysis model. The defect analysis model uses real-time algorithms to compare and judge the quantifiable features such as the area and depth of defects in the electrode and bare cell in the images, and outputs the defect judgment results.

[0043] Model Iteration and Optimization: Based on the difference between the judgment results of the defect analysis model and the results of manual review during the production process, the training dataset is continuously supplemented to update the parameters of the defect analysis model, thereby improving the accuracy of the defect analysis model's judgment and its feature comparison capability.

[0044] This method achieves comprehensive inspection of the bare cell manufacturing process through different processes (post-baking to pre-winding, post-cold or hot pressing), multiple states (electrode state, bare cell state), and multiple inspection methods (surface image acquisition, infrared thermal distribution imaging, X-ray imaging). It covers four core defect categories: electrode surface defects, internal electrode electrical defects, internal bare cell electrical defects, and internal bare cell mechanical damage defects. This method not only solves the problem of relying on a single inspection method but also overcomes the limitation of incomplete defect type coverage.

[0045] This method, building upon conventional detection methods (macroscopic, large-scale electrical defects), employs an innovative combination of micro-current excitation and thermal imaging, and high-voltage direct current and thermal imaging, to effectively detect even smaller-scale electrical defects (i.e., potential defects such as micro-short circuits and internal impurities mentioned in the background technology) within electrodes and bare cells. This significantly improves the accuracy and capability of defect detection and identification. In essence, it precisely addresses the core pain point of existing technologies' inability to detect potential minute electrical defects, filling a gap in conventional detection methods.

[0046] like Figure 2 As shown in the embodiments of this application, a bare cell defect detection device includes a memory 1 and a controller 2. The memory 1 stores a computer-readable program. When the computer-readable program is called by the controller 2, it can execute the bare cell defect detection method as described in the embodiments of this application.

[0047] The above embodiments are preferred embodiments of the present invention, but the embodiments of the present invention are not limited to the above embodiments. Any changes, modifications, substitutions, combinations, or simplifications made without departing from the spirit and principle of the present invention shall be considered equivalent substitutions and shall be included within the protection scope of the present invention.

Claims

1. A method of detecting defects in a bare cell, characterized by, The method comprises the following steps: Step 1. In the interval from the completion of the baking process to the start of the winding process in the electrode sheet processing flow, for each process or part of the processes in the interval, an electrode sheet surface image is obtained after the process is completed; Step 2. A preset alternating current is applied to the electrode sheet, and an infrared thermal distribution image of the electrode sheet is obtained; Step 3. A preset direct current high voltage is applied between the positive and negative electrodes of the bare cell, and an infrared thermal distribution image of the bare cell is obtained; Step 4. An X-ray image of the bare cell is obtained; Step 5. At least one of the obtained electrode sheet surface image, the infrared thermal distribution image of the electrode sheet, the infrared thermal distribution image of the bare cell, and the X-ray image of the bare cell is input into a preset AI analysis model based on deep learning, and a defect detection result of the bare cell is output by the defect analysis model.

2. The method of claim 1, wherein the method further comprises: In step 1, the processes in the interval from the completion of the baking process to the start of the winding process include the rolling process and the die cutting process, and electrode sheet surface images after baking, rolling, and die cutting are obtained respectively.

3. The method of claim 2, wherein the method further comprises: In step 1, the electrode sheet surface image is obtained by a high-resolution line array scanning camera.

4. The method of claim 1, wherein, In step 2, a preset alternating current is applied to the electrode sheet before the winding process.

5. The method of claim 4, wherein the method further comprises: The preset alternating current adopts a micro-current excitation mode; The current value of the micro-current excitation satisfies: under the premise of not damaging the normal electrode sheet structure and performance, the maximum applicable current value is applied, and the infrared thermal distribution image of the electrode sheet is collected in real time by a thermal imager.

6. The method of detecting defects in bare cells of claim 1, wherein, In step 3, a preset direct current high voltage is applied between the positive and negative electrodes of the bare cell after cold pressing or hot pressing of the bare cell; The applied preset direct current high voltage needs to satisfy: it can stimulate the micro-short circuit between the electrode sheets in the bare cell or the identifiable thermal effect at the foreign matter overlap, and does not damage the normal bare cell.

7. The method of claim 1, wherein the method further comprises: In step 4, the X-ray image of the bare cell is obtained by a CT scanner after cold pressing or hot pressing of the bare cell.

8. The method of claim 1, wherein, In step 5, the defect analysis model is an AI analysis model based on deep learning, and the defect analysis model is trained by the following steps: Constructing a data set: collecting normal images and different types of defect images of electrode sheets and bare cells after different processes, and synchronously recording the artificial judgment results of the corresponding images; Model training and verification: dividing the data set into a training set and a verification set, training the defect analysis model until the defect analysis model meets the preset requirements.

9. The method of claim 8, wherein the method further comprises: In model training and verification, the model parameters are adjusted in combination with the artificial review results to complete verification.

10. An apparatus for detecting defects of a bare cell, characterized by, The method comprises a memory (1) and a controller (2), the memory (1) stores a computer readable program, and the computer readable program can execute the bare cell defect detection method of any one of claims 1 to 9 when called by the controller (2).

Images