Determining non-uniformities in electrode fabrication for battery cells
Through the combination of hyperspectral imaging technology and KI engine, the problem of difficult to monitor the unevenness of the electrode coating of lithium-ion batteries is solved, real-time detection and labeling of electrode webs is realized, the quality and reliability of the battery cell are improved, and the quality and reliability of the battery cell is improved. Supports automated manufacturing.
Patent Information
- Application Number
- CN202380038606.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Priority Date
- 2022-05-06
- Filing Date
- 2023-04-21
- Publication Date
- 2024-12-17
AI Technical Summary
During the manufacturing process of lithium-ion battery cells, the unevenness of the electrode coating is difficult to monitor in real time, resulting in low battery quality and low reliability. The existing technology can only find defects after production or after years of use, and lack of effectiveness. online detection method.
Hyperspectral imaging technology uses a line scanning camera to capture images on electrode webs, obtain deviations in chemical composition and particle size distribution, mark uneven locations, and perform computer-aided analysis through KI engine and deep learning to achieve online detection and closed loop feedback.
Real-time identification and labeling of electrode web inhomogeneity is achieved, online evaluation and closed-loop feedback is provided, the quality and reliability of battery cells is improved, the IT load is reduced during the production process, and automated manufacturing is supported.
Smart Images
Figure CN119156716A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to an apparatus and method for hyperspectral imaging of electrode webs in the manufacture of electrodes for battery cells. Background Art
[0002] Lithium-ion rechargeable batteries, also referred to below as lithium-ion batteries, are used as energy storage devices in mobile and stationary applications due to their high power and energy density.
[0003] Lithium-ion batteries typically include multiple battery cells. Battery cells, especially lithium-ion battery cells, include many layers. These layers typically include anodes, cathodes, separators and other components. These layers can be constructed as a stack or as a winding.
[0004] The electrode typically comprises a metal foil, which in particular comprises copper and / or aluminum, which is coated with an active material. As the active material, a lithium-containing paste (also referred to as slurry) is typically applied. The foil and the coating each have a thickness of several micrometers. As a result, deviations of a few micrometers in the thickness of the coating or deviations in material properties, in particular material composition, have a negative impact on the quality of the electrode. As a result, battery cells of lower quality can be manufactured disadvantageously in the case of irregular coatings. In addition, it is disadvantageous that reliable operation of the battery cells is not ensured.
[0005] Currently, in the prior art, defective coatings are often only detected after the entire production process of the battery cell has been completed within the scope of a so-called final offline test. In some cases, defective coatings are also detected only after the battery cell has been in operation for many years.
[0006] During production, the battery material used for the production of battery cells is deposited or applied as a viscous medium (coating paste, English: slurry) onto an electrode web.
[0007] The continuous determination of material parameters (such as composition, uniformity of distribution of ingredients, humidity, ...) along a (continuously long) manufacturing route / product web (such as battery electrodes, paper webs, steel, ...) or batch manufacturing equipment (such as cookies, biscuits, brake linings, ...) is of critical significance to the result.
[0008] In particular, the production of battery cells (here in the case of lithium-ion batteries) is nowadays very often characterized by an experimental process (that is, rather a mixture of craftsmanship and technology rather than a scientific or fully understood industrial process). In this case, the pasty active material is applied to a long electrode web on both sides (typically continuously), dried and processed. In this case, the process parameters are often stored as a cooking recipe (e.g. drying at y degrees for x hours) and are hardly monitored by means of sensors.
[0009] Since for a pasty mixture of NMC or graphite (as the main component) with a solvent and, if necessary, other components (such as conductive carbon black applied to the electrode web), the ingredients are mixed in a mixer, the so-called slurry, and the homogeneity and viscosity of the mixture are decisive for the properties of the subsequent battery cells and thus the properties of the subsequent battery as a whole, this homogeneity of the coating must be monitored unconditionally during the production process.
[0010] In the case of a suboptimally homogenized mixture, regions may occur when coating with more or less solvent (water, NMP). On the other hand, regions with different chemical compositions may also result. Furthermore, if the mixing process is subject to fluctuations, an inhomogeneous disaggregation of the solid particles may occur, that is, deviations of the particle size or particle size distribution from the target value may occur. This makes it possible to implement a closed-loop feedback loop between the detection of inhomogeneities and the slurry mixer.
[0011] During further processing, these regions can then lead to inhomogeneous properties of the produced electrode and / or cause short circuits and inefficiencies in the battery cells.
[0012] Until now, as sampling, some areas have been punched out of the produced web or complete sections have been extracted from the production. In the best case, the material thus extracted is then analyzed under a microscope. Deviations in the chemical composition of the layer or deviations in the particle size distribution can be confirmed offline (in laboratory testing) by X-ray methods. Summary of the invention
[0013] The object of the present invention is to specify a solution which improves the production of electrodes for battery cells.
[0014] The invention is derived from the features of the independent claim. Advantageous developments and embodiments are the subject matter of the dependent claims. Further features, possible applications and advantages of the invention are apparent from the following description.
[0015] One aspect of the present invention is directed to determining the local location of inhomogeneities in an electrode web during the production of coated electrodes for battery production.
[0016] The present invention claims a device for hyperspectral imaging of electrode webs in the manufacture of electrodes for battery cells. The device has:
[0017] At least one hyperspectral spectral unit having a line scan camera for imaging, wherein the spectral unit is arranged and configured to:
[0018] - recording a hyperspectral image at a predetermined position of the electrode fabrication of the electrode web moving forward;
[0019] - detecting inhomogeneities of the electrode web from the image, wherein the inhomogeneities are deviations in the chemical composition of the layer or deviations in the particle size distribution from a predefined target size; and
[0020] The local position of the inhomogeneities along the longitudinal direction of the electrode web is detected and protected (ie a position coding).
[0021] The electrode web is a carrier foil coated with a slurry. A line scan camera is known in English as a "Push-Broom-Scanner." "Protecting" is to be understood in a broad sense, also in the sense of "storing" or "marking."
[0022] The wavelength of the hyperspectral imaging is selected by a person skilled in the art depending on the inhomogeneities to be detected and the material used to form the slurry, ie also the coating of the electrode web.
[0023] For example, the inhomogeneity can be determined by comparison with a previously acquired reference image (=target size). The reference image can be acquired, for example, from a previously produced electrode web with a proven homogeneous structure or from averaging of images from the first meter of a just produced electrode web.
[0024] The device has a turning cylinder, via which the electrode web is guided, wherein the line scan camera records images at the location of the contact surface between the turning cylinder and the electrode web.
[0025] In one development, the device is designed to protect the local locations of the inhomogeneities by marking the electrode web. The markings can be found again, for example, when the electrode web is cut and used for sorting.
[0026] In one development, the device has an engraving unit which is designed to carry out the marking by engraving, preferably at the longitudinal edge of the electrode web. This can be done, for example, by a laser engraver.
[0027] In one development, the device has a memory unit which is designed to store the spectral signature of the inhomogeneity as a position code along the electrode web in such a way that the position of the inhomogeneity can be found using the spectral signature. The stored spectrum of the inhomogeneity can be found again, for example, when the electrode web is cut.
[0028] In one development, the device has a production management system to which the spectroscopic unit is connected and which is configured to trigger the spectroscopic unit for measuring. This allows for a subsequent local allocation of the measurements, for example during cutting and sorting of the electrode web.
[0029] In one development, the device has an evaluation and calculation unit at the location of the spectral unit, wherein the evaluation and calculation unit is designed to pre-compress the data of the hyperspectral imaging into vectors and assign them to local locations. This results in a measure for the lateral distribution of the material along the observed measuring line. This means that a vector of the inhomogeneity is compiled.
[0030] In a further development, the line scan camera is a SWIR line scan camera. This makes a cost-effective solution possible.
[0031] In one development, the device has a KI engine in the spectral unit, which is designed to determine the inhomogeneities with the aid of a computer.
[0032] In a refinement, the KI engine is trained with the aid of a deep learning method in order to extract inhomogeneities from the spectrum.
[0033] The invention also claims a method for detecting inhomogeneities in the production of electrodes for battery cells by means of hyperspectral imaging of an electrode web, the method comprising the following steps:
[0034] - recording a hyperspectral image at a predetermined position of the electrode fabrication of the electrode web moving forward;
[0035] - determining inhomogeneities of the electrode web from the spectrum of the image, wherein the inhomogeneities are deviations in the chemical composition of the layer or deviations in the particle size distribution from a predefined target size, and
[0036] - The local position of said inhomogeneities along the longitudinal direction of the electrode web is captured and preserved.
[0037] At the location of the contact surface between the turning cylinder, over which the electrode web is guided, and the electrode web, an image is recorded by a line scan camera.
[0038] In a development of the method, the local position of the inhomogeneity is protected by markings on the electrode web.
[0039] In a development of the method, the spectral signature of the inhomogeneity is stored as a position code along the electrode web in such a way that the position of the inhomogeneity can be found with the aid of the spectral signature.
[0040] In a development of the method, the inhomogeneities are detected with the aid of a KI engine in a computer-assisted manner.
[0041] The present invention provides the following advantages in particular:
[0042] - Measurement at the turning roller or above the supporting / stabilizing web section;
[0043] - Direct online evaluation allows closed-loop feedback with, for example, a dryer arranged upstream;
[0044] - Near-field precompression of the data reduces the load on the IT and enables physically "correct" evaluation (in the sense of real measured variables rather than "house numbers") through the accompanying semantic annotation;
[0045] - Provide semantic annotation;
[0046] -Integration into manufacturing automation;
[0047] - assigned to a location along the "product";
[0048] - traceability;
[0049] -Continuous measurement information;
[0050] - Non-interfering in the absence of sampling; and
[0051] - Continuous detection in the process providing a characterized 2D inhomogeneity indication suitable for identification / re-identification of web segments in subsequent process steps.
[0052] Further features and advantages of the invention become apparent from the following explanation of an exemplary embodiment with the aid of the schematic drawings. BRIEF DESCRIPTION OF THE DRAWINGS
[0053] In the attached figure:
[0054] Figure 1 The coating of a carrier foil is shown;
[0055] Figure 2 The drying of the coated carrier foil is shown;
[0056] Figure 3 Calendering of the dried carrier foil is shown;
[0057] Figure 4 A block diagram of a hyperspectral measurement device is shown; and
[0058] Figure 5 An extract of a hyperspectral image of an electrode web is shown. DETAILED DESCRIPTION
[0059] Figures 1 to 3 The production of electrodes for batteries is schematically and exemplarily shown. First, a carrier foil 1 is coated with a slurry 2 delivered from a reservoir 7 by means of an application tool 3, such as a slot nozzle, a doctor blade or an anilox roller ( Figure 1 ). Along the coating direction 4, either continuously or intermittently (in top view) Figure 5 The carrier foil 1 is coated with an intermittent coating 6).
[0060] The foil thickness (anode copper foil and cathode aluminum foil) fluctuates between 10 μm and 25 μm, depending on the cell design. The carrier foil 1 is coated in a roll-to-roll process over a width of up to 900 mm.
[0061] The coating of the foil top and foil bottom is carried out either sequentially or simultaneously, depending on the configuration of the production plant. The thickness of the coating 6 produced with the slurry 2 is detected in the still wet state using a wet thickness measuring device 8 .
[0062] In coating ( Figure 1 ), the coated carrier foil 1 (which can also be referred to as electrode web) is guided directly to a dryer 9 ( Figure 2 ). In the case of simultaneous double-sided coating, an overhead web dryer is used. The solvent (steam 10 via exhaust device 12) is extracted from the coating 6 by heat transport via air nozzles 11 and reused or transported to a heat utilization device.
[0063] The dryer length is decisive for the achievable throughput speed. The dryer 9 is divided into different temperature zones to achieve an individual temperature profile. After passing through the dryer, the dried electrode web (=coated carrier foil 1) is cooled to room temperature by means of a cooling device 13 with the aid of cooling rollers 14.
[0064] After calendering for 15 ( Figure 3 ), the double-sided coated carrier foil 1 is compacted by one or more rotating roller pairs 16. The foil is compacted by upper and lower rollers. The roller pairs 16 generate a precisely adjustable line pressure. A line pressure that is set too high results in a squeezing process and damages the electrode web. The cleanliness of the roller pairs 16 is crucial to prevent foreign particles from being squeezed into the base material.
[0065] The carrier foil 1 is conveyed through the calender 15 by means of a traction device 17. An infeed device 18 ensures clean entry from a storage drum 19 into the calender 15. After calendering, the electrode web is cleaned in a cleaning device 20 and stored as a "mother roll 21" for further processing into battery cells, such as cutting.
[0066] Advantageously, in drying ( Figure 2 ) or after calendering ( Figure 3 ) Hyperspectral measurements are then performed using a line scan camera 22 and an evaluation and calculation unit 23 arranged downstream.
[0067] Figure 4 The block diagram of a hyperspectral measuring device is shown, which has a spectral unit 27 with a line scan camera 22, which records a hyperspectral image of the coated carrier foil (=electrode web 26) at the location of the deflection roller 25. The line scan camera 22 is connected to an evaluation and calculation unit 23, which in turn is connected to a production management system 24 for the production of the electrode web. An engraving unit 28 is provided for marking the electrode web 26, for example at the locations of the detected inhomogeneities.
[0068] Hyperspectral imaging in the variant of a line scan camera 22 (=Push-Broom Scanner: line measurement of the spectrum with high lateral, ie spatial, resolution) is shown in Figure 4 As shown in FIG. 2 , it is preferably used at the deflection drum 25 so that
[0069] a) observing the stationary electrode web position;
[0070] b) obtaining synchronization with the relative movement of the electrode web;
[0071] c) enable location traceability; and
[0072] d) provide stable measurement conditions (robust against vibrations, dimmable, etc.)
[0073] In the SWIR (850-1700 nm), the following signatures are revealed during imaging (particularly inexpensively) which vary with the degree of homogeneity of the distribution of the starting material.
[0074] In the NIR (1200-2500 nm) still more specific bands are possible, but the price of spectroscopic equipment is high. From an economic point of view, the use of the SWIR is therefore advantageous.
[0075] The spectroscopic measuring system is advantageously connected to a production management system 24 (PLC, Edge, etc.) and receives from it a trigger for the measurement, which facilitates the subsequent local assignment of the measurement.
[0076] In this case, early processing of the data in the field close to the measuring point is advantageous. This reduces the load on the IT system, since high data rates occur due to high product speeds and can even block modern high-speed networks.
[0077] Therefore, the data set is already pre-compressed directly after the measurement and the dimension of the transverse distribution of the material along the observed measuring line, i.e. the vector of the inhomogeneity, is assigned to the web position. This information is already two to three orders of magnitude smaller than the actual measured raw data. (HIS frame: 640 x 224 pixels vs. evaluation result: 640 x 1 vector or just a scalar indicator)
[0078] Figure 5 The measurement points of an exemplary embodiment are shown. An excerpt from a hyperspectral image of an electrode web is shown in the left figure. Inhomogeneities are represented here by different gray levels. The diagram on the right shows the spectral evaluation for a pixel with the coordinates 50 / 20. For each position coordinate x, y in the left figure, there is the following spectrum, which describes the chemical properties of this image point (=pixel).
[0079] The measured spectrum can preferably be analyzed in two ways:
[0080] A) Supervised machine learning: A model is calibrated that either maps the spectrum to a target value (e.g. graphite content) or classifies the spectrum (e.g. pass, fail). Methods exist for this purpose, such as partial least squares, principal component regression, Gaussian processes or neural networks.
[0081] B) Unsupervised Machine Learning: Anomalies such as local inhomogeneities in the coating of the electrode web (e.g. differences in slurry deposition increasing at the outer edges, layer thickness variations) are identified. For this purpose, statistical features such as local differences or entropy can be calculated. As an alternative, a residual model can be compiled (e.g. low-rank PCA approximation, variational autoencoder neural network).
[0082] In both cases, the analysis is supplemented by preprocessing steps, for example by releasing the coating foreground from the substrate foil, filtering out the sensor information according to its degree of differential interpretation or increasing the contrast.
[0083] Although the present invention has been illustrated and described in detail by way of embodiments, the present invention is not restricted to the disclosed examples and other variations can be derived therefrom by those skilled in the art without departing from the scope of protection of the present invention.
[0084] List of reference numerals:
[0085] 1 Carrier foil
[0086] 2 Slurry
[0087] 3 Smudge Tool
[0088] 4 Coating direction
[0089] 5 Top view
[0090] 6 Intermittent coating
[0091] 7 Storage container for slurry 2
[0092] 8 Wet thickness measurement
[0093] 9 Dryer
[0094] 10 Steam
[0095] 11 Air nozzle
[0096] 12 Exhaust mechanism
[0097] 13 Cooling mechanism
[0098] 14 Cooling roller
[0099] 15 Calender
[0100] 16 Roller Pairs
[0101] 17 Traction mechanism
[0102] 18 Introducing institutions
[0103] 19 Storage drum
[0104] 20 Cleaning Agency
[0105] 21 Mother roll
[0106] 22-line scan camera
[0107] 23 Evaluation and calculation units
[0108] 23.1 Storage Unit
[0109] 24 Manufacturing Management System
[0110] 25 Steering roller
[0111] 26 Electrode web
[0112] 27 Spectral Unit
[0113] 28 engraving units
Claims
1. A device for hyperspectral imaging of electrode webs (26, 1) in the production of electrodes for battery cells, have: a) at least one hyperspectral spectral unit (27) having a line scan camera (21) for imaging, wherein the spectral unit (27) is arranged and configured to: - recording a hyperspectral image at a predetermined position of the electrode fabrication of the electrode web (26, 1) moving forward; - inhomogeneities of the electrode web (26, 1) are detected from the image, wherein the inhomogeneities are deviations in the chemical composition of the layer or deviations in the particle size distribution from a predetermined target size, and - capturing and protecting the local position of said inhomogeneities along the longitudinal direction of the electrode web (26, 1), and b) a turning roller (25) through which the electrode web (26, 1) is guided, wherein the line scan camera (21) records an image at the location of the contact surface between the turning roller (25) and the electrode web.
2. The device according to claim 1, The local position of the inhomogeneity is thereby protected by markings on the electrode web (26, 1).
3. The device according to claim 1 or 2, Features An engraving unit (28) is configured to carry out marking by engraving at the longitudinal edge of the electrode web (26, 1).
4. The device according to any one of the preceding claims, Features A memory unit (23.1) is designed to store a spectral signature of an inhomogeneity as a position code along the electrode web (25, 1) in such a way that the position of the inhomogeneity can be found with the aid of the spectral signature.
5. The device according to any one of the preceding claims, Features: A production management system (24), to which the spectroscopic unit (27) is connected and which is configured to trigger the spectroscopic unit (27) to perform a measurement.
6. The device according to any one of the preceding claims, Features: An evaluation and calculation unit (23) located at the location of the spectral unit (27), wherein the evaluation and calculation unit (23) is designed to compress the data of the spectral imaging into vectors and to assign them to local locations.
7. The device according to any one of the preceding claims, It is characterized in that The line scan camera (22) is a SWIR line scan camera.
8. The device according to any one of the preceding claims, The spectral unit (27) has a KI engine which is designed to determine the inhomogeneity with the aid of a computer.
9. The apparatus according to claim 8, wherein the KI engine is trained by means of a deep learning method for extracting inhomogeneities from optical spectra.
10. A method for detecting inhomogeneities in the production of electrodes for battery cells by means of hyperspectral imaging of an electrode web (26, 1), characterized in that: - recording a hyperspectral image at a predetermined position of the electrode fabrication of the electrode web (26, 1) moving forward; - determining inhomogeneities of the electrode web from the spectrum of the image, wherein inhomogeneities are deviations of the chemical composition of the layer or deviations of the particle size distribution from a predefined target size, and - capturing and protecting the local position of the inhomogeneity in the longitudinal direction of the electrode web (26, 1), wherein the line scan camera records an image at the position of the contact surface between a deflection roller (25) through which the electrode web is guided and the electrode web.
11. The method according to claim 10, The local location of the inhomogeneities is thereby protected by markings on the electrode web.
12. The method according to any one of claims 10 or 11, It is characterized in that The spectral signature of the inhomogeneity is recorded as a position code along the electrode web in such a way that the position of the inhomogeneity can be found with the aid of the spectral signature.
13. The method according to any one of claims 10 to 12, wherein the inhomogeneity is acquired with the aid of a KI engine with the aid of a computer.