Method and system for detecting an electrode defect
Patent Information
- Authority / Receiving Office
- EP · EP
- Patent Type
- Applications
- Current Assignee / Owner
- VERKOR SA
- Filing Date
- 2023-11-07
- Publication Date
- 2026-04-29
AI Technical Summary
Existing methods for detecting electrode faults during lithium-ion battery manufacturing, such as burrs on active electrode materials, are inadequate as they fail to identify internal defects and often result in unnecessary discarding of components, especially since they are typically detected at a later stage in the manufacturing process.
A method and system utilizing an inductive stimulator and thermal imager to heat and scan electrodes, generating thermographic data which is analyzed by a computer program with an extraction algorithm to detect defects, allowing for early identification and sorting of electrodes without disrupting the manufacturing process, using an eddy current probe for non-destructive testing and machine learning algorithms for efficient defect characterization.
Enables early detection of electrode faults, reducing the number of defective components discarded and ensuring battery quality by identifying surface and subsurface defects before assembly, thus preventing short circuits and potential fires.
Smart Images
Figure 1.1
Abstract
Description
[0001] Electrode defect detection method and system
[0002] Technical field of the invention
[0003] The invention relates to the technical field of electric batteries, in particular the invention relates to a method and a system for detecting a burr on an active electrode material during the manufacture of a lithium-ion cell.
[0004] Prior art
[0005] During the manufacture of an electrode intended for inclusion in a battery, rough edges or ridges may form on the electrode surface. This occurs because cutting operations are performed during the manufacture of the electrode. These rough edges or ridges are called burrs and have an impact on the quality and performance of the battery. A burr can pierce the separator and, consequently, bring two electrodes of opposite polarity (an anode and a cathode) into direct electrical contact. This results in a short circuit that can have serious consequences, such as the establishment of an electric arc and / or spontaneous combustion during use of the battery, or even a fire.
[0006] Due to the very nature of active materials and despite efforts to achieve good quality, clean cuts, the presence of burrs is often unavoidable. Therefore, the identification of surface and subsurface burrs is crucial in order to better detect the electrode with a defect so that it is not used in battery manufacturing.
[0007] Many attempts have been made to detect the presence of a defect during battery manufacturing. For example, patent application JP20150022094 discloses a method for inspecting the presence of a defect on an inner surface of a metal case for a lithium-ion battery, in which, after an electrode assembly and an electrolyte are inserted into a metal case, a surface of the metal case is pressed with a binding device, and the lithium-ion battery to be inspected is then heated. In order to inspect the presence of at least one defect on the inner surface of the metal case, a temperature distribution of the metal case is observed for a set period of time until the temperature of the battery returns to a normal state.
[0008] However, the method described in patent application JP20150022094 only identifies the defect on the metal casing and does not detect the defect of the components inside the battery. In particular, the method described in patent application JP20150022094 does not help to detect any electrode defects inside the battery. Moreover, the method described in patent application JP20150022094 only detects the defect after the battery is assembled at a later stage of the manufacturing process, which means that if a defect is detected on the metal casing, the entire battery and the components inside must be scrapped even if these components are not themselves defective.
[0009] General Description
[0010] The purpose of the present invention is to provide a defect detection method which can detect an electrode defect at an early stage of the manufacturing process without disrupting the battery manufacturing process while reducing the number of components in the battery likely to be discarded.
[0011] To this end, the present invention relates to a defect detection method for detecting a defect in an electrode intended to be included in a battery during the manufacture of the electrode, said detection method comprising the following steps:
[0012] - placing the electrode on a primary conveyor and circulating it by means of this primary conveyor through an inductive stimulator and a field of vision of a thermal imager;
[0013] - heating a surface of said electrode using the inductive stimulator;
[0014] - obtaining thermographic data from different positions on the electrode surface using the thermal imager at different times as the electrode moves on the primary conveyor, in order to assign the electrode several thermal image sequences;
[0015] - inserting the thermographic data into a computer program comprising an extraction algorithm, the extraction algorithm being configured to extract defect characteristics from the plurality of thermal image sequences;
[0016] - sorting the electrodes according to the extracted fault characteristics, such as: if the extracted fault characteristics satisfy a predefined selection criterion, the electrode is deemed functional and the electrode remains on the primary conveyor in order to be included in the battery, if the extracted fault characteristics do not satisfy the predefined selection criterion, the electrode is considered defective and the electrode is transferred to at least one secondary conveyor.
[0017] According to one feature, the inductive stimulator is configured to deliver a current at a variable operating frequency depending on the extracted fault characteristics.
[0018] According to one feature, the fault detection method further comprises a step of: - adjusting the operating frequency of the inductive stimulator as a function of the extracted fault characteristics in order to improve the detection of the electrode fault on the surface of the electrode.
[0019] According to one feature, the inductive stimulator is an eddy current probe.
[0020] According to one characteristic, the fault detection method further comprises a step of:
[0021] - formatting of thermographic data using a conditioning circuit prior to entering the thermographic data into the computer program.
[0022] According to one characteristic, the fault detection method comprises a step of adapting at least one parameter of the conditioning circuit as a function of the extracted fault characteristics.
[0023] According to one characteristic, the extraction algorithm is a machine learning algorithm, including an anomaly detection algorithm or a classification algorithm.
[0024] The invention further relates to a defect detection system for detecting an electrode defect during the manufacture of an electrode intended to be included in a battery, said defect detection system comprising:
[0025] - an inductive stimulator heating a surface of the electrode;
[0026] - a thermal imager;
[0027] - a primary conveyor on which the electrode is arranged and by means of which the electrode travels through an inductive stimulator and a field of view of a thermal imager, the thermal imager generating thermographic data from different positions on the surface of the electrode at different times as the electrode moves on the primary conveyor, so that the electrode is assigned several sequences of thermal images;
[0028] - a computer program comprising an extraction algorithm, into which the thermographic data is inserted, the extraction algorithm being configured to extract defect characteristics from the thermal image sequences;
[0029] - at least one secondary conveyor; the electrode being transferred to the at least one secondary conveyor if the extracted fault characteristics do not meet a predefined selection criterion and the electrode is deemed defective, and the electrode remaining on the primary conveyor in order to be included in the battery if the extracted fault characteristics meet the predefined selection criterion and the electrode is deemed functional. According to one characteristic, the inductive stimulator is configured to deliver a current at a variable operating frequency depending on the extracted fault characteristics.
[0030] According to one feature, the inductive stimulator is an eddy current probe.
[0031] According to one feature, the defect detection system further includes a conditioning circuit configured to format the thermographic data before inserting the thermographic data into the computer program.
[0032] Brief description of the figures
[0033] The invention will now be described, by way of example only, with reference to the attached figure:
[0034] [Fig.1] Figure 1 is a schematic representation of the fault detection system according to the invention.
[0035] In this figure, the same references are used to designate the same elements. For reasons of clarity, the figure is not necessarily reproduced to scale. Additional features may emerge from the following description.
[0036] Detailed description
[0037] The invention relates to a defect detection method for detecting a defect in an electrode 1 intended to be included in a battery, such as for example a lithium-ion battery, during the manufacture of the electrode 1.
[0038] More generally, the fault detection method described can be applied to any type of secondary cell.
[0039] The detection method comprises a first step which consists of placing the electrode 1 on a primary conveyor 5 and circulating it by means of this primary conveyor 5 through an inductive stimulator 2 and a field of vision of a thermal imager 3 as shown in Figure 1.
[0040] An electrode sheet 1 can be continuously supplied by an electrode roll 1 , and a cutter cuts the electrode sheet 1 to obtain the electrode 1 to be tested.
[0041] The cutter may have a speed of between 20 meters / minute and 100 meters / minute and preferably substantially equal to 80 meters / minute and an encoder may be used to measure the speed of the cutter.
[0042] The thermal imager 3 may, for example, be an infrared camera producing a two-dimensional image generated from the thermographic data.
[0043] The thermal imager 3 may have a resolution of between 100 and 2000 micrometers, and preferably substantially equal to 700 micrometers at a frame rate of 1700 frames per second. According to one possibility, the thermal imager 3 is configured such that the thermal imager 3 is turned towards one side of the electrode 1 as shown in FIG. 1. For example, the thermal imager 3 may be configured so as to face an upper face of the electrode 1 and another thermal imager may be configured so as to face a lower face of the electrode 1.
[0044] The inductive stimulator 2 can be configured to deliver a current at a variable operating frequency depending on the extracted fault characteristics.
[0045] The inductive stimulator 2 can be an eddy current probe.
[0046] Advantageously, the use of an eddy current probe in the described defect detection method allows non-destructive testing of the electrode 1.
[0047] The fault detection method may further comprise a step of adjusting the operating frequency of the inductive stimulator 2 as a function of the extracted fault characteristics in order to improve the detection of the electrode fault 1 on the surface of the electrode 1. The operating frequency may vary between 50Hz and 1000MHz for example.
[0048] Advantageously, adjusting the operating frequency of the inductive stimulator 2 makes it possible to refine the detection of defects and therefore to meticulously detect electrode defects.
[0049] Advantageously, having a variable frequency inductive stimulator 2 makes it possible to refine the frequency of the current generated by the inductive stimulator 2.
[0050] The detection method also comprises a step which consists of heating a surface of said electrode 1 using the inductive stimulator 2.
[0051] Following heating of the electrode surface, thermographic data of different positions on the surface of the electrode 1 are obtained using the thermal imager 3 at different times as the electrode 1 moves on the primary conveyor 5, in order to assign to the electrode 1 several thermal image sequences.
[0052] Then, a step is performed which consists of inserting the thermographic data into a computer program 4 comprising an extraction algorithm. The extraction algorithm is configured to extract defect characteristics from the several sequences of thermal images.
[0053] The extraction algorithm may be a machine learning algorithm, including an anomaly detection algorithm or a classification algorithm.
[0054] For example, the extraction algorithm can be a convolutional neural network CNN or a long short-term memory network LSTM.
[0055] Advantageously, the use of machine learning algorithms allows for the automatic detection of defects quickly and efficiently. For example, defects can be considered:
[0056] - a first image defect derived above an aspect ratio of 1.1;
[0057] - a defect in the image derived from 2 ème greater than 0.8 and
[0058] - the image defect derived from 3 ème greater than 0.7.
[0059] In order to detect the smear or anomaly, a derivative of an intensity value f(x,y) on the image made by the thermal imager is calculated in order to find a place where the derivative is maximum. The smear could then be located following a detection of a strong intensity of the gradient in the image.
[0060] The gradient components are described by the following approximation:
[0061] [Math 1] df x,y) f(x + dx,y) — f(x,y)
[0062] — - - = Ax = - - - dx dx
[0063] [Math 2]
[0064] Where the dx and dy graders measure a distance along the x and y axes of the image respectively.
[0065] In discrete images, dx and dy can be measured in terms of the number of pixels between two points. The point for which we have dx = dy = 1 is a point at which the pixel coordinates are called (i, j), so we have:
[0066] [Math 3]
[0067] Ax = (i + 1,;) - f(i,j)
[0068] [Math 4]
[0069] In order to detect the presence of a gradient discontinuity, one could calculate the change in gradient at the point with coordinates (i, j). This can be done by finding the following magnitude measure:
[0070] [Math 5]
[0071] M = Inf^ / Ax 2 + Ay 2 ) and the direction of the gradient is given by:
[0072] [Math 6]
[0073] " Ay
[0074] 6 = tan 1 — Ax
[0075] So, for example, if M > 1.1, which represents a threshold value, then the anomaly is classified as a burr.
[0076] The aspect ratio is a ratio of the thickness of the material to the depth of the defect. Alternatively, the thermographic data is formatted using a conditioning circuit 6 prior to inputting the thermographic data into the computer program 4. For example, the thermographic data may be sorted and formatted to be analyzed by the computer program 4.
[0077] Advantageously, formatting the thermographic data prior to entering the thermographic data into the computer program 4 allows for better analysis of the thermographic data and reduces the risk of error during analysis.
[0078] The conditioning circuit 6 may comprise an analog circuit or a digital circuit. For example, the conditioning circuit 6 may comprise a median filter used to remove salt and pepper noise or impulse noise from a thermal image.
[0079] Furthermore, the fault detection method may comprise a step of adapting at least one parameter of the conditioning circuit 6 as a function of the extracted fault characteristics.
[0080] The at least one parameter of the conditioning circuit 6 may be a kernel of the median filter which may be adaptively modified based on a response of the inductive stimulator 2, and thus depending on the extracted fault characteristics.
[0081] Advantageously, making the at least one parameter of the conditioning circuit 6 dependent on the extracted fault characteristics makes it possible to adapt the conditioning circuit 6 to the extracted fault characteristics in order to characterize the fault more precisely.
[0082] An electrode sorting step is then performed based on the extracted defect characteristics.
[0083] Thus, if the extracted fault characteristics satisfy a predefined selection criterion, electrode 1 is deemed functional and electrode 1 remains on the primary conveyor 5 in order to be included in the battery. If, on the contrary, the extracted fault characteristics do not satisfy the predefined selection criterion, electrode 1 is considered defective and electrode 1 is transferred to at least one secondary conveyor 5'.
[0084] The selection criterion can be a temperature-related parameter such as a logarithmic derivative of temperature.
[0085] According to one embodiment, the electrode 1 which is transferred onto the at least one secondary conveyor 5' can be discarded in a scrapping machine where the defective electrodes are placed.
[0086] Advantageously, the described defect detection method can identify electrode defects 1 quickly and efficiently without interrupting the battery manufacturing process. Advantageously, the defect detection method can be used at any stage of the battery manufacturing process to identify surface and subsurface defects of the electrodes 1.
[0087] The invention further relates to a defect detection system for detecting a defect in an electrode 1 during the manufacture of an electrode 1 intended to be included in a battery. This defect detection system is shown in FIG. 1.
[0088] The fault detection system comprises the pre-described inductive stimulator 2 which heats a surface of the electrode 1.
[0089] The inductive stimulator 2 can be configured to deliver a current at a variable operating frequency depending on the extracted fault characteristics.
[0090] Advantageously, having a variable frequency inductive stimulator 2 makes it possible to refine the frequency of the current generated by the inductive stimulator 2.
[0091] The inductive stimulator 2 can be an eddy current probe.
[0092] Advantageously, the eddy current probe allows eddy current testing which uses the principle of electromagnetic induction to detect the defect of electrode 1 in a simple, economical and non-invasive manner, which means that there is no break in the surface of electrode 1.
[0093] The fault detection system also includes:
[0094] - thermal imager 3;
[0095] - the primary conveyor 5 on which the electrode 1 is arranged and by means of which the electrode 1 circulates through the inductive stimulator 2 and the field of view of the thermal imager 3, the thermal imager 3 generating thermographic data from different positions on the surface of the electrode 1 at different times as the electrode 1 moves on the primary conveyor 5, so that the electrode 1 is assigned several sequences of thermal images;
[0096] - the computer program 4 comprising the extraction algorithm, into which the thermographic data is inserted, the extraction algorithm being configured to extract defect characteristics from the thermal image sequences;
[0097] - at least one 5' secondary conveyor.
[0098] The fault detection system may further comprise a conditioning circuit 6 configured to format the thermographic data before inserting the thermographic data into the computer program 4.
[0099] Advantageously, the conditioning circuit 6 processes the thermographic data so that said thermographic data meets the requirements of the computer program 4 for performing an analysis of said thermographic data. The conditioning circuit 6 may comprise a voltage and current limiting device and an anti-aliasing filter.
[0100] In accordance with what has been described previously, the electrode 1 is transferred to the at least one secondary conveyor 5' if the extracted defect characteristics do not meet a predefined selection criterion and the electrode 1 is deemed defective.
[0101] On the contrary, electrode 1 remains on the primary conveyor 5 in order to be included in the battery if the extracted fault characteristics satisfy the predefined selection criterion and electrode 1 is deemed functional.
[0102] Advantageously, the defect detection system allows the implementation of the defect detection method described above.
[0103] The detection method and detection system described above have industrial application in the field of secondary cell manufacturing.
[0104] What applies in this detailed description to the detection method also applies to the detection system, and vice versa. Although the invention has been described in relation to particular embodiments, it is obvious that it is in no way limited thereto and that it includes all technical equivalents of the means described as well as their combinations if these fall within the scope of the invention.
Claims
CLAIMS 1. A defect detection method for detecting a defect in an electrode (1) intended to be included in a battery during the manufacture of the electrode (1), said detection method comprising the following steps: - placing the electrode (1) on a primary conveyor (5) and circulating it by means of this primary conveyor (5) through an inductive stimulator (2) and a field of vision of a thermal imager (3); - heating a surface of said electrode (1) using the inductive stimulator (2); - obtaining thermographic data of different positions on the surface of the electrode (1) using the thermal imager (3) at different times as the electrode (1) moves on the primary conveyor (5), in order to assign to the electrode (1) several sequences of thermal images; - inserting the thermographic data into a computer program (4) comprising an extraction algorithm, the extraction algorithm being configured to extract defect characteristics from the plurality of thermal image sequences; - sorting the electrodes according to the extracted fault characteristics, such as: if the extracted fault characteristics satisfy a predefined selection criterion, the electrode (1) is deemed functional and the electrode (1) remains on the primary conveyor (5) in order to be included in the battery, if the extracted fault characteristics do not satisfy the predefined selection criterion, the electrode (1) is considered defective and the electrode (1) is transferred to at least one secondary conveyor (5').
2. A method of detecting faults according to claim 1, wherein the inductive stimulator (2) is an eddy current probe.
3. A method of detecting defects according to any one of claims 1 or 2, further comprising a step of: - shaping the thermographic data using a conditioning circuit (6) comprising a median filter, the step of shaping the thermographic data comprising the removal, by the median filter, of impulse noise from said several sequences of thermal images prior to the entry of the thermographic data into the computer program (4).
4. A fault detection method according to claim 3, comprising a step of adapting at least one parameter of the conditioning circuit (6) as a function of the extracted fault characteristics, said parameter of the conditioning circuit (6) being a kernel of the median filter capable of being modified adaptively on the basis of a response of the inductive stimulator (2).
5. A method of detecting defects according to any one of claims 1 to 4, wherein the extraction algorithm is a convolutional neural network CNN or a long short-term memory network LSTM.
6. Defect detection system for detecting an electrode defect (1) during the manufacture of an electrode (1) intended to be included in a battery, said defect detection system comprising: - an inductive stimulator (2) heating a surface of the electrode (1); - a thermal imager (3); - a primary conveyor (5) on which the electrode (1) is arranged and by means of which the electrode (1) circulates through an inductive stimulator (2) and a field of view of a thermal imager (3), the thermal imager (3) generating thermographic data from different positions on the surface of the electrode (1) at different times as the electrode (1) moves on the primary conveyor (5), so that the electrode (1) is assigned several sequences of thermal images; - a computer program (4) comprising an extraction algorithm, into which the thermographic data is inserted, the extraction algorithm being configured to extract defect characteristics from the thermal image sequences; - at least one secondary conveyor (5'); the electrode (1) being transferred to the at least one secondary conveyor (5') if the extracted fault characteristics do not meet a predefined selection criterion and the electrode (1) is deemed defective, and the electrode (1) remaining on the primary conveyor (5) in order to be included in the battery if the extracted fault characteristics meet the predefined selection criterion and the electrode (1) is deemed functional.
7. A fault detection system according to claim 6, wherein the inductive stimulator (2) is an eddy current probe.
8. A fault detection system according to any one of claims 6 or 7, further comprising a conditioning circuit (6) comprising a median filter configured to remove impulse noise from the thermal images, so as to format the thermographic data before inserting the thermographic data into the computer program (4).