Battery electrode inspection apparatus and method

Abstract

The battery electrode inspection device according to an embodiment of the present invention extracts an inspection target image including a region suspected of a defect based on an electrode image acquired from a camera, and sets and applies different types of learning models for determining the presence or absence of defects in the electrode corresponding to the inspection target image according to the amount of learning data, thereby shortening the application time due to the initial process application and the process environment change. By re-learning the learning model based on the updated learning data, highly reliable result data with improved accuracy can be obtained.

Description

Technical Field

[0001] This application claims the benefit of the filing dates of Korean Patent Application No. 10-2022-0109765, filed with the Korean Intellectual Property Office on August 31, 2022, and Korean Patent Application No. 10-2023-0103916, filed with the Korean Intellectual Property Office on August 9, 2023, and all of the contents disclosed in the documents of the Korean patent applications are incorporated herein.

[0002] The present invention relates to a battery electrode inspection apparatus and method, and more particularly, to a battery electrode inspection apparatus and method for determining the presence or absence of electrode defects due to surface defects from electrode images using a pre-trained learning model.

Background Art

[0003] Due to the depletion of fossil fuels, the price of energy sources has increased, and there has been a growing concern about environmental pollution. As an environmentally friendly alternative energy source, the demand for secondary batteries is rapidly increasing.

[0004] Among secondary batteries, lithium batteries are being applied to many industrial fields such as mobile application devices, automobiles, robots, and energy storage devices as a countermeasure against current environmental regulations and high crude oil prices.

[0005] Such lithium batteries are generally classified into cylindrical, prismatic, or pouch types according to the shape of the exterior material that houses the electrode assembly.

[0006] Among these, cylindrical batteries are provided in the structure of a battery pack (Cell To Pack, CTP) composed of a plurality of battery cells. In other words, cylindrical batteries are provided in a form in which an electrode formed from a separator between a positive electrode and a negative electrode is wound and inserted into the inside of a battery can.

[0007] In the electrode process of cylindrical batteries, a battery electrode inspection apparatus is used to detect defects on the electrode surface due to foreign substances, scratches, etc.

[0008] Generally, battery electrode inspection equipment analyzes electrode images acquired from cameras installed on the process equipment to determine whether or not there are defects on the electrode surface. However, this means that the resulting data can be affected by on-site conditions or environmental factors such as equipment vibration and changes in process conditions.

[0009] Therefore, conventional battery electrode inspection devices applied program logic designed based on defective images acquired from cameras installed on-site during the initial stages of the process.

[0010] Therefore, conventional battery electrode inspection equipment had the disadvantage of requiring a long time to apply during the initial stages of the process.

[0011] Furthermore, conventional battery electrode inspection devices had the disadvantage that engineers had to continuously redesign the program logic whenever there were changes in on-site conditions or the environment. [Overview of the project] [Problems that the invention aims to solve]

[0012] The objective of the present invention, which aims to solve the above-mentioned problems, is to provide a battery electrode inspection system.

[0013] The objective of the present invention, which aims to solve the above-mentioned problems, is to provide a battery electrode inspection device.

[0014] Another objective of the present invention, in order to solve the problems described above, is to provide a method for inspecting battery electrodes. [Means for solving the problem]

[0015] An artificial intelligence model-based battery electrode inspection apparatus according to one embodiment of the present invention for achieving the above objective includes a memory and a processor that executes at least one instruction stored in the memory, wherein the at least one instruction includes an instruction to obtain an inspection target image in which a region suspected of being defective is extracted from an electrode image in which at least one electrode surface is captured, and an instruction to input the inspection target image into a learning model that has already been trained and output result data regarding the presence or absence of defects in the battery electrode.

[0016] Here, the instruction to output the above result data may include an instruction to determine one of several learning models based on at least one amount of training data.

[0017] In this case, the instruction to determine the above-mentioned learning model may include an instruction to output the above-mentioned result data using a first machine learning-based learning model that has already been trained, if the amount of the above-mentioned learning data is less than a previously set threshold value.

[0018] On the other hand, the instruction to determine the above-mentioned learning model may include an instruction to output the above-mentioned result data using a second, already-trained deep learning-based learning model if the amount of the above-mentioned learning data is greater than or equal to a previously set threshold value.

[0019] On the other hand, the command to output the result data using the first learning model may include a command to extract at least one image feature value from the image to be inspected, and a command to input at least one image feature value into the first learning model to obtain the result data.

[0020] In this case, the instruction to extract the above image feature values ​​may include an instruction to extract the above image feature values ​​from the image to be inspected using a rule-based learning algorithm.

[0021] Here, the image feature value can include at least one data among the height (Height) of pixels, the width (Width) of pixels, the maximum pixel value (Peak White), the minimum pixel value (Peak Dark), the aspect ratio, and the circularity, extracted from the inspection target image.

[0022] Also, the first learning model may be a learning model based on Random Forest.

[0023] On the other hand, the instruction to output the result data using the second learning model can include an instruction to input at least one inspection target image into the second learning model to obtain the result data.

[0024] At this time, the second learning model may be a learning model based on Convolutional Neural Network (CNN).

[0025] Also, the at least one instruction can further include an instruction to re-learn the already learned learning model by utilizing the result data as learning data.

[0026] A battery electrode inspection method using an artificial intelligence model-based battery electrode inspection device according to another embodiment of the present invention for achieving the above object includes a step of obtaining an inspection target image by extracting a region suspected of being defective from an electrode image in which at least one electrode surface is photographed, and a step of inputting the inspection target image into an already learned learning model and outputting result data regarding the presence or absence of defects in the battery electrode.

[0027] Here, the step of outputting the result data can include a step of determining any one of a plurality of learning models according to at least one learning data amount.

[0028] At this time, the step of determining the learning model may include the step of outputting the result data using a first machine learning-based learning model that has already been learned when the amount of the learning data is less than a preset reference value.

[0029] On the other hand, the step of determining the learning model may include the step of outputting the result data using a second deep learning-based learning model that has already been learned when the amount of the learning data is greater than or equal to the preset reference value.

[0030] On the other hand, the step of outputting the result data using the first learning model may include an instruction to extract at least one image feature value from the image to be inspected, and a step of inputting at least one of the image feature values into the first learning model to obtain the result data.

[0031] At this time, the step of extracting the image feature value may include the step of extracting the image feature value from the image to be inspected using a rule-based learning algorithm.

[0032] Here, the image feature value may include at least one data among the height of the pixel, the width of the pixel, the peak white value of the pixel, the peak dark value of the pixel, the aspect ratio, and the circularity ratio extracted from the image to be inspected.

[0033] Also, the first learning model may be a random forest-based learning model.

[0034] On the other hand, the step of outputting the result data using the second learning model may include a step of inputting at least one image to be inspected into the second learning model to obtain the result data.

[0035] In this case, the second learning model described above may be a convolutional neural network (CNN) based learning model.

[0036] Furthermore, the above-described battery electrode inspection method may further include a step of using the above-described result data as training data to retrain the already-trained learning model.

[0037] An artificial intelligence model-based battery electrode inspection system according to yet another embodiment of the present invention for achieving the above objective includes a camera that captures at least one electrode surface and generates an electrode image, and a battery electrode inspection device that acquires an inspection target image by extracting a region suspected of being defective from the electrode image, inputs the inspection target image into a pre-trained learning model, and outputs result data regarding the presence or absence of defects in the battery electrodes. [Effects of the Invention]

[0038] The battery electrode inspection apparatus and method according to the embodiments of the present invention described above, as well as the system including the same, can improve inspection reliability by setting different types of learning models according to the amount of training data to inspect for defects in the image to be inspected, and by retraining the learning model based on the training data that reflects the result data. [Brief explanation of the drawing]

[0039] [Figure 1] This is a conceptual diagram illustrating the process of applying detection logic to a typical battery electrode inspection device. [Figure 2] This is a conceptual diagram illustrating the operation of a typical battery electrode inspection device when an error occurs. [Figure 3] This is a block diagram of a battery electrode inspection system according to an embodiment of the present invention. [Figure 4] This is a block diagram of a battery electrode inspection device according to an embodiment of the present invention. [Figure 5] This is a flowchart of a battery electrode inspection method using a battery electrode inspection device according to an embodiment of the present invention. [Figure 6] This image illustrates the step of extracting an image to be inspected in the battery electrode inspection method according to an embodiment of the present invention. [Figure 7] This is a flowchart illustrating the step of determining whether or not an electrode is defective in a battery electrode inspection method according to an embodiment of the present invention. [Figure 8] This is a conceptual diagram illustrating a battery electrode inspection method using a first learning model, which is one embodiment of the present invention. [Figure 9] This is a conceptual diagram illustrating a battery electrode inspection method according to another embodiment of the present invention. [Figure 10] This is a conceptual diagram illustrating a battery electrode inspection method using a second learning model, which is one embodiment of the present invention. [Modes for carrying out the invention]

[0040] The present invention can be modified in various ways and has many embodiments; therefore, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this should be understood not as limiting the present invention to specific embodiments, but rather as including all modifications, equivalents, or substitutes that fall within the spirit and technical scope of the present invention. Similar reference numerals are used for similar components in the description of each drawing.

[0041] Terms such as First, Second, A, B, etc., may be used to describe various components, but the components should not be limited by such terms. The terms are used solely for the purpose of distinguishing one component from another. For example, without departing from the scope of the present invention, the First component may be named the Second component, and similarly, the Second component may be named the First component. The term "and / or" includes a combination of multiple related items or one of multiple related items.

[0042] When it is stated that one component is "linked" or "connected" to another component, it should be understood that this may mean that it is directly linked or connected to that other component, but that there may also be another component in between. Conversely, when it is stated that one component is "directly linked" or "directly connected" to another component, it should be understood that there is no other component in between.

[0043] The terms used in this application are used solely to describe specific embodiments and are not intended to limit the invention. Singular expressions include plural expressions unless they are clearly different in context. In this application, terms such as “includes” or “having” are intended to specify the presence of features, figures, steps, actions, components, parts, or combinations thereof as described in the specification, and should not be understood to preemptively exclude the presence or possibility of adding one or more other features, figures, steps, actions, components, parts, or combinations thereof.

[0044] Unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as those generally understood by a person of ordinary skill in the art to which this invention pertains. Terms as defined in commonly used dictionaries should be interpreted as having the meaning consistent with their meaning in the context of the relevant art, and not as ideal or overly formal unless explicitly defined herein.

[0045] Preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0046] Figure 1 is a conceptual diagram of the detection logic application process in a typical battery electrode inspection device.

[0047] Referring to Figure 1, the battery electrode inspection device can detect electrodes with defects on their surface and determine whether they are defective.

[0048] A typical battery electrode inspection device uses a pre-designed detection logic to determine the type of defect and whether or not a defect exists in at least one electrode image acquired from a camera installed on-site. Here, the detection logic may be a pre-designed logic created by an engineer using the image features of at least one electrode image in which a defect type has been confirmed.

[0049] Figure 2 is a conceptual diagram of the operation of a typical battery electrode inspection device when an error occurs.

[0050] Referring to Figure 2, the detection logic of a typical battery electrode inspection device is designed based on defective electrode images acquired from a camera installed on-site during the initial application of the process. This has the disadvantage that a typical battery electrode inspection device can experience detection errors due to distortion or deformation of the electrode image if at least one of the on-site environmental factors or process conditions changes, leading to a decrease in defect detection accuracy. Therefore, a disadvantage of typical electrode inspection devices is that the detection logic must be continuously modified by engineers.

[0051] Furthermore, if the detection logic needs to be modified to prevent this, it is necessary to take the above-mentioned process factors into account, which requires acquiring defective images from cameras installed on-site and modifying the detection logic accordingly. This has the disadvantage of being time-consuming.

[0052] Furthermore, a common drawback of battery electrode inspection devices is that, when the detection logic is modified, it is difficult to determine the classification accuracy for at least one defective image that was inspected before the modification. Here, a defective image may be an image that has been determined to be defective due to the occurrence of a defect as a result of the inspection logic.

[0053] Therefore, in this invention, we will describe an electrode detection device and method that uses a pre-learned model to extract images of electrodes suspected of having surface defects from electrode images, and then determines whether or not the electrode in question is defective based on these images.

[0054] Figure 3 is a block diagram of a battery electrode inspection system according to an embodiment of the present invention.

[0055] As shown in Figure 3, the battery electrode inspection system can detect electrodes with surface defects during the electrode manufacturing process and determine whether or not the electrodes are defective.

[0056] More specifically, it may include a camera 1000 and a battery electrode inspection device 5000.

[0057] Camera 1000 can be installed at the site where the electrode process is performed. For example, camera 1000 can be fixedly installed on equipment in which at least one electrode is moved sequentially. In this way, camera 1000 can acquire at least one electrode image by individually photographing the at least one electrode.

[0058] The battery electrode inspection device 5000 can work in conjunction with at least one camera 1000 to acquire at least one electrode image from the camera 1000. However, the battery electrode inspection device 5000 is not limited to this and can acquire at least one electrode image through a variety of paths.

[0059] Subsequently, the battery electrode inspection device 5000 can determine whether or not there is a defect in the battery electrode based on at least one electrode image, using a previously learned model.

[0060] The following provides a more detailed explanation of the battery electrode inspection device 5000.

[0061] Figure 4 is a block diagram of a battery electrode inspection device according to an embodiment of the present invention.

[0062] Referring to Figure 4, the battery electrode inspection device 5000 according to an embodiment of the present invention, as described above, can extract an image of an inspection target area suspected of having a defect from at least one electrode image, and based on this, can determine whether or not the electrode is defective using a pre-learned model. Here, defects may include foreign matter, scratches, cracks, etc.

[0063] More specifically, the battery electrode inspection device 5000 may include a memory 100, a processor 200, a transceiver 300, an input interface device 400, an output interface device 500, and a storage device 600.

[0064] According to the embodiment, the components 100, 200, 300, 400, 500, and 600 included in the battery electrode inspection device 5000 are connected by a bus 700 and can communicate with each other.

[0065] Of the above configurations 100, 200, 300, 400, 500, and 600, the memory 100 and the storage device 600 can be composed of at least one of a volatile storage medium and a non-volatile storage medium. For example, the memory 100 and the storage device 600 can be composed of at least one of a read-only memory (ROM) and a random access memory (RAM).

[0066] Among these, memory 100 may contain at least one instruction executed by processor 200.

[0067] According to the embodiment, at least one command may include a command to obtain an inspection target image in which a region suspected of being defective is extracted from an electrode image in which at least one electrode surface is captured, and a command to input the inspection target image into a previously trained model and output result data regarding the presence or absence of defects in the battery electrodes.

[0068] Here, the instruction to output the above result data may include an instruction to determine one of several learning models based on at least one amount of training data.

[0069] In this case, the instruction to determine the above-mentioned learning model may include an instruction to output the above-mentioned result data using a first machine learning-based learning model that has already been trained, if the amount of the above-mentioned learning data is less than a previously set threshold value.

[0070] On the other hand, the instruction to determine the above-mentioned learning model may include an instruction to output the above-mentioned result data using a second, already-trained deep learning-based learning model if the amount of the above-mentioned learning data is greater than or equal to a previously set threshold value.

[0071] On the other hand, the command to output the result data using the first learning model may include a command to extract at least one image feature value from the image to be inspected, and a command to input at least one image feature value into the first learning model to obtain the result data.

[0072] In this case, the instruction to extract the above image feature values ​​may include an instruction to extract the above image feature values ​​from the image to be inspected using a rule-based learning algorithm.

[0073] Here, the image feature values ​​described above may include at least one of the following data extracted from the image under inspection: pixel height, pixel width, peak white, peak dark, aspect ratio, and roundness.

[0074] Furthermore, the first learning model described above may be a Random Forest-based learning model.

[0075] On the other hand, the instruction to output the result data using the second learning model may include an instruction to input at least one image to be examined into the second learning model and obtain the result data.

[0076] In this case, the second learning model described above may be a convolutional neural network (CNN) based learning model.

[0077] Furthermore, at least one of the above instructions may further include an instruction that uses the above result data as training data to retrain the already trained model.

[0078] On the other hand, processor 200 can mean a central processing unit (CPU), a graphics processing unit (GPU), or a dedicated processor on which the method according to the embodiment of the present invention is performed.

[0079] As described above, the processor 200 can execute at least one program command stored in memory 100.

[0080] Figure 5 is a flowchart of a battery electrode inspection method using a battery electrode inspection device according to an embodiment of the present invention.

[0081] Referring to Figure 5, the battery electrode inspection device 5000 can acquire at least one electrode image (S1000).

[0082] Subsequently, the battery electrode inspection device 5000 can extract an image to be inspected from at least one electrode image (S3000). Here, the electrode image may be an image of the electrode surface captured by at least one camera 1000 mounted on the electrode processing equipment.

[0083] Furthermore, the image to be examined may be an image obtained by individually extracting at least one region in which a defect has occurred within at least one electrode image. Here, the defect may include, but is not limited to, types such as scratches, foreign objects, and deterioration.

[0084] Subsequently, the battery electrode inspection device 5000 can use a previously learned model to determine whether or not there are defects in the electrodes based on the image being inspected (S5000).

[0085] Subsequently, the battery electrode inspection device 5000 can retrain its learning model based on at least one updated training data (S7000).

[0086] Figure 6 is an image illustrating the step of extracting the image to be inspected in the battery electrode inspection method according to an embodiment of the present invention.

[0087] Referring to Figure 6, the image to be inspected can be acquired using a rule-based algorithm. In other words, the battery electrode inspection device 5000 can output the image to be inspected by inputting the electrode image and predefined conditional commands into a rule-based algorithm.

[0088] Here, the rule-based algorithm calculates the average brightness value of the electrode image using predefined conditional instructions, selects at least one pixel that differs from the calculated average brightness value by more than a certain range, extracts a certain region that includes the selected pixel, obtains the image to be inspected, and outputs it. For example, the image to be inspected may be a binary image and can be provided in a size of 128 pixels × 128 pixels.

[0089] Figure 7 is a flowchart illustrating the step of determining whether or not an electrode is defective in the battery electrode inspection method according to an embodiment of the present invention.

[0090] Referring to Figure 7, the battery electrode inspection device 5000 according to an embodiment of the present invention can confirm the amount of training data classified by the data set in the memory 100 (S5100). Here, at least one training data can include at least one inspection result data that has been pre-inspected by the battery electrode inspection device 5000. In other words, the training data can include defective data or normal data. For example, defective data can be classified and stored according to the defect type in the data set.

[0091] This allows the battery electrode inspection device 5000 to select a type of already learned model.

[0092] According to one embodiment, the battery electrode inspection device 5000 can select a first learned model that has already been learned if the amount of learning data is less than a previously set reference value (S5300).

[0093] Subsequently, the battery electrode inspection device 5000 can extract image feature values ​​from the image to be inspected (S5310).

[0094] Subsequently, the battery electrode inspection device 5000 can determine whether or not there are defects in the battery electrodes using the first learned model that has already been learned (S5350).

[0095] On the other hand, according to another embodiment, the battery electrode inspection device 5000 can select a second learning model that has already been learned if the amount of learning data is equal to or greater than a previously set reference value (S5500).

[0096] Subsequently, the battery electrode inspection device 5000 can determine whether or not the battery electrodes are defective using the selected second learning model (S5700).

[0097] Figure 8 is a conceptual diagram illustrating a battery electrode inspection method using a first learning model, which is one embodiment of the present invention.

[0098] Referring to Figure 8, the battery electrode inspection device 5000 can select a first learned model that has already been learned if the amount of learning data is less than a previously set reference value, as described above.

[0099] Subsequently, the battery electrode inspection device 5000 can extract image feature values ​​from the image to be inspected.

[0100] Here, the image feature value may be at least one pixel information extracted from the image being inspected. For example, the features of the image feature value may include at least one of the following: height, width, peak white, peak dark, aspect ratio, and roundness.

[0101] According to the embodiment, the battery electrode inspection device 5000 can acquire image feature values ​​from at least one image to be inspected using a rule-based algorithm when the amount of training data is less than a pre-set value.

[0102] Subsequently, the battery electrode inspection device 5000 can determine whether or not the battery electrodes are defective using the first learned model that has already been trained.

[0103] More specifically, the battery electrode inspection device 5000 can input image feature values ​​extracted from the image to be inspected as input data into a first learning model that has already been trained. Subsequently, the battery electrode inspection device 5000 can output result data regarding the presence or absence of defects in the battery electrodes based on the image to be inspected. Here, the battery electrodes may be electrodes corresponding to the image to be inspected.

[0104] According to the embodiment, the first learned model may be a machine learning-based algorithm. For example, the first learned model may be a bagging sequence random forest pre-trained on at least one pre-classified training data set in a dataset in memory 100.

[0105] Figure 9 is a conceptual diagram illustrating a battery electrode inspection method using a second learning model, which is one embodiment of the present invention.

[0106] Referring to Figure 9, as described above, the battery electrode inspection device 5000 can determine whether or not there is a defect in the battery electrode using a second learning model that has already been learned, if the amount of learning data is equal to or greater than a previously set reference value.

[0107] More specifically, the battery electrode inspection device 5000 can receive the image to be inspected as input data into a second learning model that has already been trained. This allows the battery electrode inspection device 5000 to output result data indicating whether or not the battery electrodes are defective, based on the image to be inspected.

[0108] According to the embodiment, the already trained second training model may be a pre-trained training model based on at least one pre-classified training data set in the dataset in memory 100.

[0109] More specifically, the second learning model, which has already been trained, may be an artificial intelligence (AI)-based classification model. In other words, the second learning model may be a deep-learning-based classification model that has been pre-trained on previously stored training data based on the result data of the battery electrode inspection device 5000.

[0110] According to the embodiment, the already trained second learning model may be an artificial neural network-based learning model of a Convolutional Neural Network (CNN). For example, the battery electrode inspection device 5000 can pre-train a second learning model to extract image feature data of the image to be inspected using a Convolutional Layer, reduce the feature data to a lower dimension through a Pooling Layer, and classify it according to the defect type using a Fully Connected Layer to determine whether or not there is a defect.

[0111] Figure 10 is a conceptual diagram illustrating the relearning step in a battery electrode inspection method according to another embodiment of the present invention.

[0112] Referring to Figure 10, the battery electrode inspection device 5000 can store result data obtained from one embodiment and another embodiment, in other words, electrode defect determination result data, in memory 100 and classify it into a dataset.

[0113] Subsequently, the battery electrode inspection device 5000 can retrain its learning model based on at least one updated training data. In other words, the battery electrode inspection device 5000 can retrain its learning model by utilizing previously learned result data as training data. This increases the amount of training data, thereby improving the inspection reliability of the battery electrode inspection device 5000.

[0114] The battery electrode inspection apparatus and method according to embodiments of the present invention have been described above.

[0115] The battery electrode inspection apparatus and method according to an embodiment of the present invention extracts an inspection target image that includes a region suspected of being defective based on an electrode image acquired from a camera, and applies different types of learning models for determining whether or not the electrode corresponding to the inspection target image is defective, according to the amount of training data, thereby shortening the initial process application time and the application time due to changes in the process environment.

[0116] Furthermore, the above-described battery electrode inspection device and method can obtain highly reliable result data with improved accuracy by retraining the learning model based on updated training data.

[0117] The operation of the methods according to the embodiments and experimental examples of the present invention can be embodied as a computer-readable program or code on a computer-readable recording medium. A computer-readable recording medium includes all types of recording devices that store data that can be read by a computer system. Furthermore, computer-readable recording media can be distributed across networked computer systems, allowing computer-readable programs or code to be stored and executed in a distributed manner.

[0118] Furthermore, computer-readable recording media can include hardware devices specially configured to store and execute program instructions, such as ROM, RAM, and flash memory. Program instructions can include not only machine code generated by a compiler, but also high-level language code that can be executed by a computer using an interpreter or the like.

[0119] Some aspects of the present invention have been described in the context of apparatus, but they can also be described by corresponding methods, where a block or apparatus corresponds to a method step or a feature of a method step. Similarly, aspects described in the context of a method can be described by corresponding blocks or items or features of corresponding apparatus. Some or all of the method steps can be carried out by (or using) hardware devices such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, one or more of the most important method steps can be carried out by such devices.

[0120] While preferred embodiments of the present invention have been described above with reference to the present invention, those skilled in the art will understand that the present invention can be modified and altered in various ways without departing from the spirit and scope of the invention as set forth in the following claims. [Explanation of Symbols]

[0121] 1000: Camera 5000: Battery electrode inspection device 100: Memory 200: Processor 300: Transceiver / Receiver 400: Input Interface Device 500: Output interface device 600: Storage device 700: Bus

Claims

1. An artificial intelligence model-based battery electrode inspection device, memory; and Includes a processor that executes at least one instruction stored in the memory, At least one instruction, A command to obtain an inspection image by extracting a region suspected of being defective from an electrode image in which at least one electrode surface has been captured, and A battery electrode inspection device that includes a command to input the aforementioned image to be inspected into a pre-trained model and output result data regarding the presence or absence of defects in the battery electrodes.

2. The instruction to output the aforementioned result data is: The battery electrode inspection apparatus according to claim 1, comprising an instruction to determine one of a plurality of learning models according to the amount of at least one preclassified training data in the dataset in memory.

3. The instruction to determine the aforementioned learning model is: The battery electrode inspection apparatus according to claim 2, further comprising an instruction to output the result data using a first machine learning-based learning model that has already been trained, if the amount of the training data is less than a previously set reference value.

4. The instruction to determine the aforementioned learning model is: The battery electrode inspection apparatus according to claim 3, further comprising an instruction to output the result data using a second deep learning-based learning model that has already been trained, if the amount of the training data is equal to or greater than a previously set reference value.

5. The instruction to output the result data using the first learning model is: A command to extract at least one image feature value from the image to be examined, and The battery electrode inspection apparatus according to claim 3, comprising an instruction to input at least one of the image feature values ​​into the first learning model to obtain the result data.

6. The command to extract the aforementioned image feature values ​​is: The battery electrode inspection apparatus according to claim 5, comprising an instruction to extract image feature values ​​from the image to be inspected using a rule-based learning algorithm.

7. The aforementioned image feature values ​​are, The battery electrode inspection apparatus according to claim 5, comprising at least one of the following data extracted from the image to be inspected: pixel height, pixel width, peak white, peak dark, aspect ratio, and roundness.

8. The battery electrode inspection apparatus according to claim 3, wherein the first learning model is a random forest-based learning model.

9. The instruction to output the result data using the second learning model is: The battery electrode inspection apparatus according to claim 4, further comprising an instruction to input at least one image to be inspected into the second learning model to obtain the result data.

10. The battery electrode inspection apparatus according to claim 4, wherein the second learning model is a convolutional neural network (CNN) based learning model.

11. The aforementioned at least one instruction, The battery electrode inspection apparatus according to claim 1, further comprising an instruction to retrain the already learned learning model using the aforementioned result data as learning data.

12. A battery electrode inspection method using an artificial intelligence model-based battery electrode inspection device, A step of obtaining an inspection image by extracting a region suspected of being defective from an electrode image in which at least one electrode surface is captured; and A battery electrode inspection method, comprising the step of inputting the aforementioned image to be inspected into a pre-trained model and outputting result data regarding the presence or absence of defects in the battery electrodes.

13. The step of outputting the aforementioned result data is: The battery electrode inspection method according to claim 12, comprising the step of determining one of a plurality of learning models based on the amount of at least one preclassified training data in a dataset in memory.

14. The step of determining the aforementioned learning model is: The battery electrode inspection method according to claim 13, further comprising an instruction to output the result data using a first machine learning-based learning model that has already been trained, if the amount of the training data is less than a previously set reference value.

15. The step of determining the aforementioned learning model is: The battery electrode inspection method according to claim 14, further comprising the step of outputting the result data using a second deep learning-based learning model that has already been trained, if the amount of the training data is equal to or greater than a previously set reference value.

16. The step of outputting the result data using the first learning model is: A command to extract at least one image feature value from the image to be examined, and The battery electrode inspection method according to claim 14, comprising the step of inputting at least one of the image feature values ​​into the first learning model to obtain the result data.

17. The step of extracting the aforementioned image feature values ​​is: The battery electrode inspection method according to claim 16, further comprising the step of extracting image feature values ​​from the image to be inspected using a rule-based learning algorithm.

18. The aforementioned image feature values ​​are, The battery electrode inspection method according to claim 16, comprising at least one of the following data extracted from the image to be inspected: pixel height, pixel width, peak white, peak dark, aspect ratio, and roundness.

19. The battery electrode inspection method according to claim 14, wherein the first learning model is a random forest-based learning model.

20. The step of outputting the result data using the second learning model is: The battery electrode inspection method according to claim 15, further comprising the step of inputting at least one image to be inspected into the second learning model to obtain the result data.

21. The battery electrode inspection method according to claim 15, wherein the second learning model is a convolutional neural network (CNN) based learning model.

22. The aforementioned battery electrode inspection method is: The battery electrode inspection method according to claim 12, further comprising the step of using the aforementioned result data as training data to retrain the already trained learning model.

23. An artificial intelligence model-based battery electrode inspection system, A camera that captures images of at least one electrode surface to generate an electrode image; and A battery electrode inspection system including a battery electrode inspection device that acquires an inspection target image by extracting a region suspected of being defective from the electrode image, inputs the inspection target image into a pre-trained learning model, and outputs result data regarding the presence or absence of defects in the battery electrodes.

Images