Training device and method for predicting adhesion to electrode, electrode monitoring device and electrode manufacturing method using prediction model trained using same, and lithium secondary battery manufactured thereby

Abstract

The present invention relates to a training device and method for predicting adhesion to an electrode, an electrode monitoring device and an electrode manufacturing method using a prediction model trained using same, and a lithium secondary battery manufactured thereby. The training device for predicting adhesion to an electrode according to one embodiment of the present invention comprises: a memory in which a near-infrared spectrum for an electrode and a measurement value for the adhesion of the electrode are stored; a prediction model which receives a differential average of a plurality of wavenumber sections including characteristics for the adhesion of the electrode in the near-infrared spectrum so as to predict the adhesion of the electrode; and a processor which receives the near-infrared spectrum, performs a first-order differentiation on the near-infrared spectrum, extracts the plurality of wavenumber sections from the first-order differentiated near-infrared spectrum, calculates a differential average of the plurality of wavenumber sections, and transmits the differential average to the prediction model, wherein the processor may receive a prediction value predicted by the prediction model for the adhesion of the electrode, and train the prediction model such that the prediction value is close to the measurement value.

Description

A learning device and method for predicting adhesion to an electrode, an electrode monitoring device and an electrode manufacturing method using a prediction model learned using the same, and a lithium secondary battery manufactured thereby.

[0001] Cross-citation with related application(s)

[0002] This application claims the benefit of priority to Korean Patent Application No. 10-2023-0179489, filed December 12, 2023, and Korean Patent Application No. 10-2024-0154935, filed November 5, 2024, the entire contents of which are incorporated herein by reference.

[0003] The present invention relates to a learning device and method for predicting adhesion to an electrode, an electrode monitoring device and electrode manufacturing method using a prediction model learned using the same, and a lithium secondary battery manufactured thereby.

[0004] In order to measure the adhesive strength of an electrode, a portion of the electrode to be measured was cut to a standard, one side of the electrode was fixed to a substrate using double-sided tape, and then the adhesive strength of the electrode was measured while peeling the electrode using a material property measuring device.

[0005] This type of destructive inspection has the disadvantage that some of the electrodes produced are lost, the same process must be repeated for each electrode sample produced, and the electrode peeling process takes a rather long time, making it impossible to measure the adhesive strength of the electrodes that changes during the process in real time.

[0006] Accordingly, there is a need for a method for predicting electrode adhesion that can monitor the adhesion of electrodes in real time during the manufacturing process while reducing the loss of produced electrodes.

[0007] The purpose of the present invention is to provide a learning method and device capable of non-destructively predicting adhesion to an electrode using a near-infrared spectrum and machine learning.

[0008] According to one embodiment of the present invention, a learning device for predicting adhesion to an electrode includes a memory storing a near-infrared spectrum for the electrode and a measurement value for the adhesion of the electrode, a prediction model for predicting the adhesion of the electrode by receiving a differential average of a plurality of wavenumber sections including characteristics of the adhesion of the electrode in the near-infrared spectrum, and a processor for receiving the near-infrared spectrum, performing a first differentiation on the near-infrared spectrum, extracting the plurality of wavenumber sections from the first-differentiated near-infrared spectrum, and calculating a differential average of the plurality of wavenumber sections and transmitting the result to the prediction model, wherein the processor can receive a prediction value predicted by the prediction model for the adhesion of the electrode, and train the prediction model so that the prediction value approaches the measurement value.

[0009] The above prediction model can predict the adhesive strength after the coating process of the electrode or the adhesive strength after rolling of the electrode.

[0010] The plurality of wavenumber sections may include a moisture section in which the characteristics of the near-infrared spectrum change due to moisture within the electrode and a binder section in which the characteristics of the near-infrared spectrum change due to a binder within the electrode.

[0011] The above processor may include a differentiator that first differentiates the near-infrared spectrum using a Savitzky-Golay filter.

[0012] The processor may receive the near-infrared spectrum and the measurement value as input, calculate a plurality of differential averages for each of a plurality of wavenumber sections of the near-infrared spectrum, generate the plurality of differential averages and the measurement value as learning data, and construct a data set by generating the learning data for a plurality of electrodes.

[0013] The processor receives the near-infrared spectrum, the measured value, the thickness of the electrode, and the active material loading value of the electrode as input, calculates a plurality of differential averages for each of a plurality of wavenumber sections of the near-infrared spectrum, and generates the plurality of differential averages, the measured value, the thickness of the electrode, and the active material loading value of the electrode as learning data, and can construct a data set by generating the learning data for a plurality of electrodes.

[0014] The above prediction model can perform learning to predict the adhesive strength of the electrode by applying multiple learning data randomly extracted from the above data set to a random forest learning model.

[0015] A learning method for predicting adhesion to an electrode according to one embodiment of the present invention may include a step in which a processor first differentiates a near-infrared spectrum input to an electrode, extracts a plurality of wavenumber sections including characteristics of adhesion of the electrode from the first differentiated near-infrared spectrum, and calculates a differential average of the plurality of wavenumber sections, a step in which a prediction model receives the differential average of the plurality of wavenumber sections and an input measurement value for the adhesion of the electrode to predict the adhesion of the electrode, and a step in which the processor trains the prediction model such that a predicted value predicted by the prediction model for the adhesion of the electrode approaches the measured value.

[0016] The adhesive strength of the electrode may include the adhesive strength after the coating process of the electrode or the adhesive strength after the rolling process of the electrode.

[0017] The plurality of wavenumber sections may include a moisture section in which the characteristics of the near-infrared spectrum change due to moisture within the electrode and a binder section in which the characteristics of the near-infrared spectrum change due to a binder within the electrode.

[0018] The step of extracting the plurality of frequency intervals may include a step of the processor first differentiating the near-infrared spectrum using a Savitzky-Golay filter.

[0019] The above learning method may further include a step of generating, by the processor, a differential average and the measurement value calculated for a plurality of wavenumber sections of the near-infrared spectrum as learning data, and a step of generating the learning data for a plurality of electrodes to construct a data set.

[0020] The above learning method may further include a step of generating, by the processor, a plurality of differential averages calculated for each of a plurality of wavenumber sections of the near-infrared spectrum, the measured value, the thickness of the electrode, and the active material loading value of the electrode as learning data, and a step of generating the learning data for a plurality of electrodes to construct a data set.

[0021] The step of training the above prediction model may include a step of performing training to predict the adhesive strength of the electrode by applying a plurality of training data randomly extracted from the data set to a random forest training model.

[0022] A method for manufacturing an electrode using a prediction model learned by the above-described learning device may include a step of mixing an active material, a binder, and a conductive material through a mixer to produce a slurry, a step of coating a support with the mixed slurry through a coating device, and a step of drying the support coated with the slurry through a drying oven, a step of irradiating the coated and dried electrode with near-infrared rays through a near-infrared spectrometer to obtain a near-infrared spectrum, and a step of inputting the obtained near-infrared spectrum into the learned prediction model to predict adhesive strength for the coated and dried electrode, a step of rolling the coated and dried electrode through a rolling roller, and a step of cutting the rolled electrode through a cutting device and processing the cut electrode into a predetermined shape through a notching device.

[0023] A method for manufacturing an electrode using a prediction model learned by the above-described learning device may include a step of mixing an active material, a binder, and a conductive material through a mixer to produce a slurry, a step of coating a support with the mixed slurry through a coating device, and a step of drying the support coated with the slurry through a drying oven, a step of rolling the coated and dried electrode through a rolling roller, a step of irradiating the rolled electrode with near-infrared rays through a near-infrared spectrometer to obtain a near-infrared spectrum, and a step of inputting the obtained near-infrared spectrum into the learned prediction model to predict adhesive strength for the rolled electrode, and a step of cutting the rolled electrode through a cutting device and processing the cut electrode into a predetermined shape through a notching device.

[0024] A secondary battery according to one embodiment of the present invention may include a positive electrode manufactured by the above-described electrode manufacturing method, a negative electrode manufactured by the above-described electrode manufacturing method, and a separator interposed between the positive electrode and the negative electrode.

[0025] An electrode monitoring device using a prediction model learned by the above-described learning device includes a monitoring processor that calculates a differential average of a plurality of wavenumber sections including characteristics of the adhesive force of an electrode in a near-infrared spectrum received from the prediction model and a near-infrared spectrometer and transmits the differential average to the prediction model, and the prediction model can receive the differential average of the plurality of wavenumber sections and predict the adhesive force of the electrode.

[0026] The above near-infrared spectrum may be obtained for a coated and dried electrode.

[0027] The above near-infrared spectrum may be obtained for a rolled electrode.

[0028] According to one embodiment of the present invention, there is an advantage in that the adhesion to an electrode can be predicted in a non-destructive manner using a machine learning technique.

[0029] In addition, the change in adhesive strength of the electrode being produced can be monitored in real time using the predicted adhesive strength value, and the quality of the electrode can be determined to be good or bad.

[0030] FIG. 1 is a block diagram of a learning device for predicting adhesion to an electrode according to one embodiment of the present invention.

[0031] Figure 2 is a diagram for explaining multiple frequency ranges extracted by the processor.

[0032] FIG. 3 is a diagram for explaining a prediction model according to one embodiment of the present invention.

[0033] Figure 4 is a flowchart of a learning method for predicting adhesion to an electrode according to one embodiment of the present invention.

[0034] FIG. 5 is a drawing for explaining the result of predicting the adhesive strength of an electrode using a prediction model according to one embodiment of the present invention.

[0035] FIG. 6 and FIG. 7 are examples of predicting the adhesive strength of an electrode in real time using a prediction model learned by a learning device or learning method according to one embodiment of the present invention.

[0036] Figure 8 is a flowchart of an electrode manufacturing method using a prediction model learned by a learning device or learning method according to the first embodiment of the present invention.

[0037] Figure 9 shows an electrode manufacturing process according to the first embodiment of the present invention.

[0038] Figure 10 is a flowchart of a method for manufacturing an electrode using a prediction model learned by a learning device or learning method according to a second embodiment of the present invention.

[0039] Figure 11 shows an electrode manufacturing process according to a second embodiment of the present invention.

[0040] FIG. 12 is a drawing for explaining a lithium secondary battery manufactured by a method for manufacturing an electrode using a prediction model learned by a learning device or learning method according to one embodiment of the present invention.

[0041] In describing the embodiments disclosed in this specification, detailed descriptions of related known technologies will be omitted if it is determined that such detailed descriptions may obscure the gist of the embodiments disclosed in this specification. In addition, the attached drawings are provided solely to facilitate understanding of the embodiments disclosed in this specification, and the technical concepts disclosed in this specification are not limited by the attached drawings, and should be understood to include all modifications, equivalents, and substitutes included within the spirit and technical scope of the present invention.

[0042] Terms that include ordinal numbers, such as first, second, etc., may be used to describe various components, but the components are not limited by these terms. These terms are used solely to distinguish one component from another.

[0043] When a component is referred to as being "connected" or "connected" to another component, it should be understood that it may be directly connected or connected to that other component, but that there may be other components intervening. Conversely, when a component is referred to as being "directly connected" or "connected" to another component, it should be understood that there are no other components intervening.

[0044] In this application, terms such as “include” or “have” are intended to specify the presence of a feature, number, step, operation, component, part or combination thereof described in the specification, but should be understood not to exclude in advance the possibility of the presence or addition of one or more other features, numbers, steps, operations, components, parts or combinations thereof.

[0045] The present invention will be described in detail with reference to the attached drawings below.

[0046] FIG. 1 is a block diagram of a learning device for predicting adhesion to an electrode according to one embodiment of the present invention.

[0047] Referring to FIG. 1, a learning device (1) for predicting adhesion to an electrode according to one embodiment of the present invention may include a memory (100), a prediction model (200), and a processor (300).

[0048] The memory (100) stores a near-infrared spectrum for the electrode and a measurement value for the adhesive strength of the electrode.

[0049] The near-infrared spectrum is measured for the surface of the electrode and can be acquired through a near-infrared spectrometer (440). Here, the near-infrared spectrometer (440) can acquire the near-infrared spectrum for the electrode by scanning the surface of the electrode with near-infrared rays in the wavelength band of 750 to 25,000 nm.

[0050] In some embodiments, the near-infrared spectrum is 4,000 to 13,333 cm -1 It can be a spectrum for the wave number region.

[0051] In some embodiments, the near-infrared spectrum may include multiple characteristic intervals related to the adhesion of the electrode, and the multiple characteristic intervals may be distinguished based on wavenumber.

[0052] The measurement value is obtained by measuring the adhesive strength of the electrode, and can be obtained using a conventionally known electrode adhesive strength measurement method. For example, the measurement value may be a value measured using a 90° peel test.

[0053] Depending on the embodiment, the measured value may be differentiated by the process step of producing the electrode. For example, the measured value may be a value measuring the adhesive strength of the electrode after the coating process or a value measuring the adhesive strength of the electrode after the rolling process.

[0054] The near-infrared spectrum and measurement values ​​stored in the memory (100) can be transmitted to the processor (300) for learning the prediction model (200).

[0055] The prediction model (200) can predict the adhesive strength of an electrode by inputting the differential average of multiple wavenumber sections containing the characteristics of the adhesive strength of the electrode in the near-infrared spectrum.

[0056] Since the characteristics of the near-infrared spectrum change depending on the adhesive strength of the electrode, the prediction model (200) can predict the adhesive strength of the electrode by analyzing the change in the characteristics of the near-infrared spectrum. That is, the characteristics related to the adhesive strength of the electrode are acquired from the near-infrared spectrum, and the prediction model (200) can learn the correlation between the change in the characteristics related to the adhesive strength of the electrode in the near-infrared spectrum and the adhesive strength of the electrode. Using the learned prediction model (200), the adhesive strength of the electrode can be predicted from the characteristics related to the adhesive strength of the electrode in the near-infrared spectrum.

[0057] In some embodiments, the prediction model (200) can predict the adhesion after the coating process of the electrode or the adhesion after the rolling process of the electrode. For example, the prediction model (200) can perform learning to predict the adhesion of the electrode after the coating process by receiving differential averages of multiple wavenumber sections in the near-infrared spectrum of the electrode after the coating process. For example, when the prediction model (200) is trained using the near-infrared spectrum of the electrode after the rolling process, the trained prediction model (200) will predict the adhesion of the electrode after the rolling process. That is, depending on which process the near-infrared spectrum used for training the prediction model (200) is acquired after, an adhesion prediction model (200) for each electrode process can be generated.

[0058] In some embodiments, the prediction model (200) can predict the adhesion of the electrode after the rolling process from the near-infrared spectrum acquired for the electrode after the coating process. For example, when the prediction model (200) is trained using the near-infrared spectrum acquired for the electrode after the coating process and the measurement value of the adhesion of the electrode after the rolling process, a model that predicts the adhesion of the electrode after the rolling process from the near-infrared spectrum acquired for the electrode after the coating process can be generated. That is, the prediction model (200) trained using the near-infrared spectrum of the electrode after the coating process and the measurement value of the adhesion of the electrode after the rolling process can predict and output the adhesion of the electrode after the rolling process when the near-infrared spectrum is input.

[0059] The prediction model (200) can determine various prediction parameters necessary for prediction through learning to predict outputs based on inputs. The prediction model (200) can be implemented as a program for multiple operations that predict outputs for new inputs using the determined prediction parameters. The prediction model (200) can be implemented in hardware by processors such as a CPU or GPU that execute the above program.

[0060] The processor (300) can extract multiple wavenumber intervals from a near-infrared spectrum, and can calculate differential averages of the extracted multiple wavenumber intervals and transmit them to the prediction model (200).

[0061] In an embodiment, the processor (300) may preprocess the near-infrared spectrum to extract a plurality of wavebands. In this case, the preprocessing of the near-infrared spectrum may be performed by the processor (300) receiving the near-infrared spectrum and performing a first differentiation on the near-infrared spectrum. The processor (300) may be implemented as a program including a plurality of instructions that instruct a plurality of operations to perform the above-described operations, and hardware such as a CPU or GPU that executes the program. In an embodiment, the processor (300) may include a differentiator (310), a Savitzky-Golay filter (320), and a learning controller (330).

[0062] The differentiator (310) can first differentiate the near-infrared intensity in each of a plurality of wavenumber sections in the near-infrared spectrum and generate each of the plurality of differential values ​​as each of the plurality of characteristic values.

[0063] In some embodiments, the differentiator (310) may first differentiate the near-infrared intensity in each of a plurality of wavebands using a Savitzky-Golay filter (320).

[0064] The learning controller (330) can derive a plurality of near-infrared spectra for a plurality of electrodes from the database (2) and apply them to the differentiator (310). Each of the plurality of near-infrared spectra can be converted into a plurality of characteristic values ​​for each of the plurality of electrodes through the differentiator (310) and provided to the learning controller (330). The learning controller (330) can match the plurality of characteristic values ​​for each of the plurality of electrodes with the measurement values ​​for each electrode and store them in the memory (100). The learning controller (330) can match the plurality of characteristic values ​​and the measurement values ​​for each of all derived plurality of electrodes and store them as a learning data set in the memory (100).

[0065] The learning controller (330) can provide characteristic values ​​and measurement values ​​for an arbitrary number of electrodes in a learning data set as learning data to the prediction model (200). The prediction model (200) provides the adhesion value (hereinafter, predicted value) predicted through learning using the corresponding learning data to the learning controller (330). The learning controller (330) can adjust the parameters of the prediction model (200) so that the predicted value approaches the measured value.

[0066] Figure 2 is a diagram for explaining multiple frequency ranges extracted by the processor.

[0067] Referring to FIG. 2, the plurality of wavenumber sections are sections that change according to the adhesive force of the electrode in the near-infrared spectrum, and may be specific sections obtained experimentally.

[0068] For example, the plurality of wavenumber intervals may refer to intervals where peaks occur in a preprocessed near-infrared spectrum such as a first differentiation, and may be a plurality of preset wavenumber intervals as shown in [Table 1] below. Here, the plurality of preset wavenumber intervals may be intervals where changes in the characteristics of the near-infrared spectrum due to adhesive force mainly occur.

[0069] Interval frequency [cm -1]Section 15245 - 5060Section 25399 - 5245Section 35646 - 5415Section 46047 - 5754Section 56819 - 6063Section 67143 - 6973Section 77235 - 7143Section 87328 - 7235Section 97451 - 7343Section 108485 - 7513

[0070] In some embodiments, the plurality of wavebands may include a moisture range and a binder range. The moisture range may refer to a range in which the characteristics of the near-infrared spectrum are changed by moisture within the electrode, and the binder range may refer to a range in which the characteristics of the near-infrared spectrum are changed by a binder within the electrode.

[0071] For example, in the above [Table 1], sections 1, 2, 3, 6, 7, and 8 may be moisture sections, and sections 4, 5, 9, and 10 may be binder sections.

[0072] FIG. 3 is a diagram for explaining a prediction model according to one embodiment of the present invention.

[0073] Referring to FIG. 3, a processor (300) according to one embodiment of the present invention may construct a data set (10) and randomly extract learning data from the data set (10) to train a prediction model (200). In addition, the processor (300) may apply the extracted learning data to the prediction model (200) to predict the adhesive strength of an electrode. According to an embodiment, the prediction model (200) may be implemented as a Random Forest learning model.

[0074] According to an embodiment, the processor (300) may receive a near-infrared spectrum and a measurement value, calculate a plurality of differential averages for each of a plurality of wavenumber intervals extracted from the near-infrared spectrum, generate a plurality of differential averages and the measurement values ​​as learning data, and generate the learning data for a plurality of electrodes to construct a data set (10).

[0075] For example, as illustrated in FIG. 3, the processor (300) can construct a data set (10) using 25 training data generated for each of 25 electrodes. Here, each of the 25 training data may include multiple differential averages and measurement values.

[0076] According to an embodiment, the processor (300) may receive a near-infrared spectrum, a measured value, an electrode thickness, and an electrode active material loading value, calculate a plurality of differential averages for each of a plurality of wavenumber intervals extracted from the near-infrared spectrum, generate the plurality of differential averages, the measured value, the electrode thickness, and the electrode active material loading value as learning data, and generate the learning data for a plurality of electrodes to construct a data set (10). In this case, each of the plurality of learning data included in the data set (10) may include a plurality of differential averages, the measured value, the electrode thickness, and the electrode active material loading value.

[0077] According to an embodiment, the prediction model (200) can perform learning to predict the adhesive strength of an electrode by applying a plurality of learning data randomly extracted from the data set (10) to the prediction model (200).

[0078] For example, the prediction model (200) can perform learning to predict the adhesive strength of the electrode as follows. First, the learning controller (330) extracts learning data (21) from 1 to 10 from the data set (10) and provides the learning data to the prediction model (200). The prediction model (200) can be implemented as a random forest learning model. The prediction model (200) can implement a first decision tree by performing learning using the learning data (21). The learning controller (330) randomly extracts 10 learning data (22) from among learning data from 1 to 25 from the data set (10) and provides the learning data to the prediction model (200). The prediction model (200) can implement a second decision tree by performing learning using the learning data (22). The learning controller (330) randomly extracts five learning data (23) from among 1 to 25 learning data from the data set (10) and provides them to the prediction model (200). The prediction model (200) can implement a third decision tree by performing learning using the learning data (22).

[0079] After learning is complete, the prediction model (200) can determine a prediction value using three predicted values ​​derived by applying multiple characteristic values ​​for the input near-infrared spectrum to the first to third decision trees. For example, the prediction model (200) can determine the average of the three predicted values ​​as the predicted value.

[0080] Meanwhile, the number of learning data extracted from the data set (10) and the number of decision trees constituting the random forest learning model (210) are not limited to this example.

[0081] That is, by varying the number of multiple learning data used to generate a decision tree and the method of extracting the learning data, the diversity of the decision tree can be secured, and through this, the prediction accuracy of the random forest learning model (210) can also be improved.

[0082]

[0083] Figure 4 is a flowchart of a learning method for predicting adhesion to an electrode according to one embodiment of the present invention.

[0084] Referring to FIG. 4, a learning method for predicting adhesion to an electrode according to one embodiment of the present invention may include a wave number interval extraction step (S4100), an electrode adhesion prediction step (S4200), and a prediction model learning step (S4300).

[0085] In the wavenumber section extraction step (S4100), the processor (300) performs a first differentiation on the near-infrared spectrum input to the electrode, extracts a plurality of wavenumber sections including characteristics of the adhesive strength of the electrode from the first-differentiated near-infrared spectrum, and calculates a differential average of the plurality of wavenumber sections. At this time, the adhesive strength of the electrode may include the adhesive strength after the electrode coating process or the adhesive strength after the electrode rolling process.

[0086] In some embodiments, the plurality of wavenumber intervals of the near-infrared spectrum may be divided into a moisture interval in which the characteristics of the near-infrared spectrum are changed by moisture within the electrode and a binder interval in which the characteristics of the near-infrared spectrum are changed by a binder within the electrode.

[0087] According to an embodiment, the frequency range extraction step (S4100) may include a step (S4110) in which the processor (300) first differentiates the near-infrared spectrum using a Savitzky-Golay filter (310).

[0088] In the electrode adhesion prediction step (S4200), the prediction model (200) can predict the adhesion of the electrode by receiving the differential average for multiple frequency sections and the input measurement value for the adhesion of the electrode.

[0089] In the prediction model learning step (S4300), the processor (300) can train the prediction model (200) so that the predicted value of the adhesive strength of the electrode is close to the measured value.

[0090] According to an embodiment, the prediction model learning step (S4300) may include a step (S4310) in which the prediction model (200) performs learning to predict the adhesive strength of an electrode by applying a plurality of learning data randomly extracted from a data set (10) to a random forest learning model (210).

[0091] According to an embodiment, a learning method for predicting adhesion to an electrode may further include a step (S4410) of generating differential averages and measurement values ​​calculated by a processor (300) for multiple wavenumber sections of a near-infrared spectrum as learning data, and a step (S4510) of generating learning data for multiple electrodes to build a data set (10).

[0092] According to an embodiment, a learning method for predicting adhesion to an electrode may further include a step (S4420) in which a processor (300) generates a plurality of differential averages, a measurement value, an electrode thickness, and an electrode active material loading value calculated for each of a plurality of wavenumber sections of a near-infrared spectrum as learning data, and a step (S4520) in which learning data is generated for a plurality of electrodes to build a data set.

[0093] That is, the learning data can be generated using multiple differential averages and measured values ​​calculated for each of multiple wavenumber sections of the near-infrared spectrum, or can be generated using multiple differential averages, measured values, electrode thickness, and electrode active material loading value.

[0094]

[0095] FIG. 5 is a drawing for explaining the result of predicting the adhesive strength of an electrode using a prediction model according to one embodiment of the present invention.

[0096] The graph in Fig. 5 shows the results of comparing and verifying the predicted adhesive strength using the prediction model (200) with the actual adhesive strength. In the graph, the horizontal axis represents the actual adhesive strength of the electrode, and the vertical axis represents the predicted adhesive strength predicted using the prediction model (200) according to one embodiment of the present invention. The points indicated in the graph represent the predicted adhesive strength predicted by the prediction model (200) compared to the actual adhesive strength. Therefore, the more linear the points indicated in the graph are, the higher the prediction accuracy.

[0097] Referring to FIG. 5, it can be seen that the adhesive strength of the electrode predicted using the prediction model (200) according to one embodiment of the present invention has an accuracy of 98%.

[0098]

[0099] FIG. 6 and FIG. 7 are examples of predicting the adhesive strength of an electrode in real time using a prediction model learned by a learning device or learning method according to one embodiment of the present invention.

[0100] Referring to FIGS. 6 and 7, a prediction model learned by a learning device (1) or learning method according to one embodiment of the present invention can predict and output in real time the adhesive strength of an electrode that changes during an electrode manufacturing process.

[0101] According to an embodiment, the learned prediction model may receive a near-infrared spectrum measured in real time for the electrode, or a differential average for each of a plurality of wavenumber intervals extracted from the near-infrared spectrum by the processor (300).

[0102] Here, the differential average for each of the plurality of wavenumber sections extracted from the near-infrared spectrum can be calculated through a process of receiving a near-infrared spectrum repeatedly measured at regular time intervals on the surface of the electrode by a near-infrared spectrometer (440), a process of performing first differentiation and filtering on the near-infrared spectrum, a process of extracting a plurality of preset wavenumber sections, and a process of calculating the differential average for each of the plurality of wavenumber sections.

[0103] According to an embodiment, the adhesive force predicted by the prediction model (200) may be transmitted to an external device such as a user terminal or a display screen via the processor (300), and the results predicted by the prediction model (200) may be provided in the form of a graph or table.

[0104] For example, as shown in Fig. 6, it may be provided in the form of a graph in which the horizontal axis represents time (sec) and the vertical axis represents adhesive force, or as shown in Fig. 7, it may be provided in the form of a table for adhesive force measurement time and measurement location. In Fig. 7, Top and Back indicate locations where the near-infrared spectrum was measured for one electrode sample. That is, the processor (300) can visualize changes in adhesive force occurring during the electrode process by providing them in the form of a table or graph.

[0105] Figure 8 is a flowchart of an electrode manufacturing method using a prediction model learned by a learning device or learning method according to the first embodiment of the present invention.

[0106] Referring to FIG. 8, the electrode manufacturing method according to the first embodiment of the present invention may include a mixing step (S8100), a coating process step (S8200), an electrode adhesion prediction step (S8300), a rolling process step (S8400), a slitting step (S8500), and a notching step (S8600).

[0107] Figure 9 shows an electrode manufacturing process according to the first embodiment of the present invention.

[0108] Referring to FIGS. 8 and 9, the mixing step (S8100) can produce a slurry (41) by inputting electrode materials into a mixer (410). That is, the mixer (410) can mix the input electrode materials to produce a slurry (41). At this time, the electrode materials used in the production of the slurry (41) vary depending on the capacity and output of the electrode to be produced, and one or more types of materials such as an active material, a conductive material, a binder, and a thickener can be mixed and used.

[0109] Here, the active material is a material responsible for the main electrochemical reaction of the positive electrode (610) or negative electrode (620), and lithium metal oxide, natural graphite, artificial graphite, etc. can be used alone or in combination. The conductive material is a material that increases electrical conductivity, and carbon black, etc. can be used. The binder is a material that acts as an adhesive to maintain the mechanical stability of the electrode and help the active material adhere well to the support (42), and the adhesive strength of the electrode can be controlled by adjusting the amount of binder used in the preparation of the slurry (41). The thickener is a chemical that increases the viscosity of the slurry (41), and the viscosity of the slurry (41) before coating can be controlled by the amount of the thickener included in the slurry (41).

[0110] The coating process step (S8200) can coat the mixed slurry (41) on a support (42) and dry it.

[0111] The coating process step (S8200) can unroll the support (42) rolled into a roll shape in one direction. At this time, one end of the support (42) unrolled in one direction can be transferred to the coating device (420) by a transfer roller (480). Here, the support (42) is a component that supports the electrode material and provides electrical connection, and may be, for example, aluminum foil or copper foil.

[0112] The coating process step (S8200) can coat at least one surface of a support (42) with a mixed slurry (41) through a coating device (420). At this time, the slurry (41) mixed in the mixer (410) can be transferred to the coating device (420) through a transfer pipe (490), etc. The support (42) coated with the slurry (41) can be transferred to a drying oven (430) by a transfer roller (480).

[0113] Here, the coating device (420) can coat the support (42) using at least one slurry (41). For example, the coating process step (S8200) can coat the support (42) using one type of slurry (41), or can coat the support (42) in multiple layers using two or more slurries (41) having different compositions. When the support (42) is coated in multiple layers, the upper layer slurry (41) and the lower layer slurry (41) can be designed to have different types and composition ratios of active materials, binders, thickeners, and conductive agents depending on the capacity and output of the electrode to be manufactured.

[0114] The coating device (420) can coat the slurry (41) on only one side of the support (42) or on both sides of the support (42). In FIG. 9, the slurry (41) is illustrated as being coated on both the upper and lower surfaces of the support (42), but the shape and position of the slurry (41) coating on the support (42) are not limited thereto. For example, the slurry (41) can be coated only on the upper surface of the support (42) or only on the lower surface of the support (42). In addition, in FIG. 9, the coating device (420) is illustrated as having a pentagonal cross-section and being positioned on the upper surface of the support (42), but the shape and position of the coating device (420) are not limited thereto and can be freely changed according to the design specifications of the electrode manufacturing device. For example, the coating device (420) may be applied with a slot die coating method that evenly applies a liquid slurry (41) onto a support (42) through a slit-shaped nozzle, or a comma coating method that applies a slurry (41) onto a support (42) between two rollers in which a comma-shaped gap is formed.

[0115] The coating device (420) can control the thickness and loading amount of the electrode by controlling the amount of slurry (41) discharged and the coating speed.

[0116] The coating process step (S8200) can dry the support (42) coated with the slurry (41) through a drying oven (430). That is, the drying oven (430) can dry the solvent in the slurry (41) coated on the support (42) to fix the electrode material contained in the slurry (41) to the support (42).

[0117] The drying oven (430) can optimize the drying process of the electrode so that the electrode adhesion is uniform and stable by controlling the temperature and airflow of the drying oven according to the amount of slurry (41) discharged from the coating device (420), the coating speed of the slurry (41), the thickness of the electrode, and the loading amount.

[0118] According to an embodiment, the coating process step (S8200) performs a process of coating and drying the slurry (41) on the upper surface of the support (42) and then a process of coating and drying the slurry (41) on the lower surface of the support (42), thereby manufacturing an electrode in which the slurry (41) is coated and dried on both surfaces of the support (42).

[0119] For example, the coating process step (S8200) may include a process of unrolling a support (42) rolled in a roll shape in one direction, a process of coating the slurry (41) on the upper surface of the support (42) by applying the slurry (41) to the upper surface of the support (42), a process of drying and fixing the slurry (41) coated on the upper surface of the support (42), a process of coating the slurry (41) on the lower surface of the support (42) by applying the slurry (41) to the support (42), and a process of drying and fixing the slurry (41) coated on the lower surface of the support (42).

[0120] The electrode adhesion prediction step (S8300) can predict the adhesion of the electrode by irradiating a near-infrared ray to a coated and dried electrode to obtain a near-infrared spectrum and inputting the obtained near-infrared spectrum into a monitoring device (500).

[0121] At this time, the coated and dried electrode can pass between two near-infrared spectrometers (440). In FIG. 9, the coated and dried electrode is illustrated as passing between two near-infrared spectrometers (440) arranged above and below the transport path. However, the position and arrangement of the near-infrared spectrometers (440) are not limited thereto, and the position and arrangement of the near-infrared spectrometers (440) can be freely changed depending on the position where the slurry (41) is coated on the support (42) or the design of the electrode manufacturing device. For example, when the slurry (41) is coated only on the upper surface of the support (42), the near-infrared spectrometer (440) can be arranged above the transport path of the coated and dried electrode.

[0122] The near-infrared spectrometer (440) can transmit the near-infrared spectrum obtained for the coated and dried electrode to the monitoring device (500). According to an embodiment, the near-infrared spectrometer (440) can repeatedly measure the near-infrared spectrum for the surface of the electrode at regular time intervals.

[0123] The monitoring device (500) can repeatedly receive a near-infrared spectrum from a near-infrared spectrometer (440) at regular time intervals, and can predict and provide an adhesive force for an electrode using the received near-infrared spectrum. According to an embodiment, the monitoring device (500) can include a monitoring processor (510), a prediction model (520), and a display (530). In this case, the prediction model (520) can be a prediction model (520) learned by a learning device (1) according to an embodiment of the present invention.

[0124] The monitoring device (500) can input the near-infrared spectrum received from the near-infrared spectrometer (440) into the monitoring processor (510). The monitoring processor (510) can extract a plurality of wavenumber sections from the near-infrared spectrum, and can calculate a differential average of the extracted plurality of wavenumber sections and transmit the calculated differential average to the learned prediction model (520). For example, the monitoring processor (510) can perform first differentiation and filtering on the received near-infrared spectrum to extract a plurality of preset wavenumber sections, and calculate a differential average for each of the plurality of wavenumber sections. The learned prediction model (520) can predict the adhesion for the coated and dried electrode based on the differential average for each of the plurality of wavenumber sections. The adhesion predicted by the learned prediction model (520) can be transmitted to an external device such as a user terminal or a display (530) through the monitoring processor (510) and provided to the user.

[0125] Here, the adhesion of the electrode predicted through the learned prediction model (520) may be the adhesion of the electrode after the coating process or the adhesion of the electrode after the rolling process. For example, if the learned prediction model (520) is a prediction model (520) learned using the near-infrared spectrum acquired for the electrode after the coating process and the measurement value of the adhesion of the electrode after the coating process, the adhesion of the electrode predicted in the electrode adhesion prediction step (S8300) corresponds to the adhesion of the electrode after the coating process. As another example, if the learned prediction model (520) is a prediction model (520) learned using the near-infrared spectrum acquired for the electrode after the coating process and the measurement value of the adhesion of the electrode after the rolling process, the adhesion of the electrode predicted in the electrode adhesion prediction step (S8300) corresponds to the adhesion of the electrode after the rolling process.

[0126] In the rolling process step (S8400), the coated and dried electrode can be rolled to increase the density of the electrode. At this time, the coated and dried electrode can pass between two rolling rollers (450). In the rolling process step (S8400), the gap between the two rolling rollers (450) and the temperature of the rolling rollers (450) can be adjusted to optimize the thickness and density of the electrode after rolling. As the electrode is rolled in the rolling process step (S8400), the gap of the electrode can be minimized and the adhesive strength can be improved.

[0127] In the slitting step (S8500), the electrode, which has been coated and rolled, can be cut into a required size. At this time, the electrode can be cut into various shapes according to the shape and size of the battery to be manufactured by a cutting device (460).

[0128] In Fig. 9, the cutting device (460) is illustrated as a laser cutter that uses a laser beam to cut the electrode into a desired shape, but the type of the cutting device (460) is not limited thereto. For example, the cutting device (460) may be a rotary die cutter that uses a roller-shaped cutter to cut the electrode into a desired shape, or a scissor-type slitter that uses two blades that operate like scissors to cut the electrode.

[0129] In the notching step (S8600), the cut electrode (43) can be processed into a specific shape suitable for the battery cell design. For example, in the notching step (S8600), a specific portion of the electrode (43) cut by the notching device (470) can be precisely cut to form a tab (44) for connection to the positive electrode (610) or the negative electrode (620). As another example, in the notching step (S8600), the electrode (43) cut by the notching device (470) can be processed into a shape suitable for cell assembly (e.g., cylindrical, square, etc.). Depending on the embodiment, the notching device (470) can be applied with a punching machine or a laser machine.

[0130]

[0131] FIG. 10 is a flowchart of a method for manufacturing an electrode using a prediction model learned by a learning device or learning method according to a second embodiment of the present invention, and FIG. 11 shows an electrode manufacturing process according to the second embodiment of the present invention.

[0132] Referring to FIGS. 10 and 11, a method for manufacturing an electrode using a prediction model learned by a learning device or learning method according to a second embodiment of the present invention may include a mixing step (S9100), a coating process step (S9200), a rolling process step (S9300), an electrode adhesion prediction step (S9400), a slitting step (S9500), and a notching step (S9600).

[0133] The electrode manufacturing method according to the second embodiment of the present invention is compared with the first embodiment of the present invention in which the electrode adhesion prediction step (S9400) is performed after the rolling process step (S9300), and in which the electrode adhesion prediction step (S8300) is performed after the coating process step (S8200). That is, the electrode manufacturing methods according to the first and second embodiments of the present invention differ only in the order of some components (i.e., the coating process step, the rolling process step, and the electrode adhesion prediction step), and other matters are common. Hereinafter, only the parts that are different from the first embodiment will be described with respect to the electrode manufacturing method according to the second embodiment.

[0134] In the coating process step (S9200), the electrode, after coating and drying, can be transported between two rolling rollers (450) via a transport device. In the rolling process step (S9300), the electrode, after rolling, can be transported to a near-infrared spectrometer (440) via a transport device.

[0135] The electrode adhesion prediction step (S9400) can predict the adhesion of the electrode by irradiating a near-infrared ray to a rolled electrode to obtain a near-infrared spectrum and inputting the obtained near-infrared spectrum into a monitoring device (500).

[0136] At this time, the rolled electrode can pass between two near-infrared spectrometers (440). In Fig. 11, the coated and dried electrode is illustrated as passing between two near-infrared spectrometers (440) arranged above and below the transport path. However, the position and arrangement of the near-infrared spectrometers (440) are not limited thereto, and the position and arrangement of the near-infrared spectrometers (440) can be freely changed depending on the position where the slurry (41) is coated on the support (42) or the design of the electrode manufacturing device. For example, when the slurry (41) is coated only on the upper surface of the support (42), the near-infrared spectrometer (440) can be arranged above the transport path of the coated and dried electrode.

[0137] The near-infrared spectrometer (440) can transmit the near-infrared spectrum obtained for the rolled electrode to the monitoring device (500). According to an embodiment, the near-infrared spectrometer (440) can repeatedly measure the near-infrared spectrum for the surface of the electrode at regular time intervals.

[0138] The adhesion of the electrode predicted by the learned prediction model (520) in the electrode adhesion prediction step (S9400) may be the adhesion of the electrode after the coating process or the adhesion of the electrode after the rolling process. For example, if the learned prediction model (520) is a prediction model (520) learned using the near-infrared spectrum acquired for the electrode after the rolling process and the measurement value of the adhesion of the electrode after the coating process, the adhesion of the electrode predicted in the electrode adhesion prediction step (S9400) corresponds to the adhesion of the electrode after the coating process. As another example, if the learned prediction model (520) is a prediction model (520) learned using the near-infrared spectrum acquired for the electrode after the rolling process and the measurement value of the adhesion of the electrode after the rolling process, the adhesion of the electrode predicted in the electrode adhesion prediction step (S9400) corresponds to the adhesion of the electrode after the rolling process.

[0139]

[0140] FIG. 12 is a drawing for explaining a lithium secondary battery manufactured by a method for manufacturing an electrode using a prediction model learned by a learning device or learning method according to one embodiment of the present invention.

[0141] Referring to FIG. 12, a secondary battery according to one embodiment of the present invention may include a positive electrode (610), a negative electrode (620), a separator (630), and an electrolyte (not shown in the drawing).

[0142] The positive electrode (610) may include a positive electrode current collector and a positive electrode active material disposed on at least one surface of the positive electrode current collector. The positive electrode active material is a material that accepts and releases lithium ions in a lithium secondary battery, and charging and discharging of the secondary battery occur in the process of lithium ions being released or inserted from the positive electrode active material. The positive electrode active material may be composed of a lithium metal oxide such as lithium cobalt oxide, lithium iron phosphate, lithium nickel manganese cobalt oxide, or lithium manganese oxide. The positive electrode current collector serves to transfer electrons generated from the positive electrode active material to an external circuit of the battery, and improves the efficiency of the battery by facilitating the flow of electrons. The positive electrode current collector may be composed of aluminum.

[0143] According to an embodiment, the anode (610) can be manufactured by an electrode manufacturing method using a prediction model (200) learned by a learning device and learning method according to an embodiment of the present invention.

[0144] The negative electrode (620) may include a negative electrode current collector and a negative electrode active material disposed on at least one surface of the negative electrode current collector. The negative electrode active material is a material that stores and releases lithium ions in a lithium secondary battery. When the lithium secondary battery is charged, lithium ions move from the positive electrode (610) to the negative electrode (620) and are stored in the negative electrode active material, and when the lithium secondary battery is discharged, lithium ions move from the negative electrode (620) to the positive electrode (610) to generate current. The negative electrode active material may be composed of graphite, silicon, lithium metal, metal oxide, alloy material, etc. The negative electrode current collector serves to help electrons generated from the negative electrode active material flow smoothly to an external circuit, thereby providing a path for the electrons. The negative electrode current collector may be composed of copper.

[0145] According to an embodiment, the cathode (620) can be manufactured by an electrode manufacturing method using a prediction model (200) learned by a learning device and learning method according to an embodiment of the present invention.

[0146] The separator (630) is a thin film positioned between the positive electrode (610) and the negative electrode (620) and has the property of allowing lithium ions to pass through but not electrons. Direct contact between the positive electrode (610) and the negative electrode (620) within the lithium secondary battery is prevented through the separator (630), thereby preventing a short circuit. The separator (630) may be composed of a polymer such as polypropylene (PP) or polyethylene (PE).

[0147] An electrolyte is a medium that transfers lithium ions between the positive electrode (610) and the negative electrode (620) in a lithium secondary battery. The electrolyte helps lithium ions move freely while the positive electrode (610) and the negative electrode (620) exchange electrons. The electrolyte may be composed of an organic solvent such as ethylene carbonate, dimethyl carbonate, or ethyl methyl carbonate.

[0148] Depending on the embodiment, the lithium secondary battery may be a cylindrical, square, or pouch-shaped secondary battery, but is not particularly limited as long as it corresponds to a charging / discharging device.

[0149] Meanwhile, the above-described method can be written as a program that can be executed on a computer, and can be implemented on a general-purpose digital computer that operates the program using a computer-readable recording medium. The computer-readable recording medium may include a storage medium such as a magnetic storage medium such as a ROM, RAM, USB, floppy disk, or hard disk, or an optical readable medium such as a CD-ROM or DVD.

[0150] The scope of the present invention is indicated by the claims described below rather than the detailed description above, and all changes or modifications derived from the meaning and scope of the claims and their equivalent concepts should be interpreted as being included in the scope of the present invention.

[0151] [Explanation of symbols]

[0152] 1: Learning device

[0153] 2: Database

[0154] 10: Data set

[0155] 21, 22, 23: Training data

[0156] 100: Memory

[0157] 200: Prediction model

[0158] 300: Processor

[0159] 310: Differentiator

[0160] 320: Savitzky-Golay filter

[0161] 330: Learning Controller

[0162] 41: Slurry

[0163] 42: Support

[0164] 43: Electrode

[0165] 44: Tap

[0166] 410: Blender

[0167] 420: Coating device

[0168] 430: Drying Oven

[0169] 440: Near-infrared spectrometer

[0170] 450: Rolling roller

[0171] 460: Cutting device

[0172] 470: Notching device

[0173] 480: Transport roller

[0174] 490: Transport pipe

[0175] 500: Monitoring device

[0176] 510: Monitoring Processor

[0177] 520: Prediction Model

[0178] 530: Display

[0179] 610: Bipolar

[0180] 620: Cathode

[0181] 630: Membrane

Claims

1. A memory storing the near-infrared spectrum for the electrode and the measurement values ​​for the adhesive strength of the electrode; A prediction model that predicts the adhesive strength of the electrode by inputting the differential average of multiple wavebands containing the characteristics of the adhesive strength of the electrode in the near-infrared spectrum; and A processor is included that receives the near-infrared spectrum, performs a first differentiation on the near-infrared spectrum, extracts the plurality of wavenumber sections from the first-differentiated near-infrared spectrum, calculates a differential average of the plurality of wavenumber sections, and transmits the result to the prediction model. The above processor, The prediction model receives a predicted value for the adhesive strength of the electrode, and trains the prediction model so that the predicted value approaches the measured value. A learning device for predicting adhesion to electrodes.

2. In paragraph 1, The above prediction model is, Predicting the adhesion after the coating process of the above electrode or the adhesion after the rolling process of the above electrode. A learning device for predicting adhesion to electrodes.

3. In paragraph 1, The above multiple frequency ranges are: A moisture region in which the characteristics of the near-infrared spectrum change due to moisture in the electrode; and A binder section in which the characteristics of the near-infrared spectrum are changed by a binder within the electrode, A learning device for predicting adhesion to electrodes.

4. In paragraph 1, The above processor, A differentiator comprising a first-order differentiation unit for the near-infrared spectrum using a Savitzky-Golay filter, A learning device for predicting adhesion to electrodes.

5. In paragraph 1, The above processor, The near-infrared spectrum and the measurement value are input, and a plurality of differential averages are calculated for each of a plurality of wavenumber sections of the near-infrared spectrum, and the plurality of differential averages and the measurement value are generated as learning data, and the learning data is generated for a plurality of electrodes to build a data set. A learning device for predicting adhesion to electrodes.

6. In paragraph 1, The above processor, The near-infrared spectrum, the measured value, the thickness of the electrode, and the active material loading value of the electrode are input, and a plurality of differential averages are calculated for each of a plurality of wavenumber sections of the near-infrared spectrum, and the plurality of differential averages, the measured value, the thickness of the electrode, and the active material loading value of the electrode are generated as learning data, and the learning data is generated for a plurality of electrodes to construct a data set. A learning device for predicting adhesion to electrodes.

7. In paragraph 5 or 6, The above prediction model is, By applying multiple learning data randomly extracted from the above data set to a random forest learning model, learning is performed to predict the adhesive strength of the electrode. A learning device for predicting adhesion to electrodes.

8. A step of the processor first differentiating the near-infrared spectrum input to the electrode, extracting a plurality of wavenumber sections including characteristics of the adhesive force of the electrode from the first differentiated near-infrared spectrum, and calculating a differential average of the plurality of wavenumber sections; A step for predicting the adhesive force of the electrode by inputting the differential average for the plurality of wave intervals and the input measurement value for the adhesive force of the electrode; and The above processor comprises a step of training the prediction model so that the predicted value predicted by the prediction model for the adhesion of the electrode approaches the measured value. A learning method for predicting adhesion to electrodes.

9. In paragraph 8, The adhesive strength of the above electrode is, Including the adhesion after the coating process of the electrode or the adhesion after the rolling process of the electrode. A learning method for predicting adhesion to electrodes.

10. In paragraph 8, The above multiple frequency ranges are: A moisture region in which the characteristics of the near-infrared spectrum change due to moisture in the electrode; and A binder section in which the characteristics of the near-infrared spectrum are changed by a binder within the electrode, A learning method for predicting adhesion to electrodes.

11. In paragraph 8, The step of calculating the differential average of the above multiple frequency intervals is: The above processor comprises a step of first differentiating the near-infrared spectrum using a Savitzky-Golay filter, A learning method for predicting adhesion to electrodes.

12. In paragraph 8, A step of generating the differential average and the measurement value calculated by the processor for multiple wavenumber sections of the near-infrared spectrum as learning data; and Further comprising a step of generating the above learning data for multiple electrodes to construct a data set. A learning method for predicting adhesion to electrodes.

13. In paragraph 8, A step of generating a plurality of differential averages calculated by the processor for each of a plurality of wavenumber sections of the near-infrared spectrum, the measured value, the thickness of the electrode, and the active material loading value of the electrode as learning data; and Further comprising a step of generating the above learning data for multiple electrodes to construct a data set. A learning method for predicting adhesion to electrodes.

14. In paragraph 12 or 13, The step of training the above prediction model is: The above prediction model includes a step of performing learning to predict the adhesive strength of the electrode by applying a plurality of learning data randomly extracted from the data set to a random forest learning model. A learning method for predicting adhesion to electrodes.

15. In a method for manufacturing an electrode using a prediction model learned by the learning device of paragraph 1, A step of preparing a slurry by mixing an active material, a binder, and a conductive material through a mixer; A step of coating the mixed slurry onto a support through a coating device and drying the support coated with the slurry through a drying oven; A step of irradiating the coated and dried electrode with near-infrared rays through a near-infrared spectrometer to obtain a near-infrared spectrum, and inputting the obtained near-infrared spectrum into the learned prediction model to predict the adhesion to the coated and dried electrode; A step of rolling the coated and dried electrode through a rolling roller; and A step of cutting the rolled electrode through a cutting device and processing the cut electrode into a predetermined shape through a notching device is included. Method for manufacturing electrodes.

16. In a method for manufacturing an electrode using a prediction model learned by the learning device of paragraph 1, A step of preparing a slurry by mixing an active material, a binder, and a conductive material through a mixer; A step of coating the mixed slurry onto a support through a coating device and drying the support coated with the slurry through a drying oven; A step of rolling the coated and dried electrode through a rolling roller; A step of irradiating the rolled electrode with near-infrared rays through a near-infrared spectrometer to obtain a near-infrared spectrum, and inputting the obtained near-infrared spectrum into the learned prediction model to predict the adhesive strength for the rolled electrode; and A step of cutting the rolled electrode through a cutting device and processing the cut electrode into a predetermined shape through a notching device is included. Method for manufacturing electrodes.

17. A positive electrode manufactured by the electrode manufacturing method of Article 15; A cathode manufactured by the electrode manufacturing method of Article 15; and A lithium secondary battery including a separator interposed between the positive and negative electrodes.

18. A positive electrode manufactured by the electrode manufacturing method of Article 16; A cathode manufactured by the electrode manufacturing method of Article 16; and A lithium secondary battery including a separator interposed between the positive and negative electrodes.

19. Prediction model learned by the learning device of paragraph 1; A monitoring processor is included that calculates a differential average of multiple wavebands containing characteristics of the adhesive strength of an electrode in a near-infrared spectrum received from a near-infrared spectrometer and transmits the result to the prediction model. The above prediction model receives the differential average of the multiple frequency sections and predicts the adhesive strength of the electrode. Monitoring device.

20. In paragraph 19, The above near-infrared spectrum was obtained for the coated and dried electrode. Monitoring device.

21. In paragraph 19, The above near-infrared spectrum was obtained for the rolled electrode. Monitoring device.

Images