Predictive control device and its operating method

Abstract

A predictive control device according to one embodiment disclosed herein may include a data acquisition unit that acquires data related to a base material onto which a slurry is loaded using a coating die, and a processor that analyzes slurry loading characteristics based on the data, derives candidate control values ​​including a slurry temperature based on the slurry loading characteristics, predicts quality of the candidate control values ​​using a predictive model, and derives an optimal control value based on the quality.

Description

Technical Field

[0001] The present invention claims the benefit of priority based on Korean Patent Application No. 10-2022-0052424, filed on April 27, 2022, and incorporates all the contents disclosed in the document of the Korean patent application as part of this specification.

[0002] Embodiments disclosed in the present application relate to a predictive control device and an operating method thereof.

Background Art

[0003] In recent years, research and development on secondary batteries have been actively conducted. Here, a secondary battery is a rechargeable battery, meaning it includes all conventional Ni / Cd batteries, Ni / MH batteries, etc., and recent lithium-ion batteries. Among secondary batteries, lithium-ion batteries have the advantage of much higher energy density compared to conventional Ni / Cd batteries, Ni / MH batteries, etc. Also, lithium-ion batteries can be manufactured in a small and lightweight form, are used as a power source for mobile devices, and in recent years, their range of use has expanded to the power source of electric vehicles and they have attracted attention as a next-generation energy storage medium.

[0004] In such a lithium-ion battery, in order for the electrode assembly to have the maximum charge capacity, the slurry must be formed with a uniform thickness on the opposing surfaces of the electrodes. Therefore, in recent years, the problem of adjusting the uniform coating thickness has become an important technical problem in order to realize high-quality and high-efficiency products.

Summary of the Invention

Problems to be Solved by the Invention

[0005] By automating the correction work in the electrode coating process, it is necessary to ensure the uniformity of the loading amount regardless of the skill level of the operator.

[0006] In the electrode coating process, it is necessary to automate slurry temperature control to ensure uniformity of the loading amount.

[0007] Temperature control of the slurry can affect quality elements other than the uniformity of the loading amount (e.g., coating width and mismatch), so it is necessary to automate the correction process while considering various quality elements together.

[0008] The technical problems of the embodiments disclosed herein are not limited to those mentioned above, and other technical problems not mentioned will be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0009] A predictive control device according to one embodiment disclosed herein may include a data acquisition unit that acquires data relating to a base material on which slurry is loaded using a coating die, and a processor that analyzes slurry loading characteristics based on the data, derives candidate control values ​​including slurry temperature in response to the analysis results of the slurry loading characteristics, predicts the quality of the candidate control values ​​using a predictive model, and derives an optimal control value based on the quality.

[0010] According to one embodiment, the prediction model may be modeled by learning the correlation between the slurry temperature and the slurry loading characteristics.

[0011] According to one embodiment, the data includes slurry loading amount data of the substrate, and the processor may analyze the uniformity of the slurry loading amount based on the slurry loading amount data.

[0012] According to one embodiment, the slurry loading amount data may include central loading amount data and left and right ground area loading amount data.

[0013] According to one embodiment, the processor may adjust the slurry temperature using the optimal control value.

[0014] According to one embodiment, the data acquisition unit acquires second data relating to the substrate after the processor has adjusted the slurry temperature, the processor determines whether a mismatch has occurred based on the second data, derives a second candidate control value including the RPM (rpm, revolutions per minute) of the slurry pump or the die spacing in response to the occurrence of the mismatch, predicts a second quality of the second candidate control value using a second prediction model, and derives a second optimal control value based on the second quality.

[0015] According to one embodiment, the second data includes image data obtained by converting the width of the textured portion of the substrate and the width of the plain portion of the substrate into an image, and the processor may determine whether or not the mismatch has occurred based on the image data.

[0016] According to one embodiment, the processor may adjust the RPM of the slurry pump or the spacing of the dies using the second optimal control value.

[0017] A method for operating a predictive control device according to one embodiment disclosed herein may include the steps of: acquiring data relating to a base material on which slurry is loaded using a coating die; analyzing slurry loading characteristics based on the data; deriving candidate control values ​​including slurry temperature in response to the analysis results of the slurry loading characteristics; and predicting the quality of the candidate control values ​​using a predictive model and deriving an optimal control value based on the quality.

[0018] According to one embodiment, the prediction model may be modeled by learning the correlation between the slurry temperature and the slurry loading characteristics.

[0019] According to one embodiment, the data includes slurry loading amount data of the substrate, and may further include a step of analyzing the uniformity of the slurry loading amount based on the slurry loading amount data.

[0020] According to one embodiment, the slurry loading amount data may include central loading amount data and left and right ground area loading amount data.

[0021] According to one embodiment, the process may further include adjusting the slurry temperature using the optimal control value.

[0022] According to one embodiment, the method may include the steps of: acquiring second data relating to the substrate after adjusting the slurry temperature; determining whether or not a mismatch has occurred based on the second data; deriving a second candidate control value including the RPM (rpm) of the slurry pump or the die spacing in response to the occurrence of the mismatch; predicting a second quality of the second candidate control value using a second prediction model; and deriving a second optimal control value based on the second quality.

[0023] According to one embodiment, the second data may further include image data obtained by converting the width of the textured portion of the substrate and the width of the plain portion of the substrate into an image, and may include the step of determining whether or not the mismatch has occurred based on the image data.

[0024] According to one embodiment, the step of adjusting the RPM of the slurry pump or the spacing of the dies using the second optimal control value may further be included. [Effects of the Invention]

[0025] The predictive control device disclosed in this application automates real-time correction work in the coating process, thereby enabling uniform quality coating regardless of the operator's skill level.

[0026] The predictive control device according to the disclosure of the present application can perform correction work on the temperature of the slurry, in addition to the die spacing and / or the RPM of the slurry pump.

[0027] The effects of the predictive control device according to the disclosure of the present application are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the disclosure of the present application. In relation to the description of the drawings, the same or similar reference numerals may be used for the same or similar components.

Brief Description of the Drawings

[0028] [Figure 1a] It is a diagram showing a predictive control system according to an embodiment disclosed in the present application. [Figure 1b] It is a diagram showing an example of a prediction model according to an embodiment disclosed in the present application. [Figure 2] It is a block diagram showing a predictive control device according to an embodiment disclosed in the present application. [Figure 3] It is a diagram for explaining the slurry loading characteristics referred to in various embodiments disclosed in the present application. [Figure 4] It is a diagram showing control factors and main quality factors referred to in various embodiments disclosed in the present application. [[ID=…]] [Figure 5] It is a diagram showing candidate control values derived by a predictive control device according to an embodiment disclosed in the present application. [Figure 6] It is a diagram for explaining the operation of a predictive control device according to an embodiment disclosed in the present application to derive an optimal control value from candidate control values. [Figure 7] It is a diagram for explaining a method for determining the presence or absence of occurrence of a mismatch according to an embodiment disclosed in the present application. [Figure 8] It is a diagram showing second candidate control values derived by a predictive control device according to an embodiment disclosed in the present application. [Figure 9] It is a diagram for explaining the operation of a predictive control device according to an embodiment disclosed in the present application to derive a second optimal control value from the second candidate control values. [Figure 10] This is a flowchart showing the operation method of a predictive control device according to one embodiment disclosed in this application. [Figure 11] This is a flowchart showing the operation method of a predictive control device according to one embodiment disclosed in this application. [Modes for carrying out the invention]

[0029] Various embodiments of the present invention will be described below with reference to the attached drawings. However, it is understood that the present invention is not limited to any particular embodiment, but rather includes various modifications, equivalents, and / or alternatives to the embodiments of the present invention.

[0030] The various embodiments of this Application and the terminology used herein are not intended to limit the technical features described herein to any particular embodiment, but rather to include various modifications, equivalents, or substitutions of such embodiments. In relation to the description of the drawings, similar or related reference numerals may be used for similar or related components. The singular form of a noun corresponding to an item may include one or more of such items unless the context clearly indicates otherwise.

[0031] In this application, each of the phrases "A or B," "A and at least one of B," "A or at least one of B," "A, B or C," "A, B and at least one of C," and "A, B or at least one of C" may include any one of the items listed together in the phrase in question, or any possible combination thereof. Terms such as "first," "second," "first," "second," "A," "B," "(a)," and "(b)" are used solely to distinguish one component from other components and, unless otherwise stated, do not limit the component in any other respect (e.g., importance or order).

[0032] In this application, when a component (e.g., the first) is referred to as being "connected," "joined," or "linked" to another component (e.g., the second) with or without the terms "functionally" or "communically," it means that the first component may be connected to the other component directly (e.g., by wire), wirelessly, or via the third component.

[0033] According to one embodiment, the methods according to the various embodiments disclosed herein may be provided in a computer program product. The computer program product may be traded as a commodity between a seller and a buyer. The computer program product may be distributed in the form of a device-readable storage medium (e.g., compact disc read-only memory (CD-ROM)) or online (e.g., download or upload) via an application store or directly between two user devices. In the case of online distribution, at least a portion of the computer program product may be at least temporarily stored or temporarily generated in a device-readable storage medium such as the memory of a manufacturer's server, an application store server, or an intermediary server.

[0034] According to various embodiments, each component of the above-mentioned components (e.g., a module or a program) may include one or more individuals, and some of the individuals may be separated and arranged in other components. According to various embodiments, one or more components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Alternatively or additionally, multiple components (e.g., a module or a program) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each of the multiple components in the same or similar manner as those performed by the component in the multiple components before the integration. According to various embodiments, operations performed by a module, program or other component may be performed sequentially, in parallel, iteratively, or heuristically, and one or more of the operations may be performed in a different order, omitted, or one or more other operations may be added.

[0035] Figure 1a shows a predictive control system according to one embodiment disclosed herein. Figure 1b shows an example of a predictive model according to one embodiment disclosed herein.

[0036] Referring to Figure 1a, the predictive control system may include a predictive control device 100, a predictive model 200, and a user terminal 300. However, this is an embodiment for achieving the objectives of the present invention, and some components may be added or removed as needed. Furthermore, each component of the predictive control system shown in Figure 1a represents a functionally distinct functional element, and in an actual physical environment, multiple components may be implemented in an integrated form.

[0037] The predictive control device 100 is a computing device that provides machine learning-based predictive control to multiple users. According to the embodiment, the computing device may include, but is not limited to, a notebook computer, a desktop computer, a laptop computer, a high-performance server computing device, etc. The computing device may include any type of device equipped with computing and communication functions.

[0038] Predictive control may mean a predictive control service for automatic correction of the electrode coating process provided by the predictive control device 100 based on a machine learning model. According to the embodiment, predictive control may include a predictive control service for automatic correction according to and / or mismatch in the slurry loading characteristics of the electrode coating process. According to the embodiment, slurry loading characteristics may include uniformity of the slurry loading amount. Slurry may mean a mixture of solids and liquids or a composition in which fine solid powder particles are suspended in a liquid solvent. For example, slurry may mean an electrode forming composition. Mismatch may mean that the width and / or loading amount of the slurry sprayed onto the opposing surface of the electrode are not uniform.

[0039] The predictive control device 100 can receive data input from the user terminal 300. The data may include slurry loading amount data relating to the loading amount of slurry applied to the base material. The data may also include pre-processed slurry loading amount data. The slurry loading amount data may include central loading amount data and left and right ground area loading amount data. The central loading amount data and left and right ground area loading amount data will be specifically explained in Figure 3.

[0040] The predictive control device 100 can analyze slurry loading characteristics based on data input from the user terminal 300. According to one embodiment, the slurry loading characteristics may include the uniformity of the slurry loading amount.

[0041] The predictive control device 100 can derive an optimal control value from a plurality of candidate control values ​​for correcting the slurry temperature in response to the analysis results of the slurry loading characteristics. In this case, the candidate control values ​​may include the slurry temperature. The predictive control device 100 can predict the quality of the plurality of candidate control values ​​using the predictive model 200. According to the embodiment, the predictive model 200 may learn and model the correlation between slurry temperature and slurry loading characteristics. The predictive control device 100 can predict the quality of the plurality of candidate control values ​​using the predictive model 200 and derive an optimal control value based on the predicted quality.

[0042] According to one embodiment, the predictive control device 100 can adjust the slurry temperature using the derived optimal control value.

[0043] According to one embodiment disclosed herein, when the predictive control device 100 adjusts the slurry temperature, the predictive control device 100 can receive input of second data from the user terminal 300. In this case, the second data may refer to data relating to the substrate after the slurry temperature has been adjusted. According to the embodiment, the second data may include data relating to the width of the textured portion and / or the width of the plain portion of the substrate. According to the embodiment, the data relating to the width of the textured portion and / or the width of the plain portion of the substrate may correspond to image data converted into an image.

[0044] According to the embodiment, the predictive control device 100 can determine whether or not a mismatch has occurred based on the second data. If a mismatch occurs, the predictive control device 100 can derive a second optimal control value from a plurality of second candidate control values ​​to correct the mismatch. In this case, the second candidate control values ​​may include the RPM (rpm) of the slurry pump or the die spacing. The predictive control device 100 can predict the second quality of a plurality of second candidate control values ​​using a second predictive model. In this case, according to the embodiment, the predictive model 200 may include a second predictive model. The predictive control device 100 can predict the second quality of a plurality of second candidate control values ​​using the second predictive model and derive a second optimal control value based on the predicted second quality.

[0045] According to one embodiment, the predictive control device 100 can adjust the RPM or die spacing of the slurry pump using the derived second optimal control value.

[0046] The prediction model 200 may mean a machine learning model that calculates a predictive control value for input data and provides an optimal control value. According to the embodiment, the prediction model 200 may include a second prediction model. According to the embodiment, the prediction model 200 may be constructed using traditional machine learning methods or deep learning methods. For example, the prediction model 200 may be an artificial intelligence model based on a learning method for transferring already learned knowledge (e.g., transfer learning). For example, the prediction model 200 may include an artificial intelligence model for transferring learned content between tasks (e.g., a PNN (progressive neural network)). Here, a task may mean each operation or goal of a sequence included in the artificial intelligence model. For example, the task of the prediction model 200 may be the calculation of a predictive control value (or optimal control value) for input data and the output of the calculation result. The operation of the prediction model 200 when the prediction model 200 is composed of a PNN will be described below with reference to Figure 1b. However, the prediction model 200 is not limited to the example presented. For example, the prediction model 200 may include an artificial intelligence model based on fine tuning.

[0047] In one embodiment, the prediction model 200 may include a plurality of networks 211, 212, and 213. In one embodiment, the prediction model 200 may have a structure in which each of the networks 211, 212, and 213 is connected to one another. Here, each of the networks 211, 212, and 213 may be an artificial intelligence network for generating results for a specified task.

[0048] In one embodiment, new networks may be added to the prediction model 200. For example, if the prediction model 200 only has network 211, network 212 may be added. Alternatively, if the prediction model 200 has networks 211 and 212, network 213 may be added.

[0049] In one embodiment, if a new network is added to the prediction model 200, the training of the existing network may be interrupted. Here, interruption of training may mean that the layers and output weights included in the existing network are not changed.

[0050] In one embodiment, if a new network is added to the prediction model 200, the new network may be trained. Here, training may be performed based on a backpropagation algorithm.

[0051] In one embodiment, the networks 211, 212, and 213 included in the prediction model 200 may be connected to each other in such a way that the intermediate calculation results of existing networks are input to the new networks. For example, if the first network 211, the second network 212, and the third network 213 are added to the prediction model 200 in that order, the intermediate calculation results of the first network 211 may be input to the second network 212 and the third network 213. Also, the intermediate calculation results of the second network 212 may be input to the third network 213.

[0052] In one embodiment, some of the networks 211, 212, and 213 included in the prediction model 200 can be merged. For example, the first network 211 and the second network 212 may be merged into a single network. In one embodiment, by merging the first network 211 and the second network 212, the sum of the outputs of the first layer 221 of the first network 211 and the first layer 222 of the second network 212 may be the same as the output of the first layer of the merged network. Also, by merging the first network 211 and the second network 212, the sum of the outputs of the second layer 221 of the first network 211 and the second layer 222 of the second network 212 may be the same as the output of the second layer of the merged network. Similarly, by merging the first network 211 and the second network 212, the sum of the output of the nth layer of the first network 211 and the output of the nth layer of the second network 212 may be the same as the output of the nth layer of the merged network.

[0053] In one embodiment, merging may occur if the number of some of the networks 211, 212, and 213 included in the prediction model 200 exceeds a threshold value.

[0054] In one embodiment, the input 201 of networks 211, 212, 213 may be a second candidate control value (e.g., die spacing and slurry pump RPM). In one embodiment, the outputs 241, 242, 243 of networks 211, 212, 213 may include a loading pattern (e.g., uniformity and / or mismatch of slurry loading amount) or a coating width.

[0055] In one embodiment, the first network 211 may be trained such that the input 201 (i.e., the second candidate control value) and the output 241 (i.e., the loading pattern or coating width) have a positive correlation. For example, the first network 211 may be trained such that the value of the output parameter (e.g., coating width) increases or decreases as the left spacing and / or right spacing increases or decreases according to the step size (e.g., ±1 μm). In one embodiment, the training data for the first network 211 may be arbitrarily generated data.

[0056] In one embodiment, the second network 212 may be trained based on actually acquired data (i.e., data showing the correlation between the second candidate control value and the loading pattern or coating width). In one embodiment, the second network 212 can generate an output 242 based on the input 201 (i.e., the second candidate control value) and the outputs of layers 221 and 231 of the first network 211. In one embodiment, the second network 212 may be trained such that the difference between the output 242 of the second network 212 and the actual data regarding the input 201 (i.e., the loading pattern or coating width) is minimized. Here, the training of the second network 212 may be performed based on a backpropagation algorithm. In one embodiment, the first network 211 may not be trained when the second network 212 is trained. In one embodiment, the parameters of the first network 211 may be fixed when the second network 212 is trained.

[0057] In one embodiment, the third network 213 may be trained based on data acquired after the training data of the second network 212 (i.e., data showing the correlation between the second candidate control value and the loading pattern or coating width). In one embodiment, the third network 213 can generate an output 243 based on the input 201 (i.e., the second candidate control value) and the outputs of layers 221 and 231 of the first network 211 and layers 222 and 232 of the second network 212. In one embodiment, the third network 213 may be trained such that the difference between the output 243 of the third network 213 and the actual data regarding the input 201 (i.e., the loading pattern or coating width) is minimized. Here, the training of the third network 213 may be performed based on a backpropagation algorithm. In one embodiment, the first network 211 and the second network 212 do not need to be trained when the third network 213 is trained. In one embodiment, the parameters of the first network 211 and the second network 212 may be fixed when the third network 213 is trained.

[0058] In one embodiment, if a new network is added after the third network 213 has been trained, the training data for the new network may be data acquired after the training data for the third network 213. Also, when the new network is trained, the parameters of the previously added networks 211, 212, and 213 may be fixed. Finally, if a new network is added after the third network 213 has been trained, some of the previously added networks 211, 212, and 213 may be merged in the order they were added.

[0059] In one embodiment, the optimal control value provided by the prediction model 200 may be the output 243 of the network that was added most recently (i.e., the third network 213) among the outputs 241, 242, and 243 of networks 211, 212, and 213.

[0060] According to one embodiment, the predictive control device 100 can update the predictive model 200 using a feedback learning model. According to one embodiment disclosed herein, the predictive model 200 may include a second predictive model. The feedback learning model may mean a machine learning model that has learned user feedback regarding satisfaction with the prediction results. According to this embodiment, since the predictive model 200 is updated based on user satisfaction, user satisfaction with predictive control can be improved.

[0061] According to one embodiment, the prediction model 200 may be built on a computing device physically independent of the prediction control device 100, or it may be built on the prediction control device 100. Hereinafter, the description will assume that the prediction model 200 is built on the prediction control device 100.

[0062] The user terminal 300 may mean a user terminal that transmits data providing predictive control. According to the embodiment, the user terminal 300 may correspond to a vision system. Here, the vision system may mean a system that generates data relating to a substrate coated with slurry by a coating die. Here, the substrate may mean a rechargeable battery including electrodes. For example, the substrate may include a secondary battery.

[0063] According to one embodiment, the user terminal 300 can generate slurry loading profile data. The loading profile data may include slurry loading amount data. According to one embodiment, the slurry loading amount data may be pre-processed. According to one embodiment, the user terminal 300 can generate image data obtained by converting the width of the textured portion and / or the width of the plain portion of the substrate into an image. The prediction control device 100 can acquire data from the user terminal 300. The prediction control device 100 can provide prediction results based on the data acquired from the user terminal 300. In this case, the prediction control device 100 can provide a user interface for inducing user feedback, and the user can provide user feedback information regarding satisfaction with the prediction results via the user interface. The prediction control device 100 can progressively train a feedback learning model using the user feedback information.

[0064] The components of the predictive control system can communicate over a network. According to the embodiment, the network can be any type of wired or wireless network, such as a local area network (LAN), a wide area network (WAN), a mobile radio communication network, or Wibro (Wireless Broadband Internet).

[0065] The configuration and operation of the predictive control device 100 will be described below with reference to Figures 2 to 11.

[0066] Figure 2 is a block diagram showing a predictive control device according to one embodiment disclosed in this application.

[0067] Referring to Figure 2, the predictive control device 100 may include a data acquisition unit 110 and a processor 120. The predictive control device 100 in Figure 2 may have substantially the same configuration as the predictive control device 100 in Figure 1.

[0068] The data acquisition unit 110 can acquire data related to the target of predictive control from the user terminal 300.

[0069] According to the embodiment, the target of predictive control may include a base material on which slurry is loaded using a coating die. The data acquisition unit 110 can acquire data from the user terminal 300 regarding the base material on which slurry is loaded using a coating die. According to the embodiment, the data may include slurry loading amount data of the base material. According to the embodiment, the slurry loading amount data may correspond to data pre-processed at the user terminal 300. According to the embodiment, the slurry loading amount data may include central loading amount data and / or left and right ground area loading amount data.

[0070] According to the embodiment, the target of predictive control may include the substrate after the processor 120 has adjusted the slurry temperature. The data acquisition unit 110 can acquire second data relating to the substrate after the processor 120 has adjusted the slurry temperature from the user terminal 300, as described later. According to the embodiment, the second data may include image data obtained by converting the width of the textured portion and / or the width of the plain portion of the substrate into an image. In the embodiments disclosed herein, this specification is prepared assuming that the target of predictive control is a substrate on which slurry is loaded using a coating die, but it is not limited thereto.

[0071] The processor 120 can analyze slurry loading characteristics based on the data acquired by the data acquisition unit 110. According to one embodiment, the slurry loading characteristics may include the uniformity of the slurry loading amount. For example, when the data acquisition unit 110 acquires slurry loading amount data, the processor 120 can analyze the uniformity of the slurry loading amount based on the slurry loading amount data.

[0072] The processor 120 can derive candidate control values ​​in response to the analysis results of the slurry loading characteristics. According to the embodiment, the candidate control values ​​may include the slurry temperature. For example, the processor 120 can derive candidate control values ​​including the slurry temperature based on the slurry loading characteristics. This will be explained in detail with reference to Figure 5.

[0073] The processor 120 can predict the quality of candidate control values ​​using a predictive model. According to one embodiment, the predictive model may be modeled by learning the correlation between slurry temperature and slurry loading characteristics. The processor 120 can derive an optimal control value based on the predicted quality. The process by which the processor 120 derives the optimal control value will be specifically explained with reference to Figure 6.

[0074] When the data acquisition unit acquires second data according to the embodiments disclosed herein, the processor 120 can determine whether or not a mismatch has occurred based on the second data. According to the embodiments, the processor 120 can calculate whether or not a mismatch has occurred using the width lengths of the top and / or back of two different opposing surfaces of the substrate. Of the upper or lower surface of the electrode-coated substrate, the top may mean the surface with a relatively longer coating length, and the back may mean the surface with a relatively shorter coating length. For example, the processor 120 can calculate a mismatch value using the width lengths of the textured portion and / or the plain portion of the top and / or back of the substrate. The textured portion means the area on the substrate that is loaded with slurry and coated, and the plain portion may mean the area on the substrate that is not loaded with slurry.

[0075] According to one embodiment, the processor 120 can derive a second candidate control value in response to the occurrence of a mismatch. According to another embodiment, the second candidate control value may include the RPM (rpm) of the slurry pump or the die spacing. For example, the processor 120 can derive a second candidate control value including the RPM of the slurry pump or the die spacing in response to the occurrence of a mismatch. This will be explained in detail with reference to Figure 8.

[0076] When the processor 120 derives a second candidate control value according to the embodiments disclosed herein, the second quality of the second candidate control value can be predicted using a second prediction model. Based on the predicted second quality, the processor 120 can derive a second optimal control value. The process by which the processor 120 derives the second optimal control value will be specifically explained with reference to Figure 9.

[0077] Figure 3 is a diagram illustrating the slurry loading characteristics referenced in various embodiments disclosed herein.

[0078] Referring to Figure 3, the base material may include textured sections A1, A2, A3 and plain sections B1, B2. The slurry loading characteristics analyzed by the predictive control device 100 will be described in detail below with reference to Figures 1 and 2.

[0079] The processor 120 can analyze slurry loading characteristics based on data. According to one embodiment, the processor 120 can analyze slurry loading characteristics based on slurry loading amount data of the land areas A1, A2, and A3. According to one embodiment, the slurry loading amount data may include central loading amount data and left and right land area loading amount data. According to this embodiment, the processor 120 can classify the entire and / or part of the land area into a central land area and left and right land areas according to a predetermined criterion. In this case, central loading amount data may mean the average of the slurry loading amount data of the central land area, and left and right land area loading amount data may mean the average of the slurry loading amount data of the left and right land areas. As an example, the processor 120 can classify A2 into the central land area and A1 and A3 into the left and right land areas. As another example, the processor 120 can classify l3 and l4 into the central land area and l1 and l6 into the left and right land areas. The processor 120 can determine different criteria for classifying the central land area and / or left and right land areas depending on the type and characteristics of the substrate, the purpose of optimizing the predictive control device, and the type of equipment used to implement the predictive control device.

[0080] According to one embodiment, the processor 120 can analyze slurry loading characteristics based on slurry loading amount data. According to one embodiment, the slurry loading characteristics may include the uniformity of the slurry loading amount. For example, the processor 120 can analyze the uniformity of the slurry loading amount based on the slurry loading amount data of the ground areas A1, A2, and A3. The processor 120 can calculate a loading deviation using the slurry loading amount data in order to analyze the uniformity of the slurry loading amount. According to one embodiment, the processor 120 can calculate a loading deviation using the central loading amount data and the loading amount data of the left and right ground areas.

[0081] For example, processor 120 can calculate the loading deviation using the following formula 1.

[0082] [Formula 1] Loading deviation = Central loading amount data - Left and right loading amount data

[0083] In one embodiment disclosed herein, when the processor 120 classifies A2 into the central area and A1 and A3 into the left and right areas, the processor 120 can calculate the loading deviation using the following formula 2.

[0084] [Formula 2] Loading deviation = (Average slurry loading amount for A2) - (Average slurry loading amount for A1 and A3)

[0085] In other embodiments disclosed herein, when the processor 120 classifies l3 and l4 into the central area and l1 and l6 into the left and right areas, the processor 120 can calculate the loading deviation using the following formula 3.

[0086] [Formula 3] Loading deviation = (Average slurry loading amount for l3 and l4) - (Average slurry loading amount for l1 and l6)

[0087] According to one embodiment, the processor 120 can analyze that the closer the loading deviation is to zero, the more uniform the slurry loading amount is.

[0088] The processor 120 can analyze the slurry loading characteristics and set a control target for the predictive control device 100 in response to the analysis results of the slurry loading characteristics. According to one embodiment, the processor 120 can analyze the uniformity of the slurry loading amount and set the control target for the predictive control device 100 to a correction related to the loading pattern in response to the analysis results. The details of how the processor 120 controls the control factors according to the control target are explained in detail in Figure 4.

[0089] Figure 4 shows control factors and key quality elements referenced in various embodiments disclosed in this application.

[0090] Referring to Figure 4, the processor 120 can adjust the main quality element Y by controlling the control factor X.

[0091] The control factor X may include slurry temperature, slurry pump RPM (rpm), or die spacing. According to the embodiment, the slurry pump RPM may affect the slurry injection speed of the coating die. According to the embodiment, the die spacing may include left-side spacing and / or right-side spacing of the coating dies. In this case, left-side spacing may mean the left-side spacing of the area in which the coating die used for coating the battery is connected to the motor, and right-side spacing may mean the right-side spacing of the area in which the coating die used for coating the battery is connected to the motor. The main quality element Y may include loading pattern or coating width. According to the embodiment, the loading pattern may include uniformity and / or mismatch of the slurry loading amount.

[0092] The processor 120 can control control factors X to correct a major quality element Y. According to the embodiment, the control factors X can act as mutually dependent variables. Depending on the control objective, the processor 120 can determine a portion of the control factors X as a primary control factor and another portion as a secondary control factor. For example, if the control objective is a correction related to the loading pattern, the processor 120 can determine the slurry temperature and / or die spacing as primary control factors and the slurry pump RPM as a secondary control factor. As another example, if the control objective is a correction related to the coating width, the processor 120 can determine the die spacing as a primary control factor and the slurry temperature and / or slurry pump RPM as secondary control factors.

[0093] According to embodiments disclosed herein, when the processor 120 increases the slurry temperature to compensate for the uniformity of the slurry loading amount, the slurry can be loaded onto the substrate in a concave pattern (U-shaped pattern). According to other embodiments, when the processor 120 decreases the slurry temperature to compensate for the uniformity of the slurry loading amount, the slurry can be loaded onto the substrate in a convex pattern (A-shaped pattern).

[0094] Figure 5 shows candidate control values ​​derived by a predictive control device according to one embodiment disclosed in this application.

[0095] Referring to Figure 5, the processor 120 of the predictive control device 100 can derive candidate control values ​​including slurry temperature. The processor 120 can derive candidate control values ​​using slurry temperature, die spacing, and slurry pump RPM (Pump rpm) as control factors. According to the embodiments disclosed herein, die spacing may include left gap (L Gap) and right gap (R Gap). The processor 120 can derive candidate control values ​​using the current values ​​and adjustment ranges of each control factor.

[0096] As shown in Figure 5, the processor 120 can fix the left spacing, the right spacing, and the RPM of the slurry pump, and derive candidate control values ​​by providing an adjustment range for the slurry temperature. According to the embodiment, the processor 120 can adjust the slurry temperature at preset intervals within the adjustment range. According to the embodiment shown in Figure 5, the processor 120 can set the adjustment range for the slurry temperature to ±2°C and derive candidate control values ​​in which the slurry temperature is adjusted at ±0.4°C intervals based on the current slurry temperature.

[0097] However, Figure 5 represents one embodiment of the candidate control values ​​derived by the processor 120, and the candidate control values ​​derived by the processor 120 are not limited to the values ​​shown in Figure 5. According to the embodiment, the processor 120 can be set to have different adjustment ranges and / or adjustment intervals for the slurry temperature.

[0098] The processor 120 can predict the quality of candidate control values ​​using a predictive model and predict the optimal control value based on that quality.

[0099] According to this embodiment, the processor 120 can perform filtering on candidate control values. According to this embodiment, the processor 120 can minimize the number of candidate control values ​​by filtering. In this case, the processor 120 can predict the quality of the candidate control values ​​minimized by filtering using a predictive model, and predict the optimal control value based on the quality.

[0100] Figure 6 is a diagram illustrating the operation of a predictive control device according to one embodiment disclosed in this application, which derives an optimal control value from a candidate control value.

[0101] Referring to Figure 6, the processor 120 can predict the quality of candidate control values ​​using a predictive model and derive an optimal control value based on that quality.

[0102] The processor 120 can predict the quality of candidate control values ​​using a predictive model. Quality may include, but is not limited to, central loading amount data, left and right loading amount data, and / or loading deviation.

[0103] According to the embodiment, the processor 120 can use a prediction model to calculate predicted values ​​of central loading amount data, left and right ground loading amount data, and / or loading deviation for candidate control values. According to the embodiment, if the processor 120 adjusts the slurry temperature by 0.4°C, the predicted values ​​of the central loading amount data and / or left and right ground loading amount data may change by up to 6 mg.

[0104] According to the embodiment, the processor 120 can predict quality based on the predicted values ​​of central loading amount data, left and right loading amount data, and / or loading deviation calculated using a predictive model.

[0105] The processor 120 can derive an optimal control value based on quality. According to the embodiment, the processor 120 can derive a candidate control value as the optimal control value in which the predicted central loading amount data, predicted left and right loading amount data, or predicted loading deviation is closest to the target value. For example, the processor 120 can derive a candidate control value as the optimal control value in which the predicted loading deviation is closest to the target value of 0. According to the embodiment shown in Figure 6, the processor 120 can derive a candidate control value as the optimal control value in which the predicted value of the loading deviation is 0.

[0106] According to one embodiment, the processor 120 can adjust the slurry temperature using the derived optimal control value. According to the embodiment shown in Figure 6, the processor 120 can adjust the slurry temperature to 29°C using the optimal control value for which the predicted loading deviation is 0.

[0107] Figure 7 is a diagram illustrating a method for determining whether or not a mismatch has occurred according to one embodiment disclosed in this application.

[0108] Referring to Figure 7, the data acquisition unit 110 can acquire second data. The second data may include image data obtained by converting the width of the textured portion and the width of the plain portion of the substrate into an image. According to this embodiment, the second data may correspond to data relating to the substrate after the processor 120 has adjusted the slurry temperature using an optimal control value.

[0109] The processor 120 can determine whether or not a mismatch has occurred based on second data acquired from the user terminal 300. According to the embodiment, the processor 120 can determine whether or not a mismatch has occurred based on image data obtained by converting the width of the textured portion and the width of the plain portion of the substrate into an image. For example, the processor 120 can determine that a mismatch has occurred based on the width differences M1, M2, M3, and M4 between the top and back of the substrate.

[0110] The processor 120 can calculate width differences M1, M2, M3, and M4 using the widths W1, W3, W5 of the plain areas and W2, W4 of the textured areas on the top of the substrate, and the widths X1, X3, X5 of the plain areas and X2, X4 of the textured areas on the back of the substrate, and determine whether or not a mismatch has occurred. A mismatch may mean that the width and / or loading amount of the slurry sprayed onto the opposing surface of the electrode are not uniform.

[0111] For example, processor 120 can determine whether or not a mismatch has occurred using Table 1 below.

[0112] [Table 1]

[0113] In Table 1, e may mean the value obtained by adding the total width of the top of the substrate (e.g., W1+W2+W3+W4+W5) and the total width of the back of the substrate (e.g., X1+X2+X3+X4+X5) and dividing by 2. The processor 120 can determine whether or not to reflect e depending on the type and characteristics of the substrate. According to the embodiment, the processor 120 can determine that a mismatch has occurred if the difference between the calculated width differences M1, M2, M3, and M4 is greater than or equal to a reference value.

[0114] Figure 8 shows a second candidate control value derived by a predictive control device according to one embodiment disclosed in this application.

[0115] Referring to Figure 8, the processor 120 of the predictive control device 100 can derive a second candidate control value including the slurry pump RPM (rpm) or die spacing. The processor 120 can derive a second candidate control value using slurry temperature, die spacing, and slurry pump RPM (Pump rpm) as control factors. According to the embodiments disclosed herein, the die spacing may include left gap (L Gap) and right gap (R Gap). The processor 120 can derive a second candidate control value using the current value and adjustment range of each control factor.

[0116] As shown in Figure 8, the processor 120 can derive second candidate control values ​​by fixing the slurry temperature and providing adjustment ranges for the left spacing, right spacing, and slurry pump RPM. According to the embodiment, the processor 120 can adjust the left spacing, right spacing, and / or slurry pump RPM within the adjustment range. According to the embodiment shown in Figure 8, the processor 120 can adjust the left spacing and / or right spacing by ±1 μm, and / or adjust the slurry pump RPM by ±1. In that case, the processor 120 can derive 27 second candidate control values.

[0117] However, Figure 8 represents one embodiment of the second candidate control value derived by the processor 120, and the candidate control value derived by the processor 120 is not limited to the value shown in Figure 8. According to the embodiment, the processor 120 can be set to have different adjustment ranges or units for the left spacing, right spacing, and / or RPM of the slurry pump.

[0118] The processor 120 can predict a second quality using a second prediction model for a second candidate control value, and then predict a second optimal control value based on that second quality.

[0119] According to this embodiment, the processor 120 can perform filtering on the second candidate control values. According to this embodiment, the processor 120 can minimize the number of second candidate control values ​​by filtering. In this case, the processor 120 can predict the second quality using a second prediction model for the second candidate control values ​​that have been minimized by filtering, and predict the second optimal control value based on the second quality.

[0120] Figure 9 is a diagram illustrating the operation by which a predictive control device according to one embodiment disclosed in this application derives a second optimal control value from a second candidate control value.

[0121] Referring to Figure 9, the processor 120 can predict the quality of the second candidate control value using a second prediction model and derive a second optimal control value based on the second quality.

[0122] The processor 120 can predict the second quality of the second candidate control value using a second prediction model. The second quality may include, but is not limited to, the loading amount, the width of the blank area, and / or the width of the area.

[0123] According to one embodiment, the processor 120 can use a second prediction model to calculate predicted values ​​for the loading amount, the width of the blank area, and / or the width of the area for a second candidate control value. According to another embodiment, the processor 120 can predict a second quality based on the predicted values ​​for the loading amount, the width of the blank area, and / or the width of the area calculated using the second prediction model.

[0124] The processor 120 can derive a second optimal control value based on the second quality. According to the embodiment, the processor 120 can derive the second candidate control value as the second optimal control value in which the predicted width of the area with ground and the predicted width of the area without ground are within the normal range, and the predicted loading amount is closest to the target value. For example, the processor 120 can derive the second candidate control value as the second optimal control value in which both the predicted width of the area with ground and the predicted width of the area without ground are 100, and the predicted loading amount is the target value of 100.

[0125] According to one embodiment, the processor 120 can adjust the die spacing or the RPM of the slurry pump using the derived second optimal control value. In this case, the slurry temperature may be fixed to the slurry temperature adjusted by the processor 120 using the optimal control value, referring to Figures 5 and 6.

[0126] According to the embodiment shown in Figure 9, both the predicted width of the area with ground and the predicted width of the area without ground are 100, and using a second optimal control value that makes the predicted loading amount equal to the target value of 100, the processor 120 can adjust the left spacing to 99 μm, the right spacing to 99 μm, and the RPM of the slurry pump to 1199.

[0127] According to embodiments disclosed herein, the processor 120 can use a prediction model to predict the quality of candidate control values ​​and derive an optimal control value, and can use a second prediction model to predict the second quality of the second candidate control value and derive a second optimal control value. However, not limited to this embodiment, the processor 120 can use an integrated prediction model that combines the prediction model and the second prediction model to predict integrated quality, and derive an integrated optimal control value to achieve the final correction target of the prediction control device 100 based on the predicted integrated quality. In this case, the integrated quality may include central loading amount data, left and right area loading amount data and / or loading deviation, loading amount, blank area width and / or area width.

[0128] Figure 10 is a flowchart showing the operation method of a predictive control device according to one embodiment disclosed in this application.

[0129] Referring to Figure 10, the operation method of the predictive control device 100 may include the steps of: acquiring data (S100) on the base material onto which the slurry is loaded using a coating die; analyzing the slurry loading characteristics based on the data (S110); deriving candidate control values ​​including the slurry temperature in response to the analysis results of the slurry loading characteristics (S120); predicting the quality of the candidate control values ​​using a predictive model (S130); and deriving an optimal control value based on the quality (S140). According to the embodiment, the operation method of the predictive control device 100 may further include the step of adjusting the slurry temperature using the optimal control value (S150).

[0130] The following will provide a detailed explanation of steps S100 to S150 with reference to Figures 1 to 6, and any overlapping content will be omitted or briefly explained.

[0131] In step S100, the data acquisition unit 110 can acquire data from the user terminal 300. The data acquired by the data acquisition unit 110 may include data relating to the substrate on which the slurry is loaded using the coating die. According to the embodiment, the data may also include slurry loading amount data relating to the amount of slurry applied to the substrate. The slurry loading amount data may include central loading amount data and left and right loading amount data.

[0132] In step S110, the processor 120 can analyze the slurry loading characteristics based on the data acquired by the data acquisition unit 110. According to one embodiment, the slurry loading characteristics can analyze the uniformity of the slurry loading amount. For example, the processor 120 can analyze the uniformity of the slurry loading amount based on the slurry loading amount data acquired by the data acquisition unit 110. According to another embodiment, the processor 120 can analyze the uniformity of the slurry loading amount by calculating a loading deviation based on the slurry loading amount data. For example, the processor 120 can analyze the uniformity of the slurry loading amount based on whether the loading deviation calculated based on the slurry loading amount data is close to zero.

[0133] In step S120, the processor 120 can derive candidate control values ​​in response to the results of analyzing the slurry loading characteristics. According to the embodiment, the candidate control values ​​may include the slurry temperature. For example, the processor 120 can derive candidate control values ​​including the slurry temperature in response to the results of analyzing the slurry loading characteristics. The processor 120 can set a control target for the predictive control device 100 in response to the results of analyzing the slurry loading characteristics. According to the embodiment, the processor 120 can set a correction related to a major quality element Y as the control target for the predictive control device 100. For example, the processor 120 can analyze the uniformity of the slurry loading amount and, in response to the analysis results, set the control target for the predictive control device 100 to a correction related to the loading pattern. In that case, the processor 120 can derive candidate control values ​​for correcting the loading pattern.

[0134] In step S130, the processor 120 can predict the quality of the candidate control value. According to the embodiment, the processor 120 can predict the quality of the candidate control value using a prediction model. According to the embodiment, the prediction model may be modeled by learning the correlation between slurry temperature and slurry loading characteristics. According to the embodiment, the quality may include, but is not limited to, central loading amount data, left and right ground loading amount data and / or loading deviation.

[0135] In step S140, the processor 120 can derive an optimal control value. According to the embodiment, the processor 120 can derive an optimal control value based on the quality predicted in step S130. According to the embodiment, the processor 120 can derive the predicted central loading amount data, the predicted left and right loading amount data, or a candidate control value in which the loading deviation is closest to the target value as the optimal control value. For example, referring to Figure 6, the processor 120 can derive the candidate control value in which the predicted loading deviation is closest to the target value of 0 as the optimal control value.

[0136] In step S150, the processor 120 can adjust the slurry temperature. According to this embodiment, the processor 120 can adjust the slurry temperature using the optimal control value derived in step S140.

[0137] Figure 11 is a flowchart showing the operation method of a predictive control device according to one embodiment disclosed in this application.

[0138] Referring to Figure 11, the operation method of the predictive control device 100 may further include the steps of: acquiring second data relating to the substrate (S200); determining whether or not a mismatch has occurred based on the second data (S210); deriving a second candidate control value including the RPM (rpm) of the slurry pump or the die spacing in response to the occurrence of a mismatch (S220); predicting a second quality of the second candidate control value using a second predictive model (S230); and deriving a second optimal control value based on the second quality (S240). According to the embodiment, the operation method of the predictive control device 100 may further include the step of adjusting the RPM of the slurry pump or the die spacing using the second optimal control value (S250).

[0139] The following will provide a detailed explanation of steps S200 to S250 with reference to Figures 1 to 10, and any overlapping content will be omitted or briefly explained.

[0140] In step S200, the data acquisition unit 110 can acquire second data from the user terminal 300. The second data may include data relating to the substrate to which the slurry has been applied. According to this embodiment, the data acquisition unit 110 can perform step S200 after the processor 120 has performed step S150 in Figure 10. The second data may include data relating to the substrate after the slurry temperature has been adjusted. For example, the second data may include data relating to the substrate to which the slurry has been applied after the slurry temperature has been adjusted in step S150. According to this embodiment, the second data may include image data obtained by converting the width of the textured portion of the substrate and the width of the plain portion of the substrate into an image.

[0141] In step S210, the processor 120 can determine whether or not a mismatch has occurred. According to the embodiment, the processor 120 can determine whether or not a mismatch has occurred based on the second data. According to the embodiment, the processor 120 can calculate whether or not a mismatch has occurred using the width lengths of the top and / or back of two different opposing surfaces of the substrate. For example, the processor 120 can calculate width differences M1, M2, M3, and M4 using the widths W1, W3, W5 of the plain portion and W2, W4 of the textured portion of the top of the substrate and the widths X1, X3, X5 of the plain portion and X2, X4 of the textured portion of the back of the substrate, and determine whether or not a mismatch has occurred. According to the embodiment disclosed herein, the processor 120 can determine that a mismatch has occurred if the difference between the calculated width differences M1, M2, M3, and M4 is greater than or equal to a reference value.

[0142] In step S220, the processor 120 can derive a second candidate control value in response to whether or not a mismatch has occurred. According to the embodiment, the second candidate control value may include the RPM (rpm) of the slurry pump or the die spacing. For example, the processor 120 can derive a second candidate control value including the RPM of the slurry pump or the die spacing in response to whether or not a mismatch has occurred.

[0143] In step S230, the processor 120 can predict the second quality of the second candidate control value. According to the embodiment, the processor 120 can predict the second quality of the second candidate control value using a second prediction model. According to the embodiment, the second quality may include, but is not limited to, the loading amount, the width of the blank area, and / or the width of the area with a background.

[0144] In step S240, the processor 120 can derive a second optimal control value. According to the embodiment, the processor 120 can derive a second optimal control value based on the second quality predicted in step S230. According to the embodiment, the processor 120 can derive as the second optimal control value a second candidate control value from among the second candidate control values ​​in which the predicted width of the area and the predicted width of the area without area are within the normal range, and the predicted loading amount is closest to the target value. For example, referring to Figure 9, the processor 120 can derive as the second optimal control value a second candidate control value in which both the predicted width of the area and the predicted width of the area without area are 100, and the predicted loading amount is the target value of 100.

[0145] In step S250, the processor 120 can adjust the RPM or die spacing of the slurry pump. According to one embodiment, the processor 120 can adjust the RPM or die spacing of the slurry pump using the second optimal control value derived in step S240.

[0146] According to the embodiments disclosed herein, when the processor 120 uses an integrated prediction model that combines a prediction model and a second prediction model, the processor 120 can perform steps S100 to S150 and steps S200 to S250 simultaneously.

[0147] The terms "includes," "constitutes," and "possesses," as used above, mean, unless otherwise specified, that the component in question may be inherent, and therefore should be interpreted as potentially including other components rather than excluding them. All terms, including technical and scientific terms, have the same meaning as generally understood by a person of ordinary skill in the art to which the embodiments disclosed herein belong, unless otherwise defined. Commonly used terms, such as those defined in dictionaries, should be interpreted in accordance with their meaning in the context of the relevant technology, and should not be interpreted in an ideal or overly formal sense unless explicitly defined herein.

[0148] The above description is merely illustrative of the technical concept disclosed herein, and any person with ordinary skill in the art to which the embodiments disclosed herein belong can make various modifications and variations without departing from the essential characteristics of the embodiments disclosed herein. Therefore, the embodiments disclosed herein are for illustrative purposes only, not to limit the technical concept of the embodiments disclosed herein, and the scope of the technical concept disclosed herein is not limited by such embodiments. The scope of protection of the technical concept disclosed herein shall be interpreted in accordance with the attached claims, and all technical concepts within an equivalent scope shall be interpreted as being included in the scope of rights of this application.

Claims

1. A data acquisition unit that acquires data on the substrate onto which the slurry is loaded using a coating die, A processor that analyzes slurry loading characteristics based on the aforementioned data, derives candidate control values ​​including slurry temperature in response to the analysis results of the slurry loading characteristics, predicts the quality of the candidate control values ​​using a predictive model, and derives an optimal control value based on the quality, The predictive control device is modeled by learning the correlation between the slurry temperature and the slurry loading characteristics.

2. The aforementioned data includes data on the amount of slurry loaded from the substrate. The predictive control device according to claim 1, wherein the processor analyzes the uniformity of the slurry loading amount based on the slurry loading amount data.

3. The predictive control device according to claim 2, wherein the slurry loading amount data includes central loading amount data and left and right ground area loading amount data.

4. The predictive control device according to claim 1, wherein the processor adjusts the slurry temperature using the optimal control value.

5. The data acquisition unit acquires second data relating to the substrate after the processor has adjusted the slurry temperature. The processor determines whether a mismatch has occurred based on the second data, derives a second candidate control value including the RPM (rpm) of the slurry pump or the die spacing in response to the occurrence of the mismatch, predicts a second quality of the second candidate control value using a second prediction model, and derives a second optimal control value based on the second quality. The predictive control device according to claim 4, wherein the second predictive model is modeled by learning the correlation between the RPM (rpm) of the slurry pump or the die spacing and the slurry loading characteristics.

6. The second data includes image data obtained by converting the width of the textured portion of the substrate and the width of the plain portion of the substrate into an image, The predictive control device according to claim 5, wherein the processor determines whether or not the mismatch has occurred based on the image data.

7. The predictive control device according to claim 5, wherein the processor adjusts the RPM of the slurry pump or the die spacing using the second optimal control value.

8. A step of acquiring data on the substrate onto which the slurry is loaded using a coating die, The steps include analyzing slurry loading characteristics based on the aforementioned data, The steps include: deriving candidate control values, including slurry temperature, in response to the analysis results of the slurry loading characteristics; The steps include predicting the quality of the candidate control values ​​using a predictive model and deriving an optimal control value based on the quality, The predictive control device operation method is modeled by learning the correlation between the slurry temperature and the slurry loading characteristics, as described above.

9. The aforementioned data includes data on the amount of slurry loaded from the substrate. A method for operating a predictive control device according to claim 8, further comprising the step of analyzing the uniformity of the slurry loading amount based on the slurry loading amount data.

10. The method for operating a predictive control device according to claim 9, wherein the slurry loading amount data includes central loading amount data and left and right ground loading amount data.

11. The method for operating a predictive control device according to claim 8, further comprising the step of adjusting the slurry temperature using the optimal control value.

12. A step of acquiring second data relating to the substrate after adjusting the slurry temperature, A step of determining whether or not a mismatch has occurred based on the second data mentioned above, The steps include: deriving a second candidate control value, including the RPM (rpm) of the slurry pump or the die spacing, in response to the occurrence of the aforementioned mismatch; A step of predicting the second quality of the second candidate control value using a second prediction model, The process further includes the step of deriving a second optimal control value based on the second quality, The method for operating a predictive control device according to claim 11, wherein the second predictive model is modeled by learning the correlation between the RPM (rpm) of the slurry pump or the die spacing and the slurry loading characteristics.

13. The second data includes image data obtained by converting the width of the textured portion of the substrate and the width of the plain portion of the substrate into an image, A method for operating a predictive control device according to claim 12, further comprising the step of determining whether or not the mismatch has occurred based on the image data.

14. The method for operating a predictive control device according to claim 12, further comprising the step of adjusting the RPM of the slurry pump or the die spacing using the second optimal control value.

Images