Estimation device, learning device, inspection method, and manufacturing method

Abstract

To improve accuracy in estimating characteristics of batteries.SOLUTION: An estimation device 100 disclosed herein is configured to estimate battery characteristics based on first data and a first estimation model 111. The first data includes at least one of weight of a compressed electrode body, thickness of the compressed electrode body, and weight of a case electrode body.SELECTED DRAWING: Figure 5

Description

[Technical Field]

[0001] The present disclosure relates to an estimation device, a learning device, an inspection method, and a manufacturing method. [Background technology]

[0002] For example, Japanese Patent Application Laid-Open No. 2014-71103 (Patent Document 1) discloses a prediction device for estimating the discharge capacity of a battery. This prediction device estimates the discharge capacity using measured values ​​for each item of each material produced in the manufacturing process of each battery material. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2014-71103 Summary of the Invention [Problem to be solved by the invention]

[0004] In an estimation device that estimates battery characteristics, such as discharge capacity, there is a demand for improving the accuracy of estimating battery characteristics.

[0005] The present disclosure has been made to solve the above-mentioned problems, and its purpose is to improve the accuracy of estimating battery characteristics of a battery. [Means for solving the problem]

[0006] (Item 1) The estimation device of the present disclosure is an estimation device that estimates battery characteristics of a battery. The battery manufacturing process includes fabricating a positive electrode by adding a positive electrode composite to a positive electrode substrate, fabricating a negative electrode by adding a negative electrode composite to a negative electrode substrate, fabricating an electrode assembly using a positive electrode, a negative electrode, and a separator, compressing the electrode assembly to fabricate a compressed electrode assembly, housing the compressed electrode assembly in a case to fabricate a case electrode assembly, and manufacturing a battery by sealing an electrolyte in the case electrode assembly. The estimation device includes an interface that accepts input of first data, a memory that stores a first estimation model, and a calculation device that estimates battery characteristics based on the first data and the first estimation model. The first data includes at least one of the weight of the compressed electrode assembly, the thickness of the compressed electrode assembly, and the weight of the case electrode assembly.

[0007] (Item 2) In the estimation device according to item 1, the manufacturing process of the battery further includes compressing the case electrode body to produce a compressed case electrode body. Manufacturing the battery includes sealing an electrolyte in the compressed case electrode body to manufacture the battery. The first data includes a thickness of the compressed case electrode body.

[0008] (Item 3) In the estimation device according to item 1 or 2, the memory further stores a second estimation model. The calculation device estimates second data based on the first data and the second estimation model, and estimates battery characteristics based on the first data, the second data, and the first estimation model. The second data includes at least one of a positive electrode parameter related to the basis weight of the positive electrode composite, a negative electrode parameter related to the basis weight of the negative electrode composite, a specific surface area of ​​the positive electrode composite, a specific surface area of ​​the negative electrode composite, an adhesion strength between the positive electrode substrate and the positive electrode composite, and an adhesion strength between the negative electrode substrate and the negative electrode composite.

[0009] (4) In the estimation device according to any one of paragraphs 1 to 3, the battery characteristics include at least one of a discharge capacity of the battery, a DCIR of the battery, a thickness of the battery, and a hardness of the battery.

[0010] (Clause 5) In the manufacturing process of the estimation device battery described in any one of clauses 1 to 4, batteries of multiple types are manufactured. The memory stores a first estimation model for each battery type. The interface accepts input of the battery type. The calculation device estimates battery characteristics of the battery indicated by the battery type based on the first estimation model corresponding to the battery type and the first data.

[0011] (Item 6) The learning device of the present disclosure is a learning device for a first estimation model used to estimate battery characteristics. The battery manufacturing process includes fabricating a positive electrode by adding a positive electrode composite to a positive electrode substrate, fabricating a negative electrode by adding a negative electrode composite to a negative electrode substrate, fabricating an electrode assembly using a positive electrode, a negative electrode, and a separator, compressing the electrode assembly to fabricate a compressed electrode assembly, housing the compressed electrode assembly in a case to fabricate a case electrode assembly, and manufacturing a battery by sealing an electrolyte in the case electrode assembly. The learning device includes an interface that accepts input of first learning data, which is a pair of actual measurement data and learning battery characteristics, a memory that stores the first estimation model, and an update device that updates the first estimation model based on the first learning data. The actual measurement data includes at least one of the weight of the compressed electrode assembly, the thickness of the compressed electrode assembly, and the weight of the case electrode assembly.

[0012] (Item 7) In the learning device described in item 6, the manufacturing process of the battery further includes compressing the case electrode body to produce a compressed case electrode body. Manufacturing the battery means manufacturing the battery by sealing an electrolyte in the compressed case electrode body. The measured data includes the thickness of the compressed case electrode body.

[0013] (Item 8) In the learning device according to item 6 or 7, the memory further stores a second estimation model, and the interface accepts input of second learning data in which actual measurement data and learning estimation data are paired. The learning device generates an updated estimation model by updating the second estimation model based on the second learning data, and estimates estimation data based on the actual measurement data and the updated estimation model. The first learning data is data in which the actual measurement data and estimation data are paired with learning battery characteristics. The learning estimation data includes at least one of a positive electrode parameter related to the basis weight of the positive electrode composite, a negative electrode parameter related to the basis weight of the negative electrode composite, a specific surface area of ​​the positive electrode composite, a specific surface area of ​​the negative electrode composite, an adhesion strength between the positive electrode substrate and the positive electrode composite, and an adhesion strength between the negative electrode substrate and the negative electrode composite.

[0014] (Item 9) A battery inspection method of the present disclosure includes carrying out a manufacturing process and estimating battery characteristics of a battery manufactured in the manufacturing process using the estimation device described in any one of Items 1 to 5. The inspection method further includes determining the battery as an acceptable product if the battery characteristics satisfy predetermined standards, and determining the battery as a rejected product if the battery characteristics do not satisfy the predetermined standards.

[0015] (Item 10) The battery inspection method of the present disclosure is a battery manufacturing method, which comprises carrying out the inspection method described in item 9, carrying out a final process on batteries that are determined to be acceptable, and not carrying out the final process on batteries that are determined to be unacceptable. [Effects of the Invention]

[0016] According to the present disclosure, it is possible to improve the accuracy of estimating the battery characteristics of a battery. [Brief explanation of the drawings]

[0017] [Figure 1] 3 is a flowchart of a method for manufacturing a battery according to the present embodiment. [Figure 2] 1 is a flowchart of a battery manufacturing process. [Figure 3]FIG. 2 is a diagram for explaining various data. [Figure 4] 1 is a diagram illustrating an example of a hardware configuration of an estimation system according to an embodiment of the present invention. [Figure 5] FIG. 1 is a functional block diagram of an estimation device according to a first embodiment. [Figure 6] FIG. 2 is a functional block diagram of the learning device of the first embodiment. [Figure 7] FIG. 10 is a functional block diagram of an estimation device according to a second embodiment. [Figure 8] FIG. 10 is a diagram for explaining a learning method performed by the learning device of the second embodiment. [Figure 9] FIG. 10 is a functional block diagram of an estimation device according to a third embodiment. [Figure 10] FIG. 10 is a diagram for explaining the estimation accuracy of the estimation device of the present disclosure. DETAILED DESCRIPTION OF THE INVENTION

[0018] Hereinafter, embodiments of the present disclosure will be described in detail with reference to the drawings. In the drawings, the same or corresponding parts are designated by the same reference numerals, and description thereof will not be repeated.

[0019] First Embodiment [Battery manufacturing method] 1 is a flowchart of a battery manufacturing method according to the present embodiment. The battery manufacturing method according to the present embodiment is carried out by at least one of a worker (also referred to as a user) and a manufacturing device.

[0020] In step S2, a battery manufacturing process is carried out. Details of the battery manufacturing process are described in FIG. 2. Next, in step S4, the estimation device of the present disclosure is used to estimate the battery characteristics of the battery manufactured in the battery manufacturing process. The battery characteristics indicate, for example, the characteristics of the battery manufactured in step S2. More specifically, the battery characteristics are the discharge capacity of the battery.

[0021] Next, in step S6, it is determined whether the battery characteristics estimated in step S4 satisfy predetermined criteria. For example, if the battery characteristics are the discharge capacity of the battery, the predetermined criteria include a normal range for the discharge capacity. If the estimated discharge capacity falls within the normal range, it is determined that the battery characteristics satisfy the predetermined criteria. If the estimated discharge capacity does not fall within the normal range, it is determined that the battery characteristics do not satisfy the predetermined criteria.

[0022] If the battery characteristics satisfy the predetermined criteria (YES in step S6), the process proceeds to step S8. On the other hand, if the battery characteristics do not satisfy the predetermined criteria (NO in step S6), the process proceeds to step S12.

[0023] In step S8, the batteries manufactured in the battery manufacturing process in step S2 are determined to be acceptable. Next, in step S10, the final process is performed. In the final process, for example, a battery is randomly selected from a predetermined number of batteries determined to be acceptable. An experiment is then performed to detect the battery characteristics of the selected battery. If the battery characteristics detected by this experiment meet predetermined standards, all of the predetermined number of batteries are determined to be genuine acceptable. That is, in the example of FIG. 1, double testing of the batteries is performed.

[0024] In step S12, the battery manufactured in the battery manufacturing process in step S2 is determined to be a rejected product, and the battery determined to be a rejected product is then discarded, for example.

[0025] [Battery manufacturing process] Fig. 2 is a flowchart of the battery manufacturing process in step S2 of Fig. 1. Note that the parentheses in Fig. 2 indicate that first data, third data, and actual measurement data are acquired and input to an estimation device, which will be described later.

[0026] First, in step S22, a positive electrode is fabricated by adding a positive electrode slurry to a positive electrode substrate. Specifically, the positive electrode slurry added to the positive electrode substrate is dried to form a positive electrode composite, and then the positive electrode composite added to the positive electrode substrate is compressed to fabricate the positive electrode. In other words, the positive electrode is fabricated by adding the positive electrode composite to the positive electrode substrate.

[0027] In step S22, the operator can obtain thickness measurements of the positive electrode substrate, the positive electrode slurry, and the positive electrode composite. The "measurements" in the present disclosure may refer to values ​​measured by the operator using a measuring tool or may refer to values ​​measured by a measuring device. The thickness measurements of the positive electrode substrate, the positive electrode slurry, and the positive electrode composite are, for example, measurements taken immediately after the positive electrode slurry is applied to the positive electrode substrate, measurements taken after the positive electrode slurry is dried to form the positive electrode composite, or measurements taken after the positive electrode composite is compressed.

[0028] Next, in step S24, the negative electrode slurry is applied to the negative electrode substrate to fabricate the negative electrode. Specifically, the negative electrode slurry applied to the negative electrode substrate is dried to form a negative electrode composite, and then the negative electrode composite applied to the negative electrode substrate is compressed to fabricate the negative electrode. In other words, the negative electrode is formed by applying the negative electrode composite to the negative electrode substrate.

[0029] In step S24, the operator can obtain any one of the thickness measurements of the negative electrode substrate, the negative electrode slurry, and the negative electrode composite. The thickness measurements of the negative electrode substrate, the negative electrode slurry, and the negative electrode composite are, for example, any one of a measurement taken immediately after the negative electrode slurry is added to the negative electrode substrate, a measurement taken after the negative electrode slurry is dried to form the negative electrode composite, and a measurement taken after the negative electrode composite is compressed.

[0030] Next, in step S26, an electrode assembly is produced from a positive electrode, a negative electrode, and a separator. Here, the electrode assembly may be produced, for example, by stacking the positive electrode, the negative electrode, and the separator. Alternatively, the electrode assembly may be produced, for example, by winding the positive electrode, the negative electrode, and the separator. In step S26, the operator can also obtain a thickness measurement of the separator.

[0031] Next, in step S28, the electrode body is compressed to produce a compressed electrode body. Also, in this step S28, the worker can obtain the thickness measurement value of the compressed electrode body and the weight measurement value of the compressed electrode body.

[0032] Next, in step S30, the compressed electrode body is housed in a case to create a case electrode body. Also in this step S30, the worker can obtain a weight measurement value of the case electrode body.

[0033] Next, in step S32, the case electrode body is compressed to produce a compressed case electrode body. Also, in this step S32, the worker can obtain a thickness measurement value of the compressed case electrode body.

[0034] Next, in step S34, an electrolyte is injected into the gaps in the compressed case electrode body and sealed, completing the battery and completing the battery manufacturing process.

[0035] The case electrode body may be in a state before the electrolyte is poured in. The case electrode body may have an opening, and the opening may be closed by a sealing plate.

[0036] [Explanation of each data] Next, various types of data described in the present disclosure will be described. FIG. 3 is a diagram for explaining various types of data. FIG. 3(A) is a diagram for explaining an example of first data and actual measurement data. As will be described later, the first data is data used when an estimation device 100 (described later) estimates battery characteristics of a battery. The actual measurement data is data used when a learning device 200 (described later) learns estimation models (first estimation model 211, second estimation model 212).

[0037] The first data includes at least one of the weight of the compressed electrode body, the thickness of the compressed electrode body, the weight of the case electrode body, and the thickness of the compressed case electrode body. The weight of the compressed electrode body and the thickness of the compressed electrode body are values ​​that can be acquired in step S28. The weight of the case electrode body is acquired in step S30. The thickness of the compressed case electrode body is a value that can be acquired in step S32.

[0038] FIG. 3(B) will be described in the second embodiment below. FIG. 3(C) is a diagram for explaining the third data. The third data is at least one of the thickness of the positive electrode, the thickness of the negative electrode, and the thickness of the separator. The thickness of the positive electrode may be replaced with the thickness of the positive electrode composite material. Furthermore, the thickness of the negative electrode may be replaced with the thickness of the negative electrode composite material. The thickness of the positive electrode or the thickness of the positive electrode composite material is a value that can be obtained in step S22. The thickness of the negative electrode or the thickness of the negative electrode composite material is a value that can be obtained in step S24. The first data and the third data are data that can be obtained by an operator without destroying the battery.

[0039] FIG. 3(D) is a diagram showing an example of battery characteristics estimated by the estimation device 100 described below. The battery characteristics are at least one of the battery discharge capacity, the battery's DCIR (Direct Current Internal Resistance), the battery's thickness, and the battery's hardness. The battery's DCIR is the direct current internal resistance of the battery. The "battery" in the battery thickness and battery hardness refers to the battery after the processing of step S34 in FIG. 2 has been performed. Furthermore, the battery hardness is determined, for example, by compressing the battery with two or more levels of stress and observing the slope of the stress-strain diagram (so-called Young's modulus).

[0040] The first data or the third data may also include at least one of the force, the electric power, and the displacement of a predetermined member of a manufacturing device (processing equipment) for manufacturing the battery.

[0041] The positive electrode has a thickness of 144 μm, the negative electrode has a thickness of 192 μm, the separator has a thickness of 14.7 μm, and the compressed electrode body has a thickness of 12.2 mm.

[0042] In addition, in Figure 3(A), the actual measurement data is data of the same type as the first data. For example, if the first data is the weight of the compressed electrode body, the actual measurement data is also the weight of the compressed electrode body. Furthermore, if the first data is the thickness of the compressed electrode body and the thickness of the case electrode body, the actual measurement data is also the thickness of the compressed electrode body and the thickness of the case electrode body. Similarly, in Figure 3(D), the battery characteristics and the learning battery characteristics are data of the same type.

[0043] [Estimation System] 4 shows an example of the hardware configuration of the estimation system 10 of this embodiment. The estimation system 10 includes an estimation device 100 and a learning device 200.

[0044] The estimation device 100 includes a calculation device 102, a memory 104, and an interface 106. The learning device 200 includes an update device 202, a memory 204, and an interface 206.

[0045] The arithmetic device 102 and the update device 202 are configured with a CPU (Central Processing Unit), an FPGA (Field-Programmable Gate Array), a GPU (Graphics Processing Unit), etc. The arithmetic device 102 and the update device 202 may be configured with at least one of a CPU, an FPGA, and a GPU. The arithmetic device 102 and the update device 202 may also be configured with a processing circuitry. The arithmetic device 102 and the update device 202 are also referred to as "at least one processor" or "arithmetic circuitry."

[0046] The memory 104 includes a volatile storage area (for example, a working area) that temporarily stores program code, work memory, etc. when the arithmetic device 102 executes any program. Similarly, the memory 204 includes a volatile storage area (for example, a working area) that temporarily stores program code, work memory, etc. when the update device 202 executes any program.

[0047] The memory 104 and the memory 204 include, for example, a RAM (Random Access Memory) and a ROM (Read Only Memory). The ROM stores programs executed by the arithmetic device 102. The RAM temporarily stores data generated by the execution of programs in the arithmetic device 102 or the update device 202. The RAM can function as a temporary data memory used as a working area.

[0048] The memory 104 stores a trained first estimation model 111. The memory 204 stores a first estimation model 211. The second estimation model 112 and the second estimation model 212 will be described in a second embodiment.

[0049] The interface 106 is configured to communicate with devices external to the estimator 100. These external devices include the learning device 200, an input device (not shown), and a display device (not shown). The interface 206 is also configured to communicate with devices external to the learning device 200 (e.g., the estimator 100).

[0050] The first data 151 and the third data 153 (see FIG. 3) are input by the user via the input device to the estimation device 100. As a modified example, the third data 153 may not be used.

[0051] The arithmetic device 102 of the estimation device 100 estimates and outputs battery characteristics 120 based on the first data 151, the third data 153, and the first estimation model 111. More specifically, the arithmetic device 102 applies the first data 151 and the third data to the first estimation model 111 to output the battery characteristics 120. The display device displays an image showing the output battery characteristics 120.

[0052] The data (first data 151 and third data 153) input to estimation device 100 are also referred to as explanatory variables. Battery characteristics 120 estimated by estimation device 100 are also referred to as "objective variables."

[0053] The learning device 200 typically learns (trains) the first estimation model 211 by so-called supervised learning. First learning data (teaching data) is input to the learning device 200 by a user. In the first embodiment, the first learning data is data in which actual measurement data 251 and learning battery characteristics 252 are paired. This data is also referred to as the actual measurement data 251 and learning battery characteristics 252 corresponding to each other.

[0054] The actual measurement data 251 is so-called example data, and the learning battery characteristic 252 is so-called correct answer data. The actual measurement data 251 is the data described in FIG. 3(A). The learning battery characteristic 252 is a battery characteristic actually measured by an operator for the battery from which the actual measurement data 251 was acquired. For example, if the battery characteristic is the discharge capacity of a battery, the operator acquires (measures) the discharge capacity of the battery from which the actual measurement data 251 was acquired as the learning battery characteristic 252 through an experiment or the like.

[0055] The update device 202 uses at least one piece of first training data to train the first estimation model 211. Training the first estimation model 211 typically includes updating parameters included in the first estimation model 211.

[0056] When the updating of the first estimation model 211 is completed, the updating device 202 reflects the updated first estimation model 211 in the first estimation model 111 (see the dashed arrow in FIG. 4). Therefore, the estimation device 100 can hold the latest first estimation model 111.

[0057] 4 discloses a configuration in which the estimation device 100 and the learning device 200 are separate devices (information processing devices). However, as a modified example, the estimation device 100 and the learning device 200 may be included in a single information processing device. The discharge capacity of the battery from which the actual measurement data 251 is acquired is acquired as learning battery characteristics 252 by an experiment or the like.

[0058] [Functional block diagram of the estimation device 100] 5 is a functional block diagram of the estimation device 100. The estimation device 100 mainly includes a first estimation unit 161. The first estimation unit 161 corresponds to the calculation device 102 in FIG.

[0059] When the first data 151 and the third data 153 are input to the estimation device 100, the first estimation unit 161 acquires the first data 151 and the third data 153. Then, the first estimation unit 161 applies the first data 151 and the third data to the first estimation model 111 to estimate the battery characteristics 120 and output the battery characteristics 120.

[0060] [Functional block diagram of learning device 200] 6 is a functional block diagram of the learning device 200. The learning device 200 mainly includes a first learning unit 261. The first learning unit 261 corresponds to the update device 202 in FIG.

[0061] When the first learning data is input to the learning device 200, the first learning unit 261 acquires the first learning data. Then, the first learning unit 261 updates the first estimation model 211 using the first learning data.

[0062] [Summary of the First Embodiment] (1) The inventors of the present disclosure discovered that, in the manufacture of a battery, there is a strong correlation between the battery characteristics 120 (FIG. 3(D)) and the first data 151 (FIG. 3(A)). In particular, the inventors discovered that the larger the first data (weight of the compressed electrode body, thickness of the compressed electrode body, weight of the case electrode body, and thickness of the case electrode body after compression), the larger the battery discharge capacity (battery characteristics 120) tends to be. The inventors also discovered that there is a strong correlation between the first data and other battery characteristics 120 (battery DCIR, battery thickness, battery hardness).

[0063] 2, the worker can obtain the "weight of the compressed electrode body, the thickness of the compressed electrode body, the weight of the case electrode body, and the thickness of the case electrode body" during the battery manufacturing process without destroying the battery. Therefore, the worker can accurately estimate battery characteristics 120 without wasting battery data by inputting the first data obtained during the battery manufacturing process into the estimation device.

[0064] (2) As shown in FIG. 3(D), the battery characteristics 120 are information including at least one of four characteristics: the battery discharge capacity, the battery DCIR, the battery thickness, and the battery hardness. Therefore, the operator can obtain the information without conducting an experiment to obtain the information. Note that, when the battery characteristics 120 are configured to include at least two of these four characteristics, the at least two characteristics are estimated as the battery characteristics 120.

[0065] (3) As shown in Fig. 6, the learning device 200 updates the first estimation model 211 (first estimation model 111) using first learning data that pairs actual measurement data and learning battery characteristics, allowing the estimation device 100 to appropriately estimate battery characteristics. Furthermore, an operator can collect actual measurement data during the battery manufacturing process shown in Fig. 2. Furthermore, the operator performs experiments on a battery completed during the battery manufacturing process to obtain learning battery characteristics 252 (actual measurement values) of the battery.

[0066] (4) In addition, in the battery manufacturing method of the present disclosure, as shown in Fig. 1, in step S4, the estimation device 100 estimates battery characteristics and determines whether the battery meets predetermined standards. Therefore, the battery manufacturing method of the present disclosure can prevent an operator from actually measuring the battery characteristics through experiments.

[0067] In addition, an operator may specify a correlation between the weight and thickness of a normal battery and the battery characteristics at the battery design stage, and then use the correlation to determine the weight and thickness of the battery so that the battery characteristics meet predetermined standards.

[0068] However, even if the weight and thickness of the battery are determined in this way, there are cases where the battery characteristics do not meet the predetermined standards due to the inclusion of foreign matter during the manufacturing process, etc. Even in such cases, the estimation device 100 of this embodiment can estimate the battery characteristics, thereby improving convenience for workers.

[0069] Second Embodiment [Estimation device of second embodiment] The estimation device 100A of the second embodiment estimates battery characteristics using not only the first data but also the second data, thereby enabling the estimation device 100A to estimate battery characteristics with even greater accuracy than the estimation device 100 of the first embodiment.

[0070] FIG. 3(B) is a diagram showing the second data, the estimated data and the learning estimated data described below. The inventors discovered that there is a strong correlation between the battery characteristics 120 (FIG. 3(D)) and the second data (FIG. 3(B)). Furthermore, the inventors discovered that the correlation between the second data and the battery characteristics tends to be stronger than the correlation between the first data and the battery characteristics. However, as will be described later, in order to obtain the second data, the battery must be destroyed, resulting in battery waste.

[0071] Therefore, the estimation device 100A of the second embodiment receives input of first data, estimates second data from the first data, and estimates battery characteristics using the input first data and the estimated second data.

[0072] Next, the second data will be described in detail. As shown in FIG. 3(B), the second data includes at least one of a positive electrode parameter related to the basis weight of the positive electrode composite, a negative electrode parameter related to the basis weight of the negative electrode composite, a specific surface area of ​​the positive electrode composite after drying, a specific surface area of ​​the negative electrode composite after drying, an adhesion strength between the positive electrode substrate and the positive electrode composite, and an adhesion strength between the negative electrode substrate and the negative electrode composite. Hereinafter, at least one of the positive electrode composite and the negative electrode composite will also be simply referred to as a "composite." The composite will also be referred to as a "coating."

[0073] The parameter relating to the basis weight is a parameter relating to the basis weight (weight per unit area) of the composite material. The basis weight parameter includes at least one of a value indicating the basis weight itself and a value calculated from the basis weight. The value calculated from the basis weight may be the composite material density. The composite material density is a value calculated by dividing the basis weight of the composite material by the thickness of the composite material.

[0074] Fig. 7 is a functional block diagram of the estimation device 100A. The estimation device 100A has a first estimation unit 161, a second estimation unit 162, a trained first estimation model 111, and a trained second estimation model 112. In this manner, the estimation device 100A uses the second estimation model 112 indicated by the dashed line in Fig. 4. Note that the example in Fig. 7 shows an example in which the third data is not used.

[0075] Furthermore, the first estimation model 111 of the first embodiment is a model that outputs (estimates) battery characteristics from the first data (and the third data), while the first estimation model 111 of the second embodiment is a model that outputs (estimates) battery characteristics from the first data (and the third data) and the second data.

[0076] When the first data 151 is input to the estimation device 100A, the first data 151 is input to a first estimation unit 161 and a second estimation unit 162.

[0077] The second estimation unit 162 estimates the second data by applying the first data 151 to the second estimation model 212. The second data is input to the first estimation unit 161. The first estimation unit 161 estimates the battery characteristics 120 by applying the first data 151 input by the user and the second data 152 estimated by the second estimation unit 162 to the first estimation model 111.

[0078] [Learning device of the second embodiment] 8A and 8B are diagrams illustrating a learning method performed by the learning device 200A of the second embodiment. First, the learning device 200A updates the second estimation model 212 as shown in FIG. 8A. Next, the learning device 200A updates the first estimation model 211 as shown in FIG. 8B.

[0079] First, updating of the second estimation model 212 will be described with reference to Fig. 8(A). Second learning data (teacher data) is input to the learning device 200 by a user. The first learning data is data in which actual measurement data 251 (see Fig. 3(A)) and learning-use estimation data 265 (see Fig. 3(B)) are paired. The actual measurement data 251 is so-called example data, and the learning-use estimation data 265 is so-called correct answer data.

[0080] Here, the learning estimation data is the data shown in FIG. 3(B). As described above, this learning estimation data is a value obtained by destroying a battery. Therefore, there is a concern about the waste of batteries due to the destruction of the batteries to obtain the learning estimation data. However, since the learning estimation data is data used only to update the second estimation model 212, only a small number of batteries need to be destroyed.

[0081] Furthermore, the second learning unit 262 updates the second estimation model 212 using the actual measurement data 251 and the learning estimation data 265. The updated second estimation model 212 corresponds to the “updated estimation model” of the present disclosure.

[0082] Next, updating of the first estimation model 211 will be described with reference to Fig. 8(B). In Fig. 8(B), actual measurement data 251 included in the second learning data and learning battery characteristics 252 corresponding to the actual measurement data 251 are input to the learning device 200A. The actual measurement data 251 in Fig. 8(B) is the same as the actual measurement data 251 in Fig. 8(A).

[0083] The input actual measurement data 251 is output to the first learning section 261 and the second learning section 262. The input learning battery characteristic 252 is output to the first learning section 261.

[0084] The second learning unit 262 applies the actual measurement data 251 to the second estimation model 212 (updated estimation model) to estimate the estimation data 253. Then, the second learning unit 262 outputs the estimation data 253 to the first learning unit 261.

[0085] The first learning unit 261 updates the first estimation model 211 based on the first learning data. Here, the first learning data is data in which the actual measurement data 251 and the estimated data 253 are paired with the learning battery characteristics 252. The actual measurement data 251 and the estimated data 253 are so-called example data, and the learning battery characteristics 252 are so-called correct answer data.

[0086] [Summary of the second embodiment] (1) The estimation device 100A estimates the battery characteristics 120 using not only the first data but also the second data. As described above, the second data tends to have a stronger correlation with the battery characteristics than the first data, but is data that requires destruction of the battery. However, in the second embodiment, the second data is data that can be estimated by the second estimation unit 162 without destroying the battery. Therefore, the estimation device 100A of the second embodiment can improve the accuracy of the battery characteristics compared to the first embodiment, without destroying the battery.

[0087] (2) Furthermore, the learning device 200A can update the first estimation model 211 and the second estimation model 212. Therefore, the estimation device 100A can appropriately estimate battery characteristics.

[0088] Third Embodiment In some cases, multiple types of batteries are manufactured in a battery manufacturing process. In consideration of such cases, the estimation device 100 may store estimation models (first estimation model 111 and second estimation model 112) for each of the multiple types of batteries. In the following, it is assumed that N (N is an integer equal to or greater than 2) types of batteries can be manufactured in the manufacturing process.

[0089] 9 is a functional block diagram of an estimation device 100B according to the third embodiment. The estimation device 100B includes N first estimation models (first estimation models 1111, ..., 111N). Here, the N first estimation models are estimation models corresponding to N types of batteries, respectively.

[0090] Furthermore, the interface of the estimation device 100B accepts input of battery type 155 in addition to first data 151 and third data. When the first estimation unit 161 acquires the battery type 155, it identifies a first estimation model corresponding to the battery type 155 from the N first estimation models. Then, based on the first data 151, third data 153, and the identified first estimation model, it estimates the battery characteristics of the battery indicated by the battery type 155.

[0091] With this configuration, even if a plurality of types of batteries can be manufactured in a battery manufacturing process, the estimation device 100B can estimate the battery characteristics of the battery of the corresponding type.

[0092] <Experimental Results> FIG. 10 is a diagram for explaining the estimation accuracy of the estimation device 100 and the estimation device 100A. FIG. 10 discloses Comparative Examples 1 and 2 and Examples 1 to 3. Comparative Example 1 is an example in which the third data (positive electrode thickness, negative electrode thickness, and separator thickness) are used as explanatory variables. Example 1 is an example in which the explanatory variables of Comparative Example 1 and the thickness of the compressed electrode body of the first data are used as explanatory variables. Example 2 is an example in which the explanatory variables of Example 1 and the estimated composite material area weight of the second data are used as explanatory variables. Example 3 is an example in which the explanatory variables of Example 2 and the estimated composite material density of the second data are used as explanatory variables. Comparative Example 2 is an example in which the explanatory variables of Example 1 and the measured composite material area weight and the estimated composite material density are used as explanatory variables.

[0093] Example 1 is an experimental example using the estimation device 100, and Example 2 is an experimental example using the estimation device 100A. In these experiments, a random forest was used as the estimation algorithm.

[0094] In addition, the standardized error and relative error were calculated for each example. The standardized error and relative error were calculated using a predetermined first calculation formula and a predetermined second calculation formula, respectively. Note that the smaller the standardized error and relative error, the higher the estimation accuracy.

[0095] As shown in FIG. 10, it is shown that Examples 1 to 3 have higher estimation accuracy than Comparative Example 1. Furthermore, it is shown that the estimation accuracy increases in the order of Example 3, Example 2, and Example 1. Furthermore, when Example 3 is compared with Comparative Example 2, the standardized error and relative error are the same. However, in Comparative Example 2, it is necessary to destroy the battery to obtain the actual measurement value and the actual composite density measurement value. It is shown that Example 3 ensures estimation accuracy equivalent to that of Comparative Example 2 without destroying the battery.

[0096] <Modification> In the above-described FIG. 2, a configuration has been described in which a process for compressing the case electrode body is performed in step S32 of the battery manufacturing process. However, step S32 may be a process for measuring the thickness of the case electrode body after compression. In this case, a configuration in which the process of step S32 is not performed may be adopted. In such a configuration, the process of step S34 is to manufacture a battery by sealing an electrolyte solution in the case electrode body produced in step S30. In this case, the "thickness of the case electrode body after compression" in FIG. 3(A) is omitted.

[0097] The embodiments disclosed herein should be considered to be illustrative in all respects and not restrictive. The scope of the present disclosure is defined by the claims, not by the above description, and is intended to include all modifications within the meaning and scope of the claims. [Explanation of symbols]

[0098] 10 Estimation system, 100, 100A, 100B Estimation device, 102 Arithmetic device, 104, 204 Memory, 106, 206 Interface, 111, 211 First estimation model, 112, 212 Second estimation model, 120 Battery characteristics, 151 First data, 152 Second data, 153 Third data, 155 Battery type, 161 First estimation unit, 162 Second estimation unit, 200, 200A Learning device, 202 Update device, 251 Actual measurement data, 252 Learning battery characteristics, 253 Estimation data, 261 First learning unit, 262 Second learning unit, 265 Learning estimation data.

Claims

1. An estimation device for estimating battery characteristics of a battery, The battery manufacturing process is as follows: preparing a positive electrode in which a positive electrode composite is added to a positive electrode substrate; preparing a negative electrode in which a negative electrode composite is added to a negative electrode substrate; preparing an electrode assembly using the positive electrode, the negative electrode, and a separator; compressing the electrode body to produce a compressed electrode body; producing a case electrode body by housing the compressed electrode body in a case; and manufacturing the battery by sealing an electrolyte in the case electrode body, The estimation device includes: an interface that accepts input of first data; a memory that stores a first estimation model; a calculation device that estimates the battery characteristics based on the first data and the first estimation model; The first data is The weight of the compressed electrode body; the thickness of the compressed electrode body; and the weight of the case electrode body.

2. the manufacturing process for the battery further includes compressing the case electrode body to produce a compressed case electrode body; manufacturing the battery means manufacturing the battery by sealing the electrolyte in the compressed case electrode body, The estimation device according to claim 1 , wherein the first data includes a thickness of the case electrode body after compression.

3. The memory further stores a second estimation model; The computing device estimating second data based on the first data and the second estimation model; estimating the battery characteristics based on the first data, the second data, and the first estimation model; The second data is a positive electrode parameter related to the basis weight of the positive electrode mixture; a negative electrode parameter related to the basis weight of the negative electrode composite; a specific surface area of ​​the positive electrode mixture; and a specific surface area of ​​the negative electrode mixture; and Adhesion strength between the positive electrode substrate and the positive electrode composite; and an adhesion strength between the negative electrode substrate and the negative electrode composite material.

4. The battery characteristics are a discharge capacity of the battery; and DCIR (Direct Current Internal Resistance) of the battery; a thickness of the battery; and The estimation device according to claim 1 or claim 2, wherein the estimation device further includes at least one of the following: a hardness of the battery;

5. In the battery manufacturing process, a plurality of types of batteries are manufactured, the memory stores the first estimation model for each type of battery; the interface accepts input of the type of the battery; 3 . The estimation device according to claim 1 , wherein the calculation device estimates the battery characteristics of the battery indicated by the battery type based on the first estimation model corresponding to the battery type and the first data.

6. A learning device for a first estimation model used to estimate battery characteristics of a battery, The battery manufacturing process is as follows: preparing a positive electrode in which a positive electrode composite is added to a positive electrode substrate; preparing a negative electrode in which a negative electrode composite is added to a negative electrode substrate; preparing an electrode assembly using the positive electrode, the negative electrode, and a separator; compressing the electrode body to produce a compressed electrode body; producing a case electrode body by housing the compressed electrode body in a case; and manufacturing the battery by sealing an electrolyte in the case electrode body, The learning device an interface that receives input of first learning data in which actual measurement data and learning battery characteristics are paired; a memory that stores the first estimation model; an updating device that updates the first estimation model based on the first learning data; The measured data is The weight of the compressed electrode body; the thickness of the compressed electrode body; and the weight of the case electrode body.

7. the manufacturing process for the battery further includes compressing the case electrode body to produce a compressed case electrode body; manufacturing the battery means manufacturing the battery by sealing the electrolyte in the compressed case electrode body, The learning device according to claim 6 , wherein the actual measurement data includes a thickness of the case electrode body after compression.

8. The memory further stores a second estimation model; the interface accepts input of second learning data in which the actual measurement data and learning estimation data are paired; The learning device updating the second estimation model based on the second training data to generate an updated estimation model; estimating estimation data based on the actual measurement data and the updated estimation model; the first learning data is data in which the actual measurement data and the estimated data are paired with the learning battery characteristics, The learning estimation data is a positive electrode parameter related to the basis weight of the positive electrode mixture; a negative electrode parameter related to the basis weight of the negative electrode composite; a specific surface area of ​​the positive electrode mixture; and a specific surface area of ​​the negative electrode mixture; and Adhesion strength between the positive electrode substrate and the positive electrode composite; and an adhesion strength between the negative electrode substrate and the negative electrode composite material.

9. A battery inspection method comprising: performing the manufacturing process; estimating the battery characteristics of the battery manufactured in the manufacturing process using the estimation device according to claim 1 or 2; determining that the battery is an acceptable product when the battery characteristics satisfy predetermined standards; and determining the battery as a reject if the battery characteristics do not satisfy the predetermined standards.

10. A method for manufacturing a battery, comprising: Executing the inspection method of claim 9; performing a final process on the battery that has been determined to be an acceptable product; and not performing the final process on the battery determined to be a rejected product.

Images