Battery cell sorting method and battery cell handling device

Abstract

To achieve shortening of a read time, appropriate determination of reuse of a battery cell, and appropriate maintenance management of a battery cell.SOLUTION: Regularity information among respective data regarding non-defective or defective battery cells is extracted from a plurality of past data regarding non-defective or defective battery cells in respective steps, and the regularity information among the respective data regarding the non-defective or defective battery cells is input to extract common points of the non-defective or defective battery cells and singular points of the non-defective and defective battery cells. Then, a determination model based on a conditional expression that can discriminate a non-defective battery cell and a defective battery cell from each other on the basis of the singular points is used to collect data necessary for determination using the determination model from data of the battery cells obtained in the respective steps of a manufacturing process, etc., and each management process, and the determination model is used to determine whether a battery cell is non-defective or defective.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to a battery cell sorting method and a battery cell handling device. [Background technology]

[0002] Regarding the characteristics of battery cells for long-term use, a cycle charge-discharge test was conducted in which manufactured battery cells were repeatedly charged and discharged. For example, if the discharge capacity after 3,000 cycles (one cycle is defined as one charge and one discharge of a battery cell) was 80% or more of the capacity at the time of initial manufacture, the battery was deemed to be of good quality.

[0003] Furthermore, in an aging test, a charged battery cell is left for a long period of time, and the presence or absence of an internal short circuit or the like in the battery cell is determined based on changes in voltage.

[0004] However, conducting these tests requires a considerable amount of time. Therefore, rather than conducting these tests on all battery cells, the above tests are conducted on randomly sampled battery cells for each production lot, and if the results indicate that the battery cells are non-defective, other battery cells in the same lot are also deemed to be non-defective.

[0005] However, with the above-mentioned sorting method, the evaluation of the battery cells in the entire lot depends on the judgment results of the sampled battery cells, and therefore, it cannot be said to be sufficient in terms of reliability.

[0006] Furthermore, it is known that a single battery cell that has been determined to be a non-defective product during the manufacturing process, or a battery module (also called a battery pack) that combines multiple such battery cells into a module, is installed in the device as the storage unit of the energy storage device, and is connected to a charge / discharge control unit that controls the charging and discharging of the battery cells, or a CMU (Cell Management Unit: this monitors and manages each battery cell in the storage unit to prevent abnormalities, and includes the case of a "Cell Monitoring Unit" that only has the function of monitoring the voltage and temperature of the battery cell) or a BMU (Battery Management Unit: this monitors and manages the battery cells that make up the battery module in the storage unit to prevent abnormalities), which measures and monitors the battery voltage and temperature of the battery cells, and in addition to controlling the charging and discharging, detects battery abnormalities and estimates the battery life based on the data obtained by measuring the battery cell voltage and temperature.

[0007] In recent years, various predictions have been made using machine learning, particularly by organizing large amounts of information obtained through improvements in artificial intelligence (AI) technology, and analyzing this information using machine learning to find patterns.These patterns are then used in sales strategies, advertising strategies, and so on.Attempts to utilize the patterns found through such AI-based analysis are also being considered for the manufacture and sale of battery cells.

[0008] Examples of technologies using AI include a technology that uses the initial characteristics of a battery to predict its long-term characteristics, thereby quickly determining whether the battery is defective (see, for example, Patent Document 1), and a technology that enables early prediction of battery life from life trends according to various characteristic factors of the battery cell before the battery cell is manufactured and its life is evaluated, thereby improving the reliability of the life prediction (see, for example, Patent Document 2). [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Special Publication No. 2010-539473 [Patent Document 2] Japanese Patent Application Laid-Open No. 2013-217897 Summary of the Invention [Problem to be solved by the invention]

[0010] As mentioned above, advances in AI technology have led to proposals to use machine learning to find patterns and laws among various data obtained through measurements, etc., and to classify the objects being measured based on these patterns. In particular, it has become possible to classify data and discover patterns from it using a wide variety of large amounts of digitized data known as big data. This type of machine learning makes it possible to automatically extract feature points using deep learning, without the need for manual extraction, and it is now possible to extract feature points that would not be noticed by manual extraction.

[0011] However, although the above-described techniques are applicable to predicting the life span of battery cells, there is room for improvement when it comes to applying them as management indicators in the production process, monitoring process, and management process of battery cells.

[0012] The present invention has been made in consideration of the above-mentioned problems, and has an object to provide a battery cell sorting method and battery cell handling device that derives a pattern for distinguishing between good and defective battery cells at each process from data obtained by measurements, etc., extracts specific points between good and defective battery cells based on this pattern, generates a determination model, and performs calculations using the determination model with battery cell data at each process as input data to appropriately determine whether a battery cell is good or defective at each process, thereby shortening lead time, making appropriate decisions about battery cell reuse, and achieving appropriate maintenance and management of battery cells. [Means for solving the problem]

[0013] Form 1: One or more embodiments of the present invention propose a battery cell sorting method including the steps of: creating a judgment model capable of distinguishing between good and bad battery cells using data obtained in each of a plurality of steps in battery cell handling; acquiring various data for each battery cell in each of the steps; using the judgment model to judge whether the battery cell is good or bad, using data obtained in each of the steps necessary for judging the battery cell using the judgment model as input data; and handling battery cells judged as good or bad as a result of the judgment differently depending on the judgment results, wherein one of the steps is a charge / discharge step, and further including the step of determining whether the battery cell requires aging if the battery cell is judged as good in the step of judging whether the battery cell is good or bad using the judgment model.

[0014] Form 2: One or more embodiments of the present invention propose a battery cell sorting method characterized in that, in the step of determining whether the battery cell is good or bad using the determination model, if the battery cell is determined to be bad, a step of determining whether the battery cell requires aging in order to determine whether it should be removed from the production line.

[0015] Mode 3: One or more embodiments of the present invention propose a battery cell sorting method, wherein the aging is a judgment process for judging whether a battery cell is good or bad based on the amount of self-discharge.

[0016] Mode 4: One or more embodiments of the present invention propose a battery cell sorting method, characterized in that the judgment model extracts regularity information between data on good or defective battery cells from a plurality of past data on good or defective battery cells, extracts singular points between the good and defective battery cells from common points between the good and defective battery cells based on the extracted regularity information, and creates a conditional formula that can distinguish between good and defective battery cells based on the singular points using data obtained in each process.

[0017] Mode 5: One or more embodiments of the present invention propose a battery cell sorting method, wherein the data obtained in each of the steps is data obtained during the manufacturing process of the battery cells.

[0018] Mode 6: One or more embodiments of the present invention include a judgment model generation unit that creates a judgment model that can distinguish between good and bad battery cells using data obtained in each step of a battery cell handling process that includes multiple steps; a data acquisition unit that acquires various data for each battery cell in each of the steps; and a pass / fail judgment unit that uses the judgment model to judge whether the battery cell is good or bad, using data necessary for judging whether the battery cell is good or bad, from the data acquired in each of the steps, as input data. The proposed battery cell handling device includes a battery cell handling unit that handles battery cells that are determined to be good and battery cells that are determined to be defective differently depending on the results of the judgment, one of the processes being a charge / discharge process, and is equipped with an aging judgment unit that, when the battery cell is determined to be good by the good product judgment unit, determines whether aging is necessary for the battery cell.

[0019] Mode 7: One or more embodiments of the present invention propose a battery cell handling device characterized in that, when the battery cell is determined to be defective, the aging determination unit determines whether aging is required for the battery cell in order to determine whether to remove the battery cell from the production line.

[0020] Mode 8: One or more embodiments of the present invention propose a battery cell handling device characterized in that the aging is a judgment process that judges whether a battery cell is good or bad based on the amount of self-discharge.

[0021] Mode 9: One or more embodiments of the present invention propose a battery cell handling device comprising: a regularity information extraction unit that extracts regularity information between each piece of data related to good and defective battery cells from a plurality of past data related to good and defective battery cells in each process; a memory unit that stores the regularity information between each piece of data related to good and defective battery cells extracted by the regularity information extraction unit; a singularity extraction unit that inputs the regularity information between each piece of data related to good and defective battery cells from the memory unit and extracts common points between the good and defective battery cells and singular points between the good and defective battery cells; a conditional expression calculation unit that calculates a conditional expression capable of distinguishing between good and defective battery cells based on the singular points; and a judgment model generation unit that verifies the validity of the conditional expression calculated by the conditional expression calculation unit and generates a judgment model.

[0022] Aspect 10: One or more embodiments of the present invention provide a battery cell handling device, wherein the data is data obtained during the manufacturing process of the battery cells. [Effects of the Invention]

[0023] According to one or more embodiments of the present invention, a regularity between good and bad battery cells in each process is derived from data obtained by measurement or the like, and specific points between good and bad battery cells are extracted based on this regularity to generate a judgment model. Data on the battery cells in each process is used as input data and calculation processing is performed using the judgment model, thereby appropriately judging whether the battery cells are good or bad in each process, thereby achieving the effects of shortening lead times, appropriately determining whether the battery cells should be reused, and appropriately maintaining and managing the battery cells. [Brief explanation of the drawings]

[0024] [Figure 1] 1 is an electrical configuration diagram of a battery cell handling device according to a first embodiment of the present invention. [Figure 2] 2 is a functional block diagram of the inside of a CPU in the battery cell handling device according to the first embodiment of the present invention. FIG. [Figure 3] FIG. 3 is a processing flow diagram of a CPU in the battery cell handling device according to the first embodiment of the present invention. [Figure 4] FIG. 2 is a process flow diagram of the battery cell handling device according to the first embodiment of the present invention. [Figure 5] FIG. 10 is an electrical configuration diagram of a battery cell handling device according to a second embodiment of the present invention. [Figure 6] FIG. 10 is a functional block diagram of the inside of a CPU in a battery cell handling device according to a second embodiment of the present invention. [Figure 7] FIG. 10 is a process flow diagram of a battery cell handling device according to a second embodiment of the present invention. [Figure 8] FIG. 10 is a processing flow diagram relating to a modified example of the processing of the battery cell handling device according to the second embodiment of the present invention. [Figure 9] FIG. 10 is an electrical configuration diagram of a battery cell handling device according to a third embodiment of the present invention. [Figure 10] FIG. 10 is a functional block diagram of the inside of a CPU in a battery cell handling device according to a third embodiment of the present invention. [Figure 11] FIG. 10 is a process flow diagram of a battery cell handling device according to a third embodiment of the present invention. [Figure 12] FIG. 10 is an electrical configuration diagram of a battery cell handling device according to a fourth embodiment of the present invention. [Figure 13] FIG. 10 is a functional block diagram of the inside of a CPU in a battery cell handling device according to a fourth embodiment of the present invention. [Figure 14] FIG. 10 is a process flow diagram of a battery cell handling device according to a fourth embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0025] First Embodiment Hereinafter, an embodiment of the present invention will be described with reference to FIGS.

[0026] <Electrical configuration of the battery cell process control device> As shown in FIG. 1, the battery cell handling device 10 according to this embodiment is configured to include a CPU (Central Processing Unit) 100, a ROM (Read Only Memory) 200, a RAM (Random Access Memory) 300, a judgment model storage unit 400, a non-defective product judgment unit 500, and an interface unit 600.

[0027] The CPU 100 controls the overall processing of the battery cell handling device 10 in accordance with a control program pre-stored in the ROM 200 . In particular, in this embodiment, based on various data that has been obtained up to now at each step in the manufacturing and management processes, the regularity of good and bad battery cells in the manufacturing and management processes is derived from data obtained by measurements, etc., and the singular points of good and bad products are extracted based on this regularity, and a judgment model is created.

[0028] The ROM 200 is a read-only nonvolatile memory, and stores the control program, initial parameters, etc., as described above.

[0029] The RAM 300 is a writable and readable memory, and examples thereof include a DRAM (Dynamic RAM) and an SRAM (Static RAM). The RAM 300 temporarily stores data obtained during the control processing of the CPU 100.

[0030] The judgment model storage unit 400 is a storage unit that stores the judgment model generated by the CPU 100. The judgment model stored in the judgment model storage unit 400 may be stored in the RAM 300 described above, but by providing an independent judgment model storage unit 400 for storing the judgment model as in this embodiment, it is possible to avoid delays in the entire calculation process due to delays in reading the judgment model when performing calculation processing using the judgment model, even if the RAM 300 is being used for another process, for example.

[0031] The non-defective product determination unit 500 uses the data of the battery cell in each process as input data and performs calculations using the determination model stored in the determination model storage unit 400 to determine whether the battery cell is non-defective or defective. Here, the battery cell data refers to data obtained during the manufacturing process, that is, data measured at each step in the battery cell manufacturing process and data obtained for production management, such as the battery capacity, energy density, shape, size, electrode plate material, electrode plate thickness, electrode plate length, electrode active material layer material and mixing ratio of materials used, presence or absence of binder and binder type, electrode active material layer thickness, size of uncoated part of the electrode active material layer, separator type, stacked structure of electrode plate and separator (number of windings in the case of a wound type, number of windings in the case of a stack type), etc. Examples include the number of layers, if any), type of electrolyte, composition ratio of the electrolyte, presence or absence of additives such as flame retardants and the type and concentration of the additives, drying time, drying temperature, current collector material, current collector structure, joining method and joining structure of the current collector and electrode plate, electrolyte injection time, electrolyte injection method, time required for impregnation and temperature during impregnation, formation charging time, formation charging voltage value, formation charging temperature, battery voltage during charge / discharge test, battery temperature, discharge time, charge time, charge voltage, battery capacity, production lot, weather (humidity) at the time of production, time at the time of production, name of person in charge of production, and name of manufacturer of each material. Furthermore, information and measurement data relating to the battery cells are linked to a common serial number (production number) and stored and managed for each battery cell to be manufactured, for example, in a data server on a network. Furthermore, information and data required for determination using the determination model are written to the RAM 300 from the data server, and determination processing using the determination model is performed based on this written information (data). It is also possible to configure the CPU 100 to perform the processing of the non-defective product determination unit 500 without providing a separate non-defective product determination unit 500. However, providing a separate non-defective product determination unit 500 has the advantage of being able to execute the non-defective product determination processing by the non-defective product determination unit 500 in parallel with the processing by the CPU 100.

[0032] The interface unit 600 exchanges data with various devices connected to the network via the network. For example, in each process, when the data storage unit 1000 that stores various data obtained so far is a server or the like on a network, the battery cell handling device 10 according to the present embodiment exchanges data stored in the data storage unit 1000 via the interface unit 600.

[0033] <Functional blocks inside the CPU> As shown in FIG. 2, the CPU 100 of the battery cell handling device 10 according to the present embodiment includes a regular information extraction unit 101, a regular information storage unit 102, an outlier extraction unit 103, a conditional expression calculation unit 104, and a determination model generation unit 105.

[0034] The regular information extraction unit 101 extracts regular information regarding good or defective battery cells from a plurality of past data regarding good or defective battery cells in each process from the data storage unit 1000 provided outside, for example. Specifically, for example, using machine learning, the regularity between each data regarding good or defective battery cells may be executed by analyzing the relationship of the data. As a specific example, a method of graphing the relationship between the battery voltage, battery temperature, battery capacity and time during charging or discharging of each battery cell can be mentioned. Furthermore, after classifying the battery cells by production lot or by the type of electrolyte applied, a reclassification process is performed based on information that allows further classification. In addition, it is also possible by a computer to input necessary data and graph various relationships based on the input data, and it is also possible by a computer to further classify the graph by information such as the production lot of the battery cell or the type of material used, if a link between the information regarding the battery cell and the target graph is established. In particular, in the present embodiment, it is important to link the obtained graph with information indicating whether the battery cell is good or defective. In addition, since unexpected common points that are not predicted may be found, graphs or distribution diagrams regarding the relationship may also be created even between information that seems to have little relevance.

[0035] The regularity information storage unit 102 stores regularity information between each piece of data relating to non-defective or defective battery cells extracted by the regularity information extraction unit 101 . In this embodiment, the regularity information storage unit 102 is a component of the CPU 100, but the regularity information storage unit 102 may be provided on a network as, for example, a data server.

[0036] The singularity extraction unit 103 inputs regularity information between each piece of data relating to good and defective battery cells from the regularity information storage unit 102, and extracts commonalities between the good and defective battery cells and singularities between the good and defective battery cells. Specifically, for example, by comparing several graphs of information, the system extracts commonalities between good battery cells, commonalities between defective battery cells, and differences between good and defective battery cells, and finds elements that can serve as singular points that distinguish good and defective battery cells, enabling the determination of good from defective products. In other words, even if there are differences between good and defective products, if those differences are not common to good battery cells, those differences cannot be used to determine whether the product is good or defective. Furthermore, if commonalities between good battery cells are also common to defective battery cells, it is not possible to distinguish between good and defective products. For this reason, singular points that enable the determination of good and defective products are extracted from the commonalities between good battery cells, commonalities between defective battery cells, and differences between good and defective battery cells.

[0037] The conditional expression calculation unit 104 calculates a conditional expression that can distinguish between a good battery cell and a defective battery cell based on the singular points extracted by the singular point extraction unit 103. Specifically, data required to calculate the extracted singular points is identified, an equation capable of calculating the singular points is calculated using the data, and a numerical value related to the singular points obtained from the calculated equation is prepared in the form of a conditional equation for determining whether the product is good or bad. Here, the number of conditional equations prepared is not limited to one, and if multiple singular points are found, an equation for calculating the singular points is calculated using data required for each singular point, and conditional equations based on the calculated equations are prepared for each singular point.

[0038] The decision model generation unit 105 verifies the validity of the conditional expressions calculated by the conditional expression calculation unit 104 and generates a decision model. Specifically, for multiple battery cells that have already been determined as pass or fail based on the conditional formula calculated by the conditional formula calculation unit 104, information required to determine whether the battery cells are pass or fail is applied to the conditional formula to verify whether it is possible to correctly determine whether the battery cells are pass or fail. If multiple conditional formulas are calculated, verification is performed for each conditional formula. If the verification results in a correct determination of whether a battery cell is good or defective (this can be the case when all battery cells used for verifying the judgment model can be correctly determined to be good or defective, or when 95% can be correctly determined to be good or defective), the model that uses the conditional expression to determine whether a product is good or defective is determined to be the judgment model for the process. If there are multiple verified conditional expressions, the conditional expression with the highest probability of determining a product as non-defective is selected as a result of the verification, and the model for determining whether a product is non-defective or defective is determined as the judgment model for the process. It is also possible to apply a plurality of judgment models to the process. For example, if the verification results show that a product can be judged as a good product with a similarly high probability using a plurality of conditional expressions, or if the probability of a product being judged as a good product in the verification results exceeds an arbitrarily set pass line probability (for example, 85% or more being judged as good products), a model for judging a product as good or defective using each of the plurality of conditional expressions may be used as the judgment model. A plurality of determination models, each requiring different information for determination, will be prepared. Compared with the case of one determination model, although it may become redundant because the determination of non-defective or defective products will be performed many times, for example, when a determination model that can correctly determine non-defective and defective products for 95% is adopted, it can be expected that the remaining 5% that cannot be correctly determined by this determination model can be correctly determined as non-defective or defective products by another determination model. Also, when making determinations using multiple determination models, when the determination results of X (an integer of 2 or more smaller than N) out of the determinations of N (an integer of 3 or more) determination models match, the handling can be advanced as non-defective or defective according to the result, so that the handling of non-defective or defective products can be performed with higher accuracy. In addition, prioritize in descending order of the probability of correctly determining non-defective and defective products, apply the result of the determination model with the highest priority first, and within the error range of the boundary point between non-defective and defective products in this determination model, adopt the result of the determination model with the next priority, etc., and it may be possible to make determinations in multiple stages. Then, output the determined determination model to the non-defective determination unit 500 of the battery cell handling device 10. Note that, based on the singular points extracted by a personal computer or the like as described above, it is not limited to modeling the determination model. For example, by deep learning, singular points can be automatically extracted from the input information (data), a conditional expression based on the singular points can be calculated, and the determination model can be created by verifying the conditional expression.

[0039] <CPU Processing> Using FIG. 3, the generation process of the determination model in the CPU 100 according to the present embodiment will be described.

[0040] The regular information extraction unit 101 inputs, from the data storage unit 1000, the past data related to the defective battery cells acquired so far in the target process (step S110).

[0041] Next, the regularity information extraction unit 101 extracts regularity information between data relating to non-defective or defective battery cells from the multiple past data relating to defective battery cells input from the data storage unit 1000 (step S120).

[0042] The singularity extraction unit 103 inputs regularity information between each piece of data relating to good and defective battery cells from the regularity information storage unit 102, and extracts commonalities between good and defective battery cells and singularities between good and defective battery cells, for example, by using graphs of the input data for each classification (step S130).

[0043] The conditional expression calculation unit 104 calculates a conditional expression that can distinguish between a good battery cell and a defective battery cell based on the singular points extracted by the singular point extraction unit 103 (step S140).

[0044] The judgment model generating unit 105 generates a judgment model based on the conditional expression calculated by the conditional expression calculating unit 104 (step S150).

[0045] Additionally, the judgment model generation unit 105 applies information necessary for judging whether a battery cell is good or bad to the conditional equation calculated by the conditional equation calculation unit 104, and verifies whether it is possible to correctly judge whether the battery cell is good or bad (step S160).

[0046] If, as a result of the investigation, the judgment model generation unit 105 determines that the conditional expression cannot correctly determine whether the battery cell is a good product or a defective product ("NO" in step S160), the judgment model generation unit 105 returns the processing to step S120.

[0047] On the other hand, when the judgment model generation unit 105 determines, as a result of its consideration, that the conditional expression can correctly determine whether a battery cell is good or bad ("YES" in step S160), it determines the judgment model as the judgment model for the process (step S170).

[0048] <Disposal of battery cell handling equipment> The processing performed by the battery cell handling device 10 according to this embodiment will be described with reference to FIG.

[0049] The CPU 100 inputs data obtained by performing processing in a target step from the data storage unit 1000, which is, for example, a server on a network, via the interface unit 600 (step S210).

[0050] The non-defective product determination unit 500 performs calculation processing using the battery cell data in each process as input data and the determination model stored in the determination model storage unit 400 (step S220).

[0051] Based on the results of the calculation process, the non-defective product determination unit 500 determines whether the battery cell is a non-defective product or a defective product (step S230).

[0052] The determination result is output via the CPU 100, for example, from the interface unit 600, and is used to control a functional unit that performs a sorting process (post-processing step) to separate non-defective and defective battery cells (step S240).

[0053] As explained above, the battery cell handling device 10 according to this embodiment uses information (data) acquired at each process for battery cells, and from this information, extracts characteristic points found in good battery cells, characteristic points found in defective battery cells, and peculiar points found between good and defective battery cells. It then performs arithmetic processing on the extracted characteristic points and peculiar points and various information (data) acquired at the actual processes to generate a judgment model for determining whether a battery cell is good or defective, thereby making it possible to appropriately judge whether a battery cell is good or defective at each process for battery cells.

[0054] Furthermore, by deriving the regularity between good and bad battery cells in each process from data obtained by measurements, etc., and extracting the singular points between good and bad battery cells, and appropriately determining whether a battery cell is good or bad in each process, it is possible to, for example, shorten the lead time in the manufacturing process, make appropriate decisions about reusing battery cells in the reuse process, and achieve appropriate maintenance management of battery cells in the maintenance management process.

[0055] <Second embodiment> Hereinafter, an embodiment of the present invention will be described with reference to FIGS. This embodiment illustrates the charge / discharge process among the battery cell production processes.

[0056] <Battery cell handling device electrical configuration> As shown in FIG. 5, the battery cell handling device 20 according to this embodiment is configured to include a CPU (Central Processing Unit) 110, a ROM (Read Only Memory) 200, a RAM (Random Access Memory) 300, a judgment model storage unit 410, a non-defective product judgment unit 510, an interface unit 600, and an aging judgment unit 700. Note that components with the same reference numerals as those in the first embodiment have the same functions, and therefore detailed descriptions thereof will be omitted.

[0057] The CPU 110 controls the overall processing of the battery cell handling device 10 in accordance with a control program for the charge and discharge process that is pre-stored in the ROM 200 . In particular, in this embodiment, based on various data obtained up to now during the charging and discharging process, the regularity of good and bad battery cells during the charging and discharging process is derived from data obtained by measurements, etc., and the singular points of good and bad products are extracted based on this regularity, and a judgment model is generated.

[0058] The determination model storage unit 410 is a storage unit that stores the determination model generated by the CPU 110.

[0059] The non-defective product determination unit 510 uses the data of the battery cell in the charge / discharge process as input data and performs calculation processing using the determination model stored in the determination model storage unit 410 to determine whether the battery cell is non-defective or defective. Here, examples of the battery cell data include the battery voltage, battery temperature, discharge time, charge time, charge voltage, etc. in the charge and discharge process.

[0060] If the battery cell is determined to be non-defective, the aging determination unit 700 determines whether or not the battery cell has aged. It should be noted that aging here refers to a process of determining whether a battery cell is truly good or defective based on whether an abnormality due to an internal short circuit or the like has occurred, i.e., whether the battery cell has a larger amount of self-discharge than a good battery cell. This determination process is performed by charging the battery cell to a certain extent and leaving it for a long period of time, such as one week or one month.

[0061] As shown in FIG. 6, the CPU 110 includes a regularity information extraction unit 111, a regularity information storage unit 112, a singularity extraction unit 113, a conditional expression calculation unit 114, and a judgment model generation unit 115.

[0062] The regularity information extraction unit 111 extracts regularity information between each piece of data relating to good or defective battery cells from a plurality of past data relating to good or defective battery cells in a charge / discharge process, for example, from an externally provided data storage unit 1000.

[0063] The regularity information storage unit 112 stores regularity information between each piece of data relating to non-defective or defective battery cells extracted by the regularity information extraction unit 111 .

[0064] The singularity extraction unit 113 inputs regularity information between each piece of data relating to good and defective battery cells from the regularity information storage unit 112, and extracts commonalities between the good and defective battery cells and singularities between the good and defective battery cells.

[0065] The conditional expression calculation unit 114 calculates a conditional expression that can distinguish between a good battery cell and a defective battery cell based on the singular points extracted by the singular point extraction unit 113.

[0066] The decision model generation unit 115 verifies the validity of the conditional expressions calculated by the conditional expression calculation unit 114 and generates a decision model.

[0067] <Disposal of battery cell handling equipment> The processing performed by the battery cell handling device 10 according to this embodiment will be described with reference to FIG.

[0068] The CPU 110 inputs data obtained by performing processing during the charge / discharge process from the data storage unit 1000, which is, for example, a server on a network, via the interface unit 600 (step S310). In this embodiment, there is one judgment model, and the data required for judgment using the judgment model will be described as data on the battery cell during the charge / discharge process.

[0069] The non-defective product determination unit 510 performs calculation processing using the data of the battery cell in the charge / discharge process as input data and the determination model stored in the determination model storage unit 410 (step S320). This allows information on the charge voltage, discharge voltage and capacity obtained in the charge and discharge process to be obtained, and charge and discharge characteristics to be obtained from the relationship between the charge voltage, discharge voltage and capacity.

[0070] The non-defective product determination unit 510 determines whether or not the battery cell can be determined to be non-defective based on the result of the calculation using the determination model (step S330). Here, the determination of whether the battery cell is a non-defective product or a defective product is made based on whether the result calculated using the determination model is within a range for determining that the battery cell is a non-defective product.

[0071] If the battery cell is determined to be non-defective ("NO" in step S330), the battery cell is removed from the original production line without proceeding to the next processing step and is disposed of as a defective product (such as being transported to a waste storage facility) (step S340).

[0072] On the other hand, if the non-defective product determination unit 510 determines that the battery cell is non-defective ("YES" in step S330), the non-defective product determination unit 510 then determines whether or not aging processing is necessary (step S350). Here, the determination of whether or not aging processing is necessary is made, for example, based on whether or not the result of the determination by the non-defective product determination unit 510 in step S340 is that the battery cell is determined to be non-defective, but is near the limit point for distinguishing between non-defective and defective battery cells. Specifically, the determination is made based on whether the battery cell is within a predetermined error range with respect to the boundary point between a good and a bad battery cell.

[0073] Therefore, if the charge / discharge characteristics of the battery cell are judged to be good within a predetermined error range of the boundary point between good and defective battery cells, for example, within a range of + (or -) 15% of the boundary point ("NO" in step S350), the good product judgment unit 510 determines that normal aging processing is necessary and makes a judgment to perform normal aging processing (step S370).

[0074] On the other hand, if the charge / discharge characteristics of the battery cell are judged to be good because they are outside a predetermined error range of the boundary point between good and defective products, for example, a range of + (or -) 15% of the boundary point ("YES" in step S350), the good product judgment unit 510 determines that normal aging processing is unnecessary and makes a decision not to perform the aging processing (step S360).

[0075] As described above, according to this embodiment, the pass / fail judgment unit 510 does not judge whether to perform normal aging processing on all pass / fail battery cells, but judges that normal aging processing is necessary if the charge / discharge characteristics of the battery cell are judged to be pass / fail within a predetermined error range of the boundary point between pass / fail battery cells, for example, within a range of + (or -) 15% of the boundary point, and judges that normal aging processing is unnecessary if the charge / discharge characteristics of the battery cell are judged to be pass / fail outside a predetermined error range of the boundary point between pass / fail battery cells, for example, within a range of + (or -) 15% of the boundary point. Therefore, by applying the judgment model to the charging and discharging process of the battery cells, it is possible to improve the production efficiency of the battery cells.

[0076] <Modification> In this embodiment, if the battery cell is determined to be non-defective ("NO" in step S330) in the step of determining whether the battery cell can be determined to be non-defective based on the results of calculations using the determination model in Fig. 7, the battery cell is discarded (step S340). However, even if the battery cell is determined to be non-defective, it may be determined whether aging processing is necessary. For example, if a battery cell is determined to be defective within the error range of the boundary point between good and defective battery cells as described above, it may be determined that normal aging processing is necessary, and if the charge / discharge characteristics of the battery cell are outside the predetermined error range of the boundary point between good and defective battery cells and the battery cell is determined to be defective, it may be discarded.

[0077] Specifically, as shown in FIG. 8, the non-defective product determination unit 510 determines whether or not the battery cell can be determined to be non-defective based on the result of calculation using the determination model (step S330). If the non-defective product determination unit 510 determines that the battery cell is not a non-defective product ("NO" in step S330), it determines whether or not aging processing is necessary (step S380). Here, if the charge / discharge characteristics of the battery cell are judged to be defective within a predetermined error range of the boundary point between good and bad battery cells, for example, within a range of + (or -) 15% of the boundary point ("NO" in step S380), the good product judgment unit 510 determines that normal aging processing is necessary and makes a decision to perform normal aging processing (step S390).

[0078] On the other hand, if the non-defective product determination unit 510 determines that the charge / discharge characteristics of the battery cell are defective because they are outside a predetermined error range of the boundary point between a good product and a defective product, for example, a range of + (or -) 15% of the boundary point ("YES" in step S350), the battery cell is removed from the original production line without proceeding to the next processing step, and is discarded as a defective product (such as being transported to a waste storage facility) (step S340).

[0079] In this way, by performing normal aging processing on battery cells that are judged to be good or bad within the error range of the boundary point between good and bad battery cells, the aging processing can be omitted for battery cells that are judged not to require aging processing, and the aging processing can be performed on other battery cells. Therefore, it is possible to more reliably separate non-defective and defective battery cells while reducing the number of battery cells to be aged. Furthermore, by feeding back the results of this aging process into the creation of a judgment model and creating a judgment model that can reliably determine whether a battery cell is good or bad (or updating the created judgment model), the number of battery cells that fall within the range of the boundary point between good and bad battery cells can be reduced, and a judgment model can be constructed that can more accurately determine whether a battery cell is good or bad.

[0080] Furthermore, in addition to feeding back the results of the aging process to the creation of the judgment model to update the judgment model, if the results of the aging process are used to extract regularity information about good or defective battery cells or extract singularities about good or defective battery cells using machine learning in the creation of the judgment model (S120, S130 in FIG. 3), and the regularity information or singularity information is found to be due to the raw material lot, rather than feeding it back to the creation of the judgment model, the regularity information or singularity information may be provided (feedback) to an earlier stage of the production process, such as reviewing the raw material delivery specifications. In this way, in addition to feeding back new judgment results in the battery cell handling process to which the judgment model is applied to updating (creating) the judgment model, the regularity information or singularity information may be output depending on the nature of the extracted regularity or singularity, and used at a more appropriate stage in the battery cell handling process, which is expected to improve battery performance and quality.

[0081] In the present embodiment, the data required for determination using the determination model is data on the battery cell in the charge / discharge process, and an example has been described in which the aging process is selectively performed. However, if the data required for judgment using the judgment model is other than data obtained during the charge / discharge process, for example, data obtained during a process prior to the charge / discharge process, the judgment as to whether the product is good or bad may be made at a stage prior to the charge / discharge process, that is, at the stage when all the data that can be used for judgment using the judgment model has been collected. It is preferable to link battery cells that are determined to be defective with a serial number or other information specified by a barcode attached to the battery cell, so that the battery cell can be recognized as a defective product that does not require further processing. By doing this, battery cells that are recognized as defective when their barcodes are read in subsequent processes can be managed so that they can be removed from the production line at the appropriate timing.

[0082] Further, only the battery cells that are determined to be non-defective (only the battery cells that are not determined to be defective) may be subjected to subsequent charge / discharge processes, aging processes, and the like. In this way, a CPU or the like manages whether or not the data required for the judgment model is available, and when the data required for judgment using the judgment model is available, the judgment model judges whether the battery cell is a good or defective product. By linking the judgment result of the judgment model's judgment of whether the battery cell is a good or defective product with information that can identify the battery cell being sent down the production line, the following advantages are obtained. The manufacturing process for battery cells consists of multiple steps, and one or more of these multiple steps includes a step of performing a pass / fail judgment on battery cells without using the judgment model of this embodiment and eliminating defective products. If the results of the pass / fail judgment on battery cells using the judgment model described above, which is performed in parallel with the pass / fail judgment on battery cells without using the judgment model, are linked to information that can identify the battery cell, then when making a pass / fail judgment on battery cells without using the judgment model, the information that can identify the battery cell being judged can be read and confirmed against the judgment result of the pass / fail judgment on battery cells using the judgment model described above. This allows the information that the pass / fail judgment on battery cells using the judgment model described above indicates that the battery cell is defective to be recognized, and the defective battery cell can be eliminated without having to make a pass / fail judgment on battery cells without using the judgment model.

[0083] <Third embodiment> Hereinafter, an embodiment of the present invention will be described with reference to FIGS. In this embodiment, a reuse (reuse determination) process is exemplified as a process.

[0084] <Battery cell handling device electrical configuration> As shown in FIG. 8, the battery cell handling device 30 according to this embodiment includes a CPU (Central Processing Unit) 120, a ROM (Read Only Memory) 200, a RAM (Random Access Memory) 300, a judgment model storage unit 420, a non-defective product judgment unit 520, an interface unit 600, and a reuse judgment unit 800. Note that components with the same reference numerals as those in the first and second embodiments have similar functions, and therefore detailed descriptions thereof will be omitted.

[0085] The CPU 120 controls the overall processing of the battery cell handling device 10 in accordance with a reuse process control program that is pre-stored in the ROM 200 . In particular, in this embodiment, based on various data obtained in the production process or the reuse process, which is made up of processes including charge / discharge processing for charging / discharging recovered battery cells and measuring the capacity when fully charged, a regularity between good and bad battery cells in the reuse process is derived from data obtained by measurements, etc., and specific points between good and bad products are extracted based on this regularity, and a process is performed to generate a judgment model.

[0086] The determination model storage unit 420 is a storage unit that stores the determination model generated by the CPU 120.

[0087] The pass / fail judgment unit 520 uses as input data battery cell data obtained in the production process or the reuse process, which is made up of processes including charging and discharging processes for charging and discharging recovered battery cells and measuring the capacity when fully charged, and performs calculations using the judgment model stored in the judgment model storage unit 420 to judge whether the battery cell is pass or defective. Here, examples of battery cell data include battery temperature, battery voltage, battery capacity, number of charge / discharge cycles, number of years of use, whether or not the battery has been unused for a long period of time, number of errors, time when the error occurred, and information regarding the error content.

[0088] When the battery cell is determined to be non-defective, the reuse determination unit 800 determines whether or not the full charge characteristics and temperature characteristics are both within ranges that allow the battery cell to be reused. Here, reuse means that the collected battery cells and energy storage systems are used again in their original form. It is required that the battery cells and energy storage systems can be charged to the desired capacity even if the charge capacity at full charge has decreased due to deterioration compared to when they were delivered, and that the temperature of the battery cells does not rise due to charging and discharging. In the following, we will particularly explain the case where a judgment model is used that can classify battery cells into good and bad products based on the relationship between the charge / discharge voltage and capacity of the battery cell (charge / discharge characteristics), the capacity value at full charge (full charge characteristics), and the relationship between the charge / discharge voltage and temperature change of the battery cell (temperature characteristics).

[0089] As shown in FIG. 9, the CPU 120 includes a regularity information extraction unit 121, a regularity information storage unit 122, a singularity extraction unit 123, a conditional expression calculation unit 124, and a judgment model generation unit 125.

[0090] The regularity information extraction unit 121 extracts regularity information between each piece of data relating to good or defective battery cells from a plurality of past data relating to good or defective battery cells obtained, for example, from an externally provided data storage unit 1000 in a production process or a reuse process consisting of processes including a charge / discharge process for charging / discharging recovered battery cells and measuring the capacity when fully charged.

[0091] The regularity information storage unit 122 stores regularity information between each piece of data relating to non-defective or defective battery cells extracted by the regularity information extraction unit 121 .

[0092] The singularity extraction unit 123 inputs regularity information between each piece of data relating to good and defective battery cells from the regularity information storage unit 122, and extracts commonalities between the good and defective battery cells and singularities between the good and defective battery cells.

[0093] The conditional expression calculation unit 124 calculates a conditional expression that can distinguish between a good battery cell and a defective battery cell based on the singular points extracted by the singular point extraction unit 123.

[0094] The judgment model generation unit 125 verifies the validity of the conditional expressions calculated by the conditional expression calculation unit 124 and generates a judgment model.

[0095] <Disposal of battery cell handling equipment> The processing performed by the battery cell handling device 30 according to this embodiment will be described with reference to FIG.

[0096] The CPU 120 inputs data obtained in the production process or the reuse process, which includes processes such as charging and discharging processes for charging and discharging recovered battery cells and measuring the capacity when fully charged, from the data storage unit 1000, which is, for example, a data server, via the interface unit 600 (step S410).

[0097] The non-defective product determination unit 520 performs calculation processing using the determination model stored in the determination model storage unit 420, using as input data the data of the battery cells obtained in the production process or the reuse process, which includes a charge / discharge process for charging / discharging recovered battery cells and a process including measuring the capacity when fully charged (step S420). This allows obtaining the relationship between the charge / discharge voltage and capacity of the battery cell (charge / discharge characteristics), the capacity value at full charge (full charge characteristics), and the relationship between the charge / discharge voltage and the temperature change of the battery cell (temperature characteristics).

[0098] The non-defective product determination unit 520 determines whether or not the battery cell can be determined to be non-defective based on the result of the calculation using the determination model (step S430). Here, the determination of whether the battery cell is a good or bad product is made, for example, by checking whether the result calculated using a determination model based on the charge / discharge characteristics is within a range for determining that the battery cell is a good product.

[0099] If the battery cell is determined to be non-defective ("NO" in step S430), the battery cell is removed from the original production line without proceeding to the next processing step and is disposed of as a defective product (such as being transported to a waste storage facility) (step S440).

[0100] On the other hand, if the non-defective product determination unit 520 determines that the battery cell is non-defective ("YES" in step S430), the reuse determination unit 800 then determines whether or not the battery cell is reusable (step S450). Here, whether or not the battery can be reused is determined based on, for example, full charge characteristics and temperature characteristics.

[0101] If the reuse determination unit 800 determines that at least one of the full charge characteristics or temperature characteristics of the battery cell is outside the range for determining that the battery cell is reusable ("NO" in step S450), the battery cell is disposed of as a defective product that cannot be reused (step S440).

[0102] On the other hand, if the reuse determination unit 800 determines that both the full charge characteristics and the temperature characteristics are within the ranges that determine that the battery can be reused ("YES" in step S450), the battery is reused as a battery cell and its state is managed (step S460).

[0103] As described above, according to this embodiment, the reuse determination unit 800 determines whether or not a battery cell determined to be a good product by the good product determination unit 520 is reusable, based on whether or not at least one of the full charge characteristics or temperature characteristics of the battery cell is within a range that determines that the battery cell is reusable. Therefore, by applying the judgment model to the battery cell reuse process, the battery cells can be reused efficiently and appropriately.

[0104] <Fourth embodiment> Hereinafter, an embodiment of the present invention will be described with reference to FIGS. In this embodiment, as a process, for example, a monitoring process for monitoring the behavior of a battery cell that is installed in a power storage system and is in operation is illustrated.

[0105] <Battery cell handling device electrical configuration> As shown in FIG. 11 , the battery cell handling device 40 according to this embodiment is configured to include a CPU (Central Processing Unit) 130, a ROM (Read Only Memory) 200, a RAM (Random Access Memory) 300, a judgment model storage unit 430, a non-defective product judgment unit 530, an interface unit 600, and a life prediction unit 900. Note that components with the same reference numerals as those in the first, second and third embodiments have the same functions, and therefore detailed descriptions thereof will be omitted.

[0106] The CPU 130 controls the overall processing of the battery cell handling device 10 in accordance with a control program for the monitoring process that is pre-stored in the ROM 200 . In particular, in this embodiment, based on various data obtained in the production process or monitoring process, the regularity of good and bad battery cells in the monitoring process is derived from data obtained by measurements, etc., and based on this regularity, singular points between good and bad products are extracted, and a judgment model is generated.

[0107] The determination model storage unit 430 is a storage element that stores the determination model generated by the CPU 130.

[0108] The pass / fail judgment unit 530 uses battery cell data obtained in the production process or monitoring process as input data and performs calculations using the judgment model stored in the judgment model storage unit 430 to judge whether the battery cell is pass or fail. Here, examples of battery cell data include battery temperature, battery voltage, battery capacity, number of charge / discharge cycles, number of years of use, whether or not the battery has been unused for a long period of time, number of errors, time when the error occurred, and information regarding the error content.

[0109] When a battery cell is determined to be non-defective, the life prediction unit 900 predicts the life of the battery cell. Here, the life prediction is made when the battery voltage and battery temperature of the battery cell measured in the monitoring process are in a range of voltage values ​​and temperatures that cannot be measured in the case of a normal battery cell, and for battery cells that are determined to be unable to be used normally, a warning is issued to notify the user that continued use is prohibited. This can be determined in the actual monitoring process by checking whether the battery voltage and battery temperature of the battery cell being monitored are within normal ranges. In this embodiment, even if the battery voltage and battery temperature of the monitored battery cell are judged to be normal, the state showing signs that the battery cell is nearing the end of its life (advancing deterioration of the battery cell) is determined from the measured data as a judgment model and judged by calculation. In the following, an example of a determination model that can determine deterioration using information related to the battery voltage and battery temperature of the battery cell measured in the battery cell monitoring process will be described.

[0110] As shown in FIG. 12, the CPU 130 includes a regularity information extraction unit 131, a regularity information storage unit 132, a singularity extraction unit 133, a conditional expression calculation unit 134, and a judgment model generation unit 135.

[0111] The regularity information extraction unit 131 extracts regularity information between each piece of data relating to good or defective battery cells from a plurality of past data relating to good or defective battery cells obtained in a production process or a monitoring process from an externally provided data storage unit 1000, for example.

[0112] The regularity information storage unit 132 stores regularity information between each piece of data relating to non-defective or defective battery cells extracted by the regularity information extraction unit 131 .

[0113] The singularity extraction unit 133 inputs regularity information between each piece of data relating to good and defective battery cells from the regularity information storage unit 132, and extracts commonalities between the good and defective battery cells and singularities between the good and defective battery cells.

[0114] The conditional expression calculation unit 134 calculates a conditional expression that can distinguish between a good battery cell and a defective battery cell based on the singular points extracted by the singular point extraction unit 133.

[0115] The judgment model generation unit 135 verifies the validity of the conditional expressions calculated by the conditional expression calculation unit 134 and generates a judgment model.

[0116] <Disposal of battery cell handling equipment> The processing performed by the battery cell handling device 40 according to this embodiment will be described with reference to FIG.

[0117] The CPU 130 inputs data such as the battery voltage and battery temperature of the battery cells obtained in the production process or monitoring process from the data storage unit 1000, which is, for example, a data server, via the interface unit 600 (step S510).

[0118] The non-defective product determination unit 530 performs calculation processing using the determination model stored in the determination model storage unit 420, using as input data battery cell data such as the battery voltage and battery temperature obtained in the production process or monitoring process (step S520). This provides the relationship between the charge / discharge voltage of the battery cell and temperature change (temperature characteristics).

[0119] The non-defective product determination unit 530 determines whether or not the battery cell can be determined to be non-defective (normal) based on the result of the calculation using the determination model (step S530). Here, the determination of whether the battery cell is a good or bad product is made, for example, by checking whether the result calculated using a determination model based on temperature characteristics is within a range for determining that the battery cell is a good product.

[0120] If the battery cell is determined to be non-defective (normal) ("NO" in step S530), it is determined that there is some kind of abnormality in the battery cell, and the user of the power storage system in which the battery cell is installed is notified of the abnormality in the battery cell (step S540). At this time, for safety reasons, the operation of the power storage system itself may be prohibited by remote control.

[0121] On the other hand, if the non-defective product determination unit 520 determines that the battery cell is non-defective (normal) ("YES" in step S530), the life prediction unit 900 then determines whether the battery cell has deteriorated (step S550). Here, whether or not a battery cell has deteriorated can be determined based on the calculation results obtained by calculating the measured battery voltage and battery temperature using a determination model.

[0122] If the life prediction unit 900 determines that the battery cell shows signs of deterioration and that the battery cell has a short life ("NO" in step S550), it warns the user and urges them to take measures such as strengthening monitoring of the battery cell and replacing the battery cell in advance to prevent the battery cell from rapidly deteriorating and being unable to function properly in the event of a power outage or the like (step S570).

[0123] On the other hand, if it is determined that there are no signs of deterioration in the battery cell ("YES" in step S550), the battery cell continues to be monitored without issuing any particular warning to the user (step S560).

[0124] As described above, according to this embodiment, the life prediction unit 900 determines whether or not the battery voltage and battery temperature of the battery cell being monitored in the monitoring process, among the battery cells that have been determined to be good by the good product determination unit 530, are within normal ranges.

[0125] In addition, the service life of the battery cell in question will be predicted, and if the service life of the battery cell is short, the user will be warned of this fact and monitoring of the battery cell in question will be strengthened, encouraging the user to take measures such as replacing the battery cell in advance to prevent the battery cell from rapidly deteriorating and being unable to function properly in the event of a power outage, etc. Therefore, by applying the judgment model to the battery cell monitoring process, the battery cells can be monitored more precisely.

[0126] In the above embodiment, a judgment model is created and various selections are made using this judgment model (selection of good or defective products, selection of whether or not to reuse products, selection of monitoring degree), but this is not limited to this.

[0127] For example, as described above, the number of generated determination models may not be one but multiple.

[0128] For example, if multiple judgment models are created, each with different information required for judgment, they can be used appropriately to make a judgment using each judgment model after a processing step that allows judgment, and selection processing can be performed appropriately each time.

[0129] Furthermore, for example, if a selection process is performed by determining whether a battery cell is good or bad using a plurality of judgment models and this is performed for each appropriate processing step, it is possible to remove defective battery cells as appropriate in each processing step, thereby making processing more efficient without performing unnecessary processing on defective battery cells.

[0130] Furthermore, for example, if one judgment model can determine whether a battery cell is defective with a probability of 90%, and another judgment model using information different from the information required by the first judgment model can be used to determine whether the remaining 10% of battery cells are defective with a probability of about 90%, then after performing the selection process for good or defective battery cells using the first judgment model, the other judgment model can be used to select good or defective battery cells for battery cells that could not be determined to be defective. In this way, by preparing different judgment models and performing selection processing at two or more stages, it is possible to extract defective battery cells with higher accuracy, thereby improving the efficiency of processing defective battery cells.

[0131] The battery cell handling device of the present invention can be realized by recording the battery cell sorting method on a computer-readable recording medium and having the program recorded on this recording medium read and executed by the battery cell handling device or the like. The computer system here includes hardware such as an OS and peripheral devices.

[0132] Furthermore, if a WWW (World Wide Web) system is used, the "computer system" also includes the homepage providing environment (or display environment). The program may also be transmitted from a computer system in which the program is stored in a storage device or the like to another computer system via a transmission medium or by transmission waves in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has the function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.

[0133] The program may also be for realizing part of the above-mentioned functions. Furthermore, the above-mentioned functions may be realized in combination with a program already recorded in the computer system, that is, a so-called differential file (differential program).

[0134] The above describes an embodiment of the present invention in detail with reference to the drawings, but the specific configuration is not limited to this embodiment, and includes designs within the scope of the gist of the present invention. [Explanation of symbols]

[0135] 10. Battery cell handling device 20. Battery cell handling device 30: Battery cell handling device 40: Battery cell handling device 100;CPU 101; Regularity information extraction part 102;Regularity information storage unit 103;Singularity extraction part 104: Conditional expression calculation section 105: Decision model generation unit 110;CPU 111; Regularity information extraction part 112;Regularity information storage unit 113;Singularity extraction part 114: Conditional expression calculation section 115: Decision model generation unit 120;CPU 121; Regularity information extraction part 122;Regularity information storage unit 123;Singularity extraction part 124: Conditional expression calculation section 125: Decision model generation unit 130;CPU 131; Regularity information extraction part 132;Regularity information storage unit 133;Singularity extraction part 134; Conditional expression calculation section 135: Decision model generation unit 200;ROM 300;RAM 400: Decision model memory unit 410: Decision model memory unit 420: Decision model memory unit 430: Decision model memory unit 500; Good product judgment department 510;Good product judgment department 520;Good product judgment department 530;Good product judgment department 600;Interface section 700: Aging determination section 800;Reuse Judgment Department 900; Life expectancy prediction section 1000: Data storage unit

Claims

1. creating a determination model that can distinguish between a good battery cell and a defective battery cell using data obtained in each step of a battery cell handling process that includes a plurality of steps; In each of the steps, a step of acquiring various data for each battery cell; a step of using the judgment model to judge whether the battery cell is a good product or a defective product, using the judgment model as input data, data necessary for judging the battery cell using the judgment model, out of the data acquired in each step; a step of treating the battery cells determined as non-defective and the battery cells determined as defective differently according to the results of the determination; Including, One of the steps is a recovery step, a step of determining whether the battery cell is good or bad using the determination model, and if the battery cell is determined to be good, determining whether the capacity value when fully charged and the temperature characteristics are both within a reusable range.

2. 2. The battery cell sorting method according to claim 1, further comprising a step of discarding the battery cell if the battery cell is determined to be defective in the step of determining whether the battery cell is defective or not using the determination model.

3. 3. The battery cell sorting method according to claim 1, wherein the judgment model is created by extracting regularity information between each piece of data relating to the good or defective battery cells from a plurality of past data relating to the good or defective battery cells, extracting singular points between the good and defective battery cells from common points between the good or defective battery cells based on the extracted regularity information, and using data obtained in each process to create a conditional expression that can distinguish between the good battery cells and the defective battery cells based on the singular points.

4. 3. The battery cell sorting method according to claim 1, wherein the data obtained in each step is data obtained during the manufacturing process of the battery cells.

5. a judgment model generation unit that generates a judgment model that can distinguish between a good battery cell and a defective battery cell using data obtained in each process of handling the battery cell, which includes a plurality of processes; a data acquisition unit that acquires various data for each battery cell in each of the steps; a quality determination unit that uses data necessary for determining the battery cell using the determination model as input data from among the data acquired in each of the steps, and determines whether the battery cell is a quality or a defect using the determination model; a battery cell handling unit that handles the battery cells that are determined to be non-defective and the battery cells that are determined to be defective differently depending on the results of the determination; Including, One of the steps is a recovery step, a reuse determination unit that, when the battery cell is determined to be a good product in the step of determining whether the battery cell is good or bad using the determination model, determines whether the capacity value and temperature characteristics when fully charged are both within a range in which the battery cell can be reused.

6. 6. The battery cell handling device according to claim 5, wherein the reuse determination unit determines whether or not the battery cells determined to be non-defective by the non-defective determination unit are reusable, and battery cells determined to be non-reusable as a result of this determination are subject to disposal.

7. a regularity information extraction unit that extracts regularity information between data on the non-defective or defective battery cells from a plurality of past data on the non-defective or defective battery cells in each of the processes; a storage unit that stores the regularity information between the data regarding the non-defective or defective battery cells extracted by the regularity information extraction unit; a singularity extraction unit that receives the regularity information between the data relating to the non-defective or defective battery cells from the storage unit and extracts commonalities between the non-defective or defective battery cells and singularities between the non-defective and defective battery cells; a conditional expression calculation unit that calculates a conditional expression that can distinguish between the non-defective battery cell and the defective battery cell based on the singular point; a judgment model generation unit that verifies the validity of the conditional expression calculated by the conditional expression calculation unit and generates the judgment model; 7. The battery cell handling device according to claim 5, further comprising:

8. 7. The battery cell handling device according to claim 5, wherein the data is data obtained during the manufacturing process of the battery cells.

Images