Battery management device, battery management method, and battery management system

Abstract

According to some embodiments, a battery management device comprises: an interface for acquiring battery data of a battery; and a controller, which extracts physical parameters of the battery on the basis of the battery data, classifies a state of the battery corresponding to the physical parameters by using a first model trained, with a supervised learning algorithm, on a plurality of degradation experimental data for the battery, and calculates the probability that the battery is in an abnormal state by using a second model fitted on the basis of the classification results of the first model.

Description

Battery management device, battery management method, and battery management system

[0001] Cross-citation with related applications

[0002] This application claims the benefit of priority from Republic of Korea Patent Application No. 10-2024-0080287, filed June 20, 2024, the entire disclosure of which is incorporated herein by reference.

[0003] Technology field

[0004] Embodiments disclosed in this document relate to a battery management device, a battery management method, and a battery management system.

[0005] Recently, active research and development has been conducted on secondary batteries. The term "secondary battery" refers to a rechargeable battery, encompassing both conventional Ni / Cd and Ni / MH batteries, as well as recent lithium-ion batteries. Among secondary batteries, lithium-ion batteries can boast higher energy densities than conventional Ni / Cd and Ni / MH batteries. They can be manufactured in small and lightweight designs, making them highly versatile power sources for mobile devices. Recently, their use has expanded to include power sources for electric vehicles, attracting attention as a next-generation energy storage medium.

[0006] State of Health (SOH) can indicate a battery's remaining lifespan and current performance status. A battery's SOH can tend to decline rapidly under certain circumstances. Previously, SOH values ​​were measured periodically, and only the current condition could be determined based on the slope. Because battery abnormalities caused by declining SOH can cause inconvenience and risk to users, predicting the timing of SOH decline in advance may be necessary. However, the difficulty of such prediction using conventional methods can be problematic.

[0007] One of the objects of the embodiments disclosed in this document includes providing a battery management device, a battery management method, and a battery management system capable of estimating the occurrence time of an abnormal state regarding SOH reduction based on physical parameters of a battery.

[0008] The technical objectives of the embodiments disclosed in this document are not limited to the technical tasks mentioned above, and other technical tasks not mentioned will be clearly understood by those skilled in the art from the descriptions below.

[0009] According to some embodiments, a battery management device includes an interface configured to acquire battery data of a battery; and a controller configured to extract physical parameters of the battery based on the battery data, classify a state of the battery corresponding to the physical parameters using a first model learned by a supervised learning algorithm for a plurality of degradation experimental data of the battery, and calculate a probability that the state of the battery is abnormal using a second model fitted by a classification result of the first model.

[0010] According to some embodiments, the first model is configured to be learned based on physical parameters extracted from each of the plurality of degenerate experimental data and anomaly labeling data added in advance for the plurality of degenerate experimental data.

[0011] According to some embodiments, the physical parameters include a first feature and a second feature used for learning the first model, wherein the first feature includes loss of active material (LAM) and the second feature includes loss of lithium source (LLI).

[0012] According to some embodiments, the controller is configured to derive a first OCV value at a beginning of life (BOL) and a second OCV value at a middle of life (MOL) based on the battery data, and to extract the physical parameter based on the first OCV value and the second OCV value.

[0013] According to some embodiments, the second model comprises a logistic regression model trained to calculate the probability based on the physical parameter and the state of the battery.

[0014] According to some embodiments, the controller is further configured to provide a notification function to a user of the battery based on the state of the battery and the probability.

[0015] According to some embodiments, the controller is configured to provide the user with a time at which the state of the battery is expected to become abnormal based on the state of the battery and the probability.

[0016] According to some embodiments, a battery management method includes the steps of: obtaining battery data of a battery; extracting physical parameters of the battery based on the battery data; classifying a state of the battery corresponding to the physical parameters using a first model learned by a supervised learning algorithm for a plurality of degradation experimental data of the battery; and calculating a probability that the state of the battery is abnormal using a second model fitted by a classification result of the first model.

[0017] According to some embodiments, the first model is configured to be learned based on physical parameters extracted from each of the plurality of degenerate experimental data and anomaly labeling data added in advance for the plurality of degenerate experimental data.

[0018] According to some embodiments, the physical parameters include a first feature and a second feature used for learning the first model, wherein the first feature includes loss of active material (LAM) and the second feature includes loss of lithium source (LLI).

[0019] According to some embodiments, the step of extracting the physical parameter includes: deriving a first OCV value at a beginning of life (BOL) and a second OCV value at a middle of life (MOL) based on the battery data; and extracting the physical parameter based on the first OCV value and the second OCV value.

[0020] According to some embodiments, the second model comprises a logistic regression model trained to calculate the probability based on the physical parameter and the state of the battery.

[0021] According to some embodiments, the method further comprises providing a notification function to a user of the battery based on the state of the battery and the probability.

[0022] According to some embodiments, the step of providing the notification function includes the step of providing the user with a time at which the state of the battery is expected to become abnormal based on the state of the battery and the probability.

[0023] According to some embodiments, a battery management system includes: a battery; a charger / discharger configured to charge or discharge the battery; and a battery management device configured to obtain battery data of the battery, extract physical parameters of the battery based on the battery data, classify a state of the battery corresponding to the physical parameters using a first model learned by a supervised learning algorithm for a plurality of degradation experimental data for the battery, and calculate a probability that the state of the battery is abnormal using a second model fitted by a classification result of the first model.

[0024] According to the embodiments disclosed in this document, a battery management device, a battery management method, and a battery management system capable of estimating the occurrence time of an abnormal state regarding SOH reduction based on physical parameters of a battery can be provided.

[0025] The technical effects according to the embodiments disclosed in this document are not limited to the effects mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art according to the disclosure of this document.

[0026] FIG. 1 illustrates elements constituting a battery management system according to some embodiments.

[0027] FIG. 2 illustrates elements constituting a battery management device according to some embodiments.

[0028] Figure 3 illustrates a conventional technique for determining a battery abnormality based on the SOH slope.

[0029] FIG. 4 illustrates a process of training an AI model according to some embodiments and a process of calculating a battery status and abnormality probability using the AI ​​model.

[0030] FIG. 5 illustrates a method for training a first model using physical parameters of a battery according to some embodiments.

[0031] Figure 6 illustrates a method of training a second model using the classification results of a first model according to some embodiments.

[0032] FIG. 7 illustrates steps of a battery management method according to some embodiments.

[0033] Hereinafter, embodiments described in this document are described with reference to the attached drawings. However, this is not intended to limit the disclosure of this document to specific embodiments, and it should be understood that various modifications, equivalents, and / or alternatives of the embodiments described in this document are included.

[0034] The embodiments and terminology used in this document are not intended to limit the technical features described in this document to a specific embodiment, but should be understood to encompass various modifications, equivalents, or alternatives of the embodiment. In connection with the description of the drawings, similar reference numerals may be used to refer to similar or related components. The singular form of a noun corresponding to an item may include one or more of the item, unless the context clearly indicates otherwise.

[0035] In this document, the phrases "A or B", "at least one of A and B", "at least one of A or B", "A, B, or C", "at least one of A, B, and C", and "at least one of A, B, or C" can each include any one of the items listed together in that phrase, or all possible combinations thereof. Terms such as "first", "second", "first", "second", "A", "B", "(a)", or "(b)" may be used merely to distinguish the corresponding component from other corresponding components, and do not limit the corresponding components in any other respect (e.g., importance or order) unless specifically stated otherwise.

[0036] In this document, when a component (e.g., a first component) is referred to as being “connected,” “coupled,” or “connected,” with or without the terms “functionally” or “communicatively,” or “coupled” or “connected,” it means that the component can be connected to the other component directly (e.g., wired or wirelessly), or indirectly (e.g., via a third component).

[0037] The methods according to various embodiments disclosed in this document may be provided as a computer program product. The computer program product may be traded as a commodity between a seller and a buyer. The computer program product may be distributed in the form of a machine-readable storage medium (e.g., compact disc read-only memory, CD-ROM), or may be distributed online (e.g., downloaded or uploaded) through an application store or directly between two driver devices. In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily generated in a machine-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or a relay server.

[0038] According to the embodiments disclosed in this document, each component (e.g., a module or a program) of the above-described components may include one or more entities, and some of the entities may be separated and placed in other components. According to the embodiments disclosed in this document, one or more components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Alternatively or additionally, a plurality of components (e.g., a module or a program) may be integrated into a single component. In this case, the integrated component may perform one or more functions of each of the plurality of components identically or similarly to those performed by the corresponding component among the plurality of components prior to the integration. According to the embodiments disclosed in this document, the operations performed by a module, program, or other component may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, omitted, or one or more other operations may be added.

[0039] FIG. 1 may illustrate elements constituting a battery management system according to some embodiments.

[0040] Referring to FIG. 1, the battery management system (100) may include a charger / discharger (110), a battery (120), and a battery management device (130). However, the present invention is not limited thereto, and some components may be omitted from the battery management system (100) or other general-purpose components may be further included in the battery management system (100).

[0041] The charger / discharger (110) may be configured to charge or discharge the battery (120). For example, the charger / discharger (110) may include a power supply device, etc. The charger / discharger (110) may apply charge / discharge cycles to the battery (120) to perform a degradation experiment of the battery (120). According to an embodiment, the charger / discharger (110) may include a power-using device, and the power-using device may include a mobility device such as an electric vehicle (EV), a hybrid electric vehicle (HEV), or an electric bike.

[0042] The battery (120) may include a battery pack or the like that is managed by the system (100). The battery pack of the battery (120) may include a plurality of battery modules, and each battery module may include a plurality of battery cells. According to an embodiment, the battery (120) may be mounted on various types of mobility devices. To artificially create various degradation scenarios, charge and discharge cycles by the charger (110) may be applied to the battery (120).

[0043] The battery management device (130) can perform operations for diagnosing or managing the battery (120). The battery management device (130) can acquire battery data of the battery (120) and diagnose or manage the status of the battery (120) to be diagnosed based on the battery data. According to an embodiment, the battery management device (130) can include a battery management system (BMS) configured together with the battery (120) in an on-board manner, and / or an external device remotely placed from the battery (120) in an off-board manner. The external device can include a charger of a battery charging station, a battery diagnostic device, a cloud computing server, etc.

[0044] The battery management system (100) may further include a management server. The management server may manage the management results of the battery management device (130). The management server may exchange data with the battery management device (130) via wired / wireless communication. When a defect in the battery (120) is diagnosed or its lifespan is predicted, the results may be transmitted to the management server and recorded in a database. According to an embodiment, the battery management device (130) may perform diagnostic operations by executing battery management software, and the management server may provide update information of the battery management software to the battery diagnostic device (130).

[0045] FIG. 2 may illustrate elements constituting a battery management device according to some embodiments.

[0046] Referring to FIG. 2, the battery management device (130) may include an interface (131) and a controller (132). However, the present invention is not limited thereto, and some components may be omitted from the battery management device (130), or other general-purpose components may be further included in the battery management device (130).

[0047] The interface (131) can obtain battery data of the battery (120). According to an embodiment, the interface (131) may include a communication unit configured to receive battery data and / or a sensor unit configured to measure battery data. According to an embodiment, when the battery management device (130) is implemented in an off-board form, the communication unit may receive battery data in a wired data communication, wireless data communication, or the like. Alternatively, when the battery management device (130) is implemented in an on-board form, the sensor unit may be configured to measure values ​​such as voltage, current, temperature, and resistance from the battery (120).

[0048] The controller (132) may have a structure for executing commands that implement the operations of the battery management device (130). The controller (132) may be implemented as an array of multiple logic gates for processing various operations or as a general-purpose microprocessor, and may be composed of a single processor or multiple processors. For example, the controller (132) may be implemented in the form of at least one of a microprocessor, a CPU, a GPU, and an AP.

[0049] The controller (132) can operate with a memory configured to store various data, commands, mobile applications, computer programs, etc. The memory can be configured separately from or integrally with the controller (132). The controller (132) can process various operations by executing commands stored in the memory. For example, the memory can be implemented as a non-volatile device such as a ROM, a PROM, an EPROM, an EEPROM, a flash memory, a PRAM, an MRAM, an RRAM, an FRAM, etc., or a volatile device such as a DRAM, an SRAM, an SDRAM, a PRAM, etc., and can be implemented in the form of an HDD, an SSD, an SD, a Micro-SD, etc., or a combination thereof.

[0050] The interface (131) may be configured to acquire battery data of the battery (120). The battery data may include voltage, current, temperature, resistance, etc. of the battery (120). The battery data may be acquired in units such as cells, modules, and packs of the battery (120). The battery data may include time-series data composed of values ​​collected at regular intervals. According to an embodiment, the battery data may include data measured while the battery (120) is being charged or discharged by the charger / discharger (110).

[0051] The controller (132) may be configured to extract physical parameters of the battery (120) based on battery data. The physical parameters may refer to parameters representing physical characteristics of the battery (120). For example, the physical parameters may include loss of active material (LAM), loss of lithium source (LLI), etc. The physical parameters may function as machine learning features utilized to identify a state in which the SOH of the battery (120) rapidly decreases.

[0052] The controller (132) may be configured to classify the state of the battery (120) corresponding to the physical parameters using a first model learned by a supervised learning algorithm for a plurality of degradation experimental data for the battery (120). The plurality of degradation experimental data may be generated by repeatedly charging and discharging the battery (120) under different degradation conditions. For example, the charger / discharger (110) may apply different charge / discharge cycles to the battery (120), and data output by the battery (120) during the cycle may be collected as the plurality of degradation experimental data. For example, the plurality of degradation experimental data may include combinations of different loss of active material (LAM) values ​​and loss of lithium source (LLI) values, and a label value may be assigned to each combination. The label value may indicate whether the battery (120) is abnormal according to each degradation condition. The plurality of degradation experimental data may be used as training data for training the first model. For example, the first model may include a support vector machine (SVM) model. The first model, after completing training, can receive physical parameters as input and output the corresponding battery status.

[0053] The controller (132) may be configured to calculate the probability that the battery (120) is in an abnormal state using a second model fitted by the classification result of the first model. Once the training of the first model is completed, the fitting of the second model may be performed using the first model. Through the fitting, the probability of the battery (120) being in an abnormal state output by the second model may be optimized. For example, the second model may be trained to output the probability of the battery (120) being in an abnormal state using logistic regression based on physical parameters and the output of the first model. Referring to the outputs of the first and second models, it may be determined whether the battery (120) is currently in an abnormal state, and if not, what the probability of the battery (120) being in an abnormal state is.

[0054] According to an embodiment, the first model may be configured to learn based on physical parameters extracted from each of a plurality of degradation experimental data and anomaly labeling data added in advance to the plurality of degradation experimental data. The anomaly labeling data may be added in advance to the plurality of degradation experimental data, through which it may be determined whether the degradation experimental conditions of each degradation experimental data cause anomalies or failures in the battery (120). The state of the battery (120) classified by the anomaly labeling data may be expressed by physical parameters. The first model may learn the relationship between the physical parameters and the state of the battery (120).

[0055] In an embodiment, the physical parameters may include first features and second features used for training the first model, wherein the first feature may include loss of active material (LAM) and the second feature may include loss of lithium source (LLI). The physical parameters may represent a combination of LAM values ​​and LLI values, and the first model may learn how the combination of LAM values ​​and LLI values ​​is related to the state of the battery (120).

[0056] According to an embodiment, the controller (132) may be configured to derive a first OCV value at the beginning of life (BOL) and a second OCV value at the middle of life (MOL) based on battery data, and extract physical parameters based on the first OCV value and the second OCV value. A LAM value and an LLI value may be calculated based on the OCV difference at the BOL and MOL points.

[0057] In an embodiment, the second model may include a logistic regression model trained to calculate probabilities based on physical parameters and the state of the battery (120). Once the first model is trained, the first model may output the state of the battery (120). Furthermore, by taking the physical parameters into account, the second model may output an abnormality probability of the battery (120). Alternatively, other suitable models besides logistic regression may be applied to the second model.

[0058] In an embodiment, the controller (132) may be further configured to provide a notification function to the user of the battery (120) based on the state and probability of the battery (120). For example, the notification function may be provided to the user's mobile terminal and / or a vehicle equipped with the battery (120). The notification content may include whether there is currently an abnormality in the battery (120) and the probability of the abnormality.

[0059] In an embodiment, the controller (132) may be configured to provide the user with a time point at which the battery (120) is expected to enter an abnormal state based on the state and probability of the battery (120). Although the battery (120) is not currently in an abnormal state, if the probability of the battery (120) entering an abnormal state is high, the battery (120) may enter an abnormal state in the near future. Considering the physical parameters of the battery and the probability of the abnormal state, the expected time point for the abnormal state can be calculated. In an embodiment, the second model may be additionally trained to output the expected time point along with the probability of the battery (120) entering an abnormal state. Alternatively, the controller (132) may be configured to additionally introduce a third model separate from the first and second models to calculate the expected time point for the abnormal state. For example, the third model may be configured to output the expected time point earlier as the probability of the abnormality increases.

[0060] Figure 3 illustrates a conventional technique for determining a battery abnormality based on the SOH slope.

[0061] Referring to FIG. 3, a graph (300) illustrating a conventional technique for determining a battery abnormality based on an SOH slope can be illustrated.

[0062] In the prior art, a trend of fluctuation in the SOH value as the battery discharges is checked, and if the magnitude of the SOH slope exceeds a specific value and / or a specific ratio, it is determined that an abnormal condition has occurred.

[0063] However, because prior techniques require accumulated past data to determine SOH slope trends, anomaly detection can be difficult when only current data is available. Furthermore, even when considering past data, only the current state can be determined, making it difficult to predict future conditions.

[0064] FIG. 4 illustrates a process of training an AI model according to some embodiments and a process of calculating a battery status and abnormality probability using the AI ​​model.

[0065] Referring to FIG. 4, a process of training an AI model (410) and a process of calculating a battery status and abnormality probability using the AI ​​model (420) can be illustrated.

[0066] In step (411), multiple degradation experiment data can be collected using various types of DOEs. In step (412), physical parameters such as LLI and LAM can be extracted for each experiment. These can be utilized as features for AI model training. In step (413), a labeling task can be performed to distinguish normal and abnormal states for each experiment. This labeling task can be prepared in advance before the model training process.

[0067] In step (414), a classification AI model may be trained. For multiple DOEs, how combinations of LLI and LAM values ​​are associated with labeled battery states may be trained. In step (415), a probabilistic AI model providing anomaly probabilities may be trained. Once both the classification AI model and the probabilistic AI model are trained, model training may be completed in step (416).

[0068] In step (421), physical parameters such as LLI values ​​and LAM values ​​can be extracted from the battery that is the target of state estimation. In step (422), the physical parameters can be input into an AI model. In step (423), the AI ​​model can provide battery status and abnormality probability. In step (424), a notification function can be provided to the user based on the battery status and abnormality probability. The notification function can notify the current battery status and the expected time of abnormality occurrence.

[0069] FIG. 5 illustrates a method for training a first model using physical parameters of a battery according to some embodiments.

[0070] Referring to FIG. 5, a graph (500) illustrating a method of training a first model using physical parameters of a battery may be illustrated.

[0071] Referring to graph (500), data points corresponding to combinations of LAM values ​​on the horizontal axis and LLI values ​​on the vertical axis can be plotted on a two-dimensional graph. Each of the plotted data points can correspond to multiple degenerate experimental data. The data points can be labeled as either normal or abnormal.

[0072] The first model can establish a baseline (510) based on the distribution of data points. The baseline (510) can be utilized to distinguish between normal and abnormal combinations of LAM and LLI values. In addition to a straight line, the baseline (510) can be configured as various curved shapes or a combination of straight and curved shapes. The first model can adjust the position of the baseline (510) through learning to most appropriately classify data points of multiple degenerate experimental data.

[0073] Once the training of the first model is completed, the location of the baseline (510) can be determined. Then, when a combination of LAM and LLI values ​​of the target battery is input into the first model, the state of the target battery can be classified by comparing the data points of the combination with the baseline (510).

[0074] Figure 6 illustrates a method of training a second model using the classification results of a first model according to some embodiments.

[0075] Referring to FIG. 6, a graph (600) illustrating a method of training a second model using the classification results of a first model can be illustrated.

[0076] The two-dimensional region of the graph (600) can be distinguished by different colors according to the abnormality probability. For example, the legend (610) can indicate the matching relationship between colors and abnormality probability. The two-dimensional region of the graph (600) can be formed based on the reference line (510) of the graph (500) of FIG. 5.

[0077] For example, once the training of the first model is complete, the location of the baseline (510) can be determined. The location of the baseline (510) may correspond to an abnormality probability of approximately 0.5. Accordingly, the second model can be trained to adjust the boundary line representing an abnormality probability of 0.5 to be identical to the location of the baseline (510). Once the boundary line representing an abnormality probability of 0.5 is determined, the boundaries corresponding to other values ​​of abnormality probability can also be adjusted based on the parallel translation thereof.

[0078] Once the locations of the boundaries representing each abnormality probability are determined, the training of the second model can be completed. When a combination of LAM and LLI values ​​of the target battery is input into the second model, it can first be determined whether the combination is classified as normal or abnormal. If the combination is classified as normal, an abnormality probability value can be provided based on the boundaries of the second model. In an embodiment, the second model can be configured to provide an abnormality probability value for a combination of LAM and LLI values ​​that is classified as abnormal.

[0079] FIG. 7 illustrates steps of a battery management method according to some embodiments.

[0080] Referring to FIG. 7, the battery management method (700) may include steps (710) to (740). However, the present invention is not limited thereto, and some steps may be omitted or other general steps may be added, and the steps of the battery management method (700) may be executed in a different order than the illustrated order.

[0081] The battery management method (700) may be composed of steps that are processed in a time-series manner in the battery management device (130). Therefore, even if the content is omitted below, the content described above for the battery management device (130) may be equally applied to the battery management method (700).

[0082] Steps (710) to (740) of the battery management method (700) can be performed by the interface (131) and controller (132) of the battery management device (130).

[0083] In step (710), the battery management device (130) may perform a step of acquiring battery data of the battery.

[0084] In step (720), the battery management device (130) may perform a step of extracting physical parameters of the battery based on battery data.

[0085] In step (730), the battery management device (130) may perform a step of classifying the state of the battery corresponding to the physical parameter using a first model learned by a supervised learning algorithm for multiple degradation experimental data for the battery.

[0086] In step (740), the battery management device (130) may perform a step of calculating the probability that the battery is in an abnormal state using a second model fitted by the classification result of the first model.

[0087] According to an embodiment, the battery management method (700) may be implemented in the form of a computer program stored on a computer-readable storage medium. That is, the computer program may include instructions for implementing the battery management method (700), and the instructions of the program may be stored on the computer-readable storage medium. The computer program may include a mobile application.

[0088] According to an embodiment, the computer-readable storage medium may include magnetic media such as hard disks, floppy disks, and magnetic tapes, optical media such as CD-ROMs, DVDs, magneto-optical media such as floptical disks, and hardware devices specially configured to store and execute computer program instructions such as ROMs, RAMs, flash memories, and the like. The computer program instructions may include machine language codes generated by a compiler and high-level language codes that can be executed by a computer using an interpreter, etc.

[0089] The terms "include," "comprise," or "have" used herein, unless otherwise specifically stated, imply that the corresponding component may be included, and therefore should be interpreted to include other components rather than to exclude other components. All terms, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art to which the embodiments disclosed in this document pertain, unless otherwise defined. Commonly used terms, such as terms defined in dictionaries, should be interpreted to be consistent with their contextual meaning in the relevant art, and shall not be interpreted in an idealized or overly formal sense, unless explicitly defined in this document.

[0090] The above description is merely an example of the technical idea disclosed in this document, and those skilled in the art to which the embodiments disclosed in this document pertain may make various modifications and variations without departing from the essential characteristics of the embodiments disclosed in this document. Therefore, the embodiments disclosed in this document are not intended to limit the technical idea of ​​the embodiments disclosed in this document, but to explain it, and the scope of the technical idea disclosed in this document is not limited by these embodiments. The scope of protection of the technical idea disclosed in this document should be interpreted by the claims below, and all technical ideas within a scope equivalent thereto should be interpreted as being included in the scope of rights of this document.

[0091] [Explanation of symbols]

[0092] 100: Battery management system 110: Charger / discharger

[0093] 120: Battery 130: Battery management unit

[0094] 131: Interface 132: Controller

Claims

1. An interface configured to obtain battery data of a battery; and Extracting the physical parameters of the battery based on the above battery data, Classifying the state of the battery corresponding to the physical parameter using a first model learned by a supervised learning algorithm on multiple degradation experiment data for the battery, A battery management device comprising a controller configured to calculate a probability that the battery is in an abnormal state using a second model fitted by the classification result of the first model.

2. In paragraph 1, A battery management device, wherein the first model is configured to be learned based on physical parameters extracted from each of the plurality of degenerate experimental data and ideal labeling data added in advance to the plurality of degenerate experimental data.

3. In paragraph 1, The above physical parameters include first features and second features used for learning the first model, A battery management device, wherein the first feature includes loss of active material (LAM) and the second feature includes loss of lithium source (LLI).

4. In paragraph 1, The controller derives a first OCV value at the beginning of life (BOL) and a second OCV value at the middle of life (MOL) based on the battery data, A battery management device configured to extract the physical parameter based on the first OCV value and the second OCV value.

5. In paragraph 1, A battery management device, wherein the second model comprises a logistic regression model trained to calculate the probability based on the physical parameters and the state of the battery.

6. In paragraph 1, A battery management device, wherein the controller is further configured to provide a notification function to a user of the battery based on the state of the battery and the probability.

7. In paragraph 6, A battery management device, wherein the controller is configured to provide the user with a time at which the state of the battery is expected to become abnormal based on the state of the battery and the probability.

8. Step of acquiring battery data of the battery; A step of extracting physical parameters of the battery based on the battery data; A step of classifying the state of the battery corresponding to the physical parameter using a first model learned by a supervised learning algorithm for multiple degradation experimental data for the battery; and A battery management method, comprising a step of calculating the probability that the battery is in an abnormal state using a second model fitted by the classification result of the first model.

9. In paragraph 8, A battery management method, wherein the first model is configured to be learned based on physical parameters extracted from each of the plurality of degenerate experimental data and ideal labeling data added in advance to the plurality of degenerate experimental data.

10. In paragraph 8, The above physical parameters include first features and second features used for learning the first model, A battery management method, wherein the first feature includes loss of active material (LAM) and the second feature includes loss of lithium source (LLI).

11. In paragraph 8, The step of extracting the above physical parameters is: A step of deriving a first OCV value at the beginning of life (BOL) and a second OCV value at the middle of life (MOL) based on the above battery data; and A battery management method, comprising a step of extracting the physical parameter based on the first OCV value and the second OCV value.

12. In paragraph 8, A battery management method, wherein the second model comprises a logistic regression model trained to calculate the probability based on the physical parameters and the state of the battery.

13. In paragraph 8, A battery management method further comprising the step of providing a notification function to a user of the battery based on the state of the battery and the probability.

14. In paragraph 13, The steps for providing the above notification function are: A battery management method comprising the step of providing the user with a time at which the state of the battery is expected to become abnormal based on the state of the battery and the probability.

15. Battery; A charger configured to charge or discharge the battery; and A battery management system comprising a battery management device configured to obtain battery data of the battery, extract physical parameters of the battery based on the battery data, classify the state of the battery corresponding to the physical parameters using a first model learned by a supervised learning algorithm for a plurality of degradation experiment data for the battery, and calculate the probability that the state of the battery is abnormal using a second model fitted by the classification result of the first model.

Images