Battery management device and battery management method

Abstract

A battery management device disclosed in the present document comprises: a communication unit for receiving battery data of a battery cell; and a control unit. The control unit: derives an average voltage value for each of a plurality of sections included in the battery data; inputs some of the average voltage values for the plurality of sections into a self-supervised learning-based learning model; and uses a voltage prediction value output by the learning model to detect whether the battery cell is defective.

Description

Battery management device and battery management method

[0001] Cross-citation with related applications

[0002] This application claims the benefit of priority from Republic of Korea Patent Application No. 10-2023-0174556, filed December 5, 2023, the entire contents of which are incorporated herein by reference.

[0003] Technology field

[0004] The embodiments disclosed in this document relate to a battery management device and a battery management method for diagnosing a battery condition.

[0005] Recently, active research and development is being conducted on secondary batteries. Here, 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 have the advantage of a much higher energy density than conventional Ni / Cd and Ni / MH batteries. Furthermore, lithium-ion batteries can be manufactured in a compact and lightweight form, making them suitable for use as power sources for mobile devices. Furthermore, lithium-ion batteries are attracting attention as a next-generation energy storage medium, as their use is expanding to include power sources for electric vehicles.

[0006] Therefore, research is being conducted on faulty battery cell detection algorithms, as defective battery cells, such as tab open or lithium deposition, can pose safety issues during battery use. Typically, when machine learning techniques are applied to battery safety diagnosis, supervised learning is utilized, with battery cell voltage values ​​as input and defective battery information as output. However, supervised learning has been limited by the lack of data on defective battery information, resulting in poor learning performance and poor output accuracy.

[0007] According to one embodiment disclosed in this document, a battery management device and a battery management method are provided for diagnosing a battery using a battery safety diagnosis index derived by utilizing self-supervised learning.

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

[0009] A battery management device according to one embodiment includes a communication unit that receives battery data of a battery cell, and a control unit that derives voltage average values ​​for a plurality of sections included in the battery data, inputs some of the voltage average values ​​for the plurality of sections into a learning model based on self-supervised learning, and detects whether the battery cell is defective using a voltage prediction value output by the learning model.

[0010] The above control unit can determine the average value of the voltage as the average value of each section in which the voltage is maintained constant in the charge / discharge profile of the battery cell.

[0011] The above learning model may include a plurality of linear regression models formed by excluding some of the voltage average values.

[0012] The above control unit can input some of the voltage average values ​​into each of the plurality of linear regression models to derive a plurality of output values.

[0013] The above control unit can determine the difference between the voltage prediction value, which is the plurality of output values, and the actual value, which is a part of the voltage average value, as an error for each of the plurality of sections.

[0014] The above control unit can determine a diagnostic index of the battery cell by adding the errors for the multiple sections.

[0015] The control unit may determine that the battery cell is defective based on the diagnostic indicator exceeding a preset diagnostic criterion.

[0016] The above self-supervised learning-based learning model can detect whether the battery cell is defective, including a linear mixed model (LMM).

[0017] A battery management method according to one embodiment includes receiving battery data of a battery cell, deriving voltage average values ​​for a plurality of sections included in the battery data, inputting some of the voltage average values ​​for the plurality of sections into a learning model based on self-supervised learning, and detecting whether the battery cell is defective using a voltage prediction value output by the learning model.

[0018] Deriving the average value of the above voltage can derive the average value of the voltage as the average value of each section in which the voltage is maintained constant in the charge / discharge profile of the battery cell.

[0019] The above learning model may include a plurality of linear regression models formed by excluding some of the voltage average values.

[0020] A battery management method according to one embodiment may further include inputting some of the voltage average values ​​into each of the plurality of linear regression models to derive a plurality of output values.

[0021] A battery management method according to one embodiment may further include determining a difference between the voltage prediction value, which is the plurality of output values, and the actual value, which is a part of the voltage average value, as an error for each of the plurality of sections.

[0022] A battery management method according to one embodiment may further include determining a diagnostic indicator of the battery cell by adding the errors for the plurality of sections.

[0023] Detecting whether the battery cell is defective may include determining the battery cell to be defective based on the diagnostic indicator exceeding a preset diagnostic criterion.

[0024] The above self-supervised learning-based learning model can detect whether the battery cell is defective, including a linear mixed model (LMM).

[0025] According to a battery management device according to one embodiment, since it is possible to detect whether a battery is defective based on self-supervised learning, a machine learning model can be trained even when labeled defective battery information is insufficient.

[0026] According to a battery management device according to one embodiment, a diagnostic algorithm having high universality can be provided because it can be applied to all battery data.

[0027] FIG. 1 illustrates a block diagram of a typical battery system including a battery management device according to one embodiment.

[0028] FIG. 2 illustrates a block diagram showing the configuration of a battery management device according to one embodiment.

[0029] FIG. 3 illustrates input and output values ​​of a machine learning model utilized in a battery management device according to one embodiment.

[0030] FIG. 4 illustrates a schematic flowchart of a battery management device detecting a defective battery according to one embodiment.

[0031] FIG. 5 illustrates an average voltage value derived by a battery management device according to one embodiment.

[0032] FIG. 6 illustrates the correlation between normal battery cells and defective battery cells derived by a battery management device according to one embodiment.

[0033] FIG. 7 is a graph illustrating diagnostic indicators derived by a battery management device according to one embodiment.

[0034] FIG. 8 illustrates voltage prediction values ​​and actual values ​​derived from one battery pack by a battery management device according to one embodiment.

[0035] FIG. 9 illustrates an error derived from one battery pack by a battery management device according to one embodiment.

[0036] FIG. 10 illustrates a control flowchart of a battery management method according to one embodiment.

[0037] Hereinafter, various embodiments disclosed in this document will be described in detail with reference to the attached drawings. In this document, identical components in the drawings are designated by the same reference numerals, and redundant descriptions of identical components are omitted.

[0038] With respect to the various embodiments disclosed in this document, specific structural and functional descriptions are merely illustrative for the purpose of explaining the embodiments, and the various embodiments disclosed in this document may be implemented in various forms and should not be construed as being limited to the embodiments described in this document.

[0039] The expressions "first," "second," "first," or "second" used in various embodiments may describe various components, regardless of order and / or importance, and do not limit the components. For example, without departing from the scope of the embodiments disclosed herein, a first component may be renamed a second component, and similarly, a second component may also be renamed a first component.

[0040] The terms used in this document are intended solely to describe specific embodiments and may not be intended to limit the scope of other embodiments. Singular expressions may include plural expressions unless the context clearly indicates otherwise.

[0041] All terms used herein, including technical or scientific terms, have the same meaning as commonly understood by those of ordinary skill in the art of the embodiments disclosed herein. Terms defined in commonly used dictionaries may be interpreted as having the same or similar meaning in the context of the relevant technology, and unless explicitly defined herein, they shall not be interpreted in an idealized or overly formal sense. In some cases, even if a term is defined herein, it cannot be interpreted to exclude the embodiments disclosed herein.

[0042] FIG. 1 illustrates a block diagram showing the configuration of a typical battery system including a battery management device according to various embodiments.

[0043] Specifically, FIG. 1 schematically illustrates a battery system (10) and an upper controller (20) included in an upper system according to one embodiment disclosed in this document.

[0044] As illustrated in FIG. 1, the battery system (10) may include a plurality of battery modules (12), a sensor unit (14), a switching unit (16), and a battery management device (1). At this time, the battery system (10) may be equipped with a plurality of battery modules (12), sensor units (14), switching units (16), and battery management devices (1).

[0045] A plurality of battery modules (12) may include at least one rechargeable battery cell (13). The battery cell (13) may include a cathode, a cathode material, a cathode material, a separator, an electrolyte, a polymer, and a case. In this case, the plurality of battery modules (12) may be connected in series or in parallel.

[0046] The sensor unit (14) may include a voltage sensor (2), a current sensor (3), and a temperature sensor (not shown).

[0047] The voltage sensor (2) can be configured to be connected in parallel to the battery, detect the battery voltage, which is the voltage across both terminals of the battery, and generate a voltage signal representing the detected battery voltage.

[0048] The current sensor (3) can detect the current used in the process of determining the SOC of the battery cell (13).

[0049] The current sensor (3) may include any configuration that generates a signal corresponding to the size of the charging current, and the current sensor (3) may be installed on a charging / discharging path, which is a path through which the charging / discharging current flows in the battery.

[0050] The current sensor (3) can measure the battery current flowing in the battery, i.e., the charging current and the discharging current, and transmit the measurement results to the battery management device (1). According to one embodiment, the current sensor (3) can measure the battery current at predetermined intervals during a charging cycle in which the battery is charged with power from an external device or a discharging cycle in which the battery is discharged, and transmit the measurement results to the battery management device (1).

[0051] The temperature sensor may be configured to measure the battery temperature and generate a temperature signal representing the measured battery temperature. The temperature sensor may be positioned within the case so as to measure a temperature close to the actual temperature of the battery. For example, the temperature sensor may be attached to the surface of at least one battery cell included in the cell group and may detect the surface temperature of the battery cell as the battery temperature.

[0052] The temperature sensor may be configured to measure the external temperature, which is the temperature at a predetermined location away from the battery, and generate a temperature signal representing the measured external temperature. The temperature sensor may be positioned at a predetermined location outside the case where heat exchange between the battery and the atmosphere occurs. According to an embodiment, the temperature sensor may be implemented using one or a combination of two or more known temperature detection elements, such as a thermocouple, a thermistor, or a bimetal. The current flowing in the battery system (10) may be detected. At this time, the detection signal may be transmitted to the battery management device (1).

[0053] In Fig. 1, the sensor unit (14) is connected between the positive electrode of the battery cell (13) and the switching unit (16), but it may also be located on a PCB (Printed Circuit Board) provided around the battery cell (13), and the configurations and connection relationships between the configurations shown in Fig. 1 are only examples and are not limited thereto.

[0054] The switching unit (16) is connected in series to the (+) terminal side or the (-) terminal side of the battery module (12) to control the charge / discharge current flow of the battery module (12). For example, the switching unit (16) may use at least one relay, magnetic contactor, etc. depending on the specifications of the battery system (10).

[0055] The battery management device (1) can monitor the voltage, current, temperature, etc. of the battery system (10) and control and manage it to prevent overcharging and overdischarging, etc., and may include, for example, a BMS (Battery Management System).

[0056] The battery management device (1) is an interface for receiving values ​​measured from various parameters, and may include a plurality of terminals and a circuit connected to these terminals to process the values ​​received. In addition, the battery management device (1) may control the ON / OFF of a switching unit (16), for example, a relay or a contactor, and may be connected to a battery module (12) to monitor the status of each battery module (12).

[0057] In addition, the battery management device (1) can receive temperature data, voltage data, and current data from the sensor unit (14) to obtain battery status information and diagnose the status of the battery.

[0058] The upper controller (20) can transmit a control signal for controlling the battery module (12) to the battery management device (1). Accordingly, the battery management device (1) can be controlled for operation based on the control signal received from the upper controller (20). In addition, the battery module (12) may be a component included in an ESS (Energy Storage System). In this case, the upper controller (20) may be a controller (BBMS) of a battery bank including a plurality of battery systems (10) or an ESS controller that controls the entire ESS including a plurality of banks. However, the battery system (10) is not limited to this purpose.

[0059] FIG. 2 illustrates a block diagram showing the configuration of a battery management device according to one embodiment.

[0060] Referring to FIG. 2, a battery management device (1) according to one embodiment includes a control unit (100) including at least one processor (110) and a memory (120) and a communication unit (200), and can diagnose a battery by communicating with an external device (4) through the communication unit (200).

[0061] According to an embodiment, an external device (4) communicating with a battery management device (1) may include a user terminal and a server device that transmit results diagnosed by the battery management device (1).

[0062] Specifically, when the external device (4) is a user terminal, the control unit (100) of the battery management device (1) can transmit the battery diagnosis results to the user terminal so that the user can check them. At this time, the user terminal may include, but is not limited to, a personal computer, a terminal, a portable telephone, a smart phone, a handheld device, a wearable device, etc.

[0063] In addition, when the external device (4) is a server device, the server device may be implemented as various computing devices such as a workstation, a cloud, a data drive, a data station, etc. The server device may be implemented as one or more server devices that are physically or logically separated based on function, detailed configuration of function, or data, etc., and may transmit and receive data and process the transmitted and received data through communication between each server device.

[0064] A battery management device (1) according to one embodiment may refer to any electronic device including a processor (110) and a memory (120), and may be mounted on a vehicle and operated. Each component of the battery management device (1) will be described in detail below.

[0065] The communication unit (200) may include a wireless communication unit (210) and a wired communication unit (220) to communicate with an external device (4). The communication unit (200) may transmit and receive programs for calculating characteristic values ​​of battery cells, class classification, and lifespan estimation, as well as various data, from a separately provided external server.

[0066] The wireless communication unit (210) may include at least one of a short-range communication module and a long-range communication module.

[0067] The short-range communication module can communicate with an external device (4) adjacent to the battery management device (1) using a short-range communication method. Here, the short-range communication module can utilize one of the following communication methods: Bluetooth, Bluetooth low energy, infrared data association (IrDA), Zigbee, Wi-Fi, Wi-Fi direct, Ultra Wideband (UWB), or near field communication (NFC).

[0068] The remote communication module may include a communication module that performs various types of remote communication and may include a mobile communication unit. The mobile communication unit may transmit and receive wireless signals with at least one of a base station and an external terminal on a mobile communication network. In addition, the remote communication module may communicate with an external device (4) or an external device (4) such as another electronic device through a surrounding access point (AP). The access point (AP) may connect a local area network (LAN) to which the battery management device (1) is connected to a wide area network (WAN) to which a communication server is connected. Accordingly, the battery management device (1) may be connected to the communication server through the wide area network (WAN) with the external device (4) and communicate with each other.

[0069] The wired communication unit (220) can connect to a wired communication network and communicate with an external device (4) through the wired communication network. For example, the wired communication unit (220) can connect to a wired communication network through Ethernet (IEEE 802.3 technology standard) or connect to a wired communication network through CAN communication, and transmit and receive data with the external devices (4) through the wired communication network.

[0070] A battery management device (1) according to one embodiment may include an input / output interface (not shown). An interface may be provided that connects an input device (not shown) such as a keyboard, mouse, or touch panel, an output device (not shown) such as a display, and a processor (110) to transmit and receive data.

[0071] The memory (120) can store various information necessary for operating the battery management device (1). Specifically, the memory (120) can store an operating system and a program necessary for operating the battery management device (1), or store data necessary for operating the battery management device (1).

[0072] Specifically, the memory (120) can store various programs related to calculating characteristic values ​​of battery cells, classifying classes, and estimating lifespan. In addition, the memory (120) can store various data such as voltage, current, and characteristic value data of each battery cell.

[0073] Additionally, the memory (120) can store the SOC and SOH of the battery cell (13) estimated by the processor (110) and can store a learning model for machine learning.

[0074] The memory (120) may include volatile memory (120) such as Static Random Access Memory (S-RAM) and Dynamic Random Access Memory (D-RAM) for temporarily storing data. In addition, the memory (120) may include nonvolatile memory (120) such as Read Only Memory (ROM), Erasable Programmable Read Only Memory (EPROM), and Electrically Erasable Programmable Read Only Memory (EEPROM) for long-term storage of data.

[0075] The processor (110) outputs control signals to control the battery management device (1) as a whole. The processor (110) may include one or more central processing units (CPUs) and graphics processing units (GPUs). In this case, the processor (110) may be implemented as an array of a plurality of logic gates, or may be implemented as a combination of a general-purpose microprocessor (110) and a memory (120) storing a program that can be executed on the microprocessor (110).

[0076] The aforementioned memory (120) and processor (110) may be included in the control unit (100), and the control unit (100) may control the aforementioned components to determine whether a defect has occurred in the battery cell.

[0077] Specifically, the control unit (100) can receive a voltage value from the voltage sensor (2) and a current value from the current sensor, thereby deriving an average voltage value of the battery cell (13). That is, the control unit (100) can derive an average voltage value for each of multiple sections included in the battery data.

[0078] At this time, the multiple sections may mean sections in which charging and discharging are performed while maintaining different current values ​​in the charge / discharge profile. The control unit (100) may derive an average voltage value for each of the multiple sections in which the current value is maintained within a certain range, and the control unit (100) may input some of the average voltage values ​​for each of the multiple sections as input values ​​to a learning model based on self-supervised learning.

[0079] Some of the voltage average values ​​used by the control unit (100) as input values ​​of the learning model may refer to one section arbitrarily set among multiple sections, and multiple sections may be used as input values.

[0080] The control unit (100) can derive voltage prediction values ​​as output values ​​using a learning model, and at this time, the learning model can include a linear regression model. That is, the control unit (100) can derive multiple voltage prediction values ​​as output values ​​by utilizing different linear regression models for each of multiple sections, and can detect whether the battery is defective by comparing the multiple voltage prediction values ​​with actual values.

[0081] Specifically, the control unit (100) can determine the difference between the voltage prediction values, which are multiple output values, and the actual measured voltage actual values ​​as an error. In this case, the actual voltage value can mean a voltage average value that is not used as an input value among the voltage average values ​​of multiple sections. In other words, the control unit (100) can use some of the voltage average values ​​for multiple sections as input values ​​and the remaining some as actual values ​​for determining the error with the output values.

[0082] In addition, the control unit (100) can determine a diagnostic index of the battery cell (13) by adding up all errors for multiple sections. Accordingly, the control unit (100) can derive a diagnostic index for all battery cells (13) included in one battery pack, and the control unit (100) can determine whether the battery cell (13) or the battery pack is defective.

[0083] In this way, the battery management device (1) according to one embodiment can diagnose whether a defect has occurred in a battery cell (13) using a learning model based on self-supervised learning, and thus has a remarkable effect of improving the reliability of diagnosis even in a situation where there is insufficient defect data.

[0084] FIG. 3 illustrates input and output values ​​of a machine learning model utilized in a battery management device according to one embodiment.

[0085] Referring to FIG. 3, the control unit (100) can train an ANN (Artificial Neural Network) model (b) using voltage average data for multiple sections as training data (a), and can derive a voltage prediction value as output data (c).

[0086] Here, the voltage average data can be derived by the control unit (100) receiving the voltage of each cell through the sensor unit (14) and calculating the average of the voltage data of each cell received. In addition, as described above, the voltage average data utilized as an input value can refer to some data among the voltage average data of multiple sections.

[0087] When the control unit (100) includes an artificial intelligence processor (110) (e.g., NPU) for training an ANN (Artificial Neural Network) model (b), the processor (110) can train the artificial neural network by utilizing the weight data stored in the memory (120) as training data for the machine learning model.

[0088] Examples of learning algorithms may include supervised learning, unsupervised learning, semi-supervised learning, or reinforcement learning, but the battery management device (1) according to one embodiment may determine whether a battery is defective based on self-supervised learning.

[0089] An artificial neural network included in a machine learning model may be composed of multiple neural network layers. Each of the multiple neural network layers has multiple weight values ​​and performs neural network operations through operations between the calculation results of the previous layer and the multiple weights. The multiple weights of the multiple neural network layers can be optimized based on the learning results of an artificial intelligence model. For example, during the learning process, the multiple weights may be updated so that the loss or cost values ​​obtained from the artificial intelligence model are reduced or minimized.

[0090] The artificial neural network may include a deep neural network (DNN), for example, a convolutional neural network (CNN), a deep neural network (DNN), a recurrent neural network (RNN), a restricted boltzmann machine (RBM), a deep belief network (DBN), a bidirectional recurrent deep neural network (BRDNN), or deep Q-networks, but is not limited to the examples described above.

[0091] The control unit (100) can learn the correlation between the voltage average data, which is training data (a), and the voltage prediction value, which is output data (c), based on the selected artificial intelligence model.

[0092] FIG. 4 illustrates a schematic flowchart of a battery management device according to one embodiment for detecting a defective battery. Configurations 101 to 104 in FIG. 4 are implemented in the form of software blocks, stored in memory (120), and executed by a processor (110).

[0093] Referring to FIG. 4, the control unit (100) can obtain voltage data included in battery data from the voltage sensor (2), and the control unit (100) can store the received voltage data of each battery cell (13) in the voltage data DB (121).

[0094] Thereafter, the voltage average data generation unit (101) of the control unit (100) can generate voltage average data based on the voltage data stored in the voltage data DB (121). The method by which the control unit (100) generates voltage average data of the battery cells (13) may generate voltage average data for each of all battery cells (13) included in one battery pack, or may generate voltage average data in parallel for all battery packs.

[0095] At this time, the voltage average data generated by the control unit (100) may include not only the voltage average value during the charging process but also the voltage average value during the discharging process. For convenience of explanation, the voltage average value during the charging process is described below.

[0096] The machine learning model learning unit (102) of the control unit (100) can learn a model based on preset hyperparameters and selected features. Specifically, the control unit (100) can learn model parameters by utilizing the selected features to predict the voltage average.

[0097] A battery management device (1) according to one embodiment can derive voltage prediction value data by applying a linear mixed model to voltage average data.

[0098] Here, the linear mixed model may mean a model that performs linear regression for each battery pack, and multiple linear regression models may be applied to each battery cell (13) included in one battery pack to derive voltage prediction value data.

[0099] The control unit (100) can determine the voltage prediction value data obtained through the above process as the target output value of the machine learning model, and the control unit (100) can train the machine learning model by changing the hyper parameters of the machine learning learning model or adding data, and the control unit (100) can store the trained machine learning model in the machine learning learning model DB (122).

[0100] Thereafter, the voltage prediction value derivation unit (103) of the control unit (100) can determine the output value of the learned machine learning model as the voltage prediction value. Specifically, the control unit (100) can derive the voltage prediction value and use it as data for comparison with the actual value.

[0101] Here, the voltage prediction value may mean multiple voltage prediction values ​​derived by using some of the multiple intervals in which the voltage is maintained within a certain range as input values ​​of the learning model.

[0102] The defective battery detection unit (104) of the control unit (100) can compare the actual value of each section corresponding to the voltage prediction value with the voltage prediction value to calculate an error. Specifically, the control unit (100) can determine the difference between the voltage prediction value derived as an output value and the actual value as an error, and can derive a battery safety diagnosis index by adding up the error values ​​generated from one battery cell (13).

[0103] The control unit (100) can determine a battery cell (13) whose derived diagnostic indicator exceeds a preset diagnostic criterion as defective. In this case, the diagnostic criterion may refer to a value pre-stored in the memory (120) according to the physicochemical characteristics of the battery.

[0104] Thereafter, the control unit (100) can transmit information on the battery cell (13) determined to be a defective battery to an external device (4), so that the user or manager of the battery can easily check the information on the defective battery remotely.

[0105] FIG. 5 illustrates an average voltage value derived by a battery management device according to one embodiment.

[0106] Figure 5 is a graph of current and voltage showing the process of battery charging over time, and the x-axis represents time.

[0107] That is, during the process of charging the battery, there can be five sections from x0 to x4 in which the battery is charged with different current values. The different current values ​​in different ranges may mean a section where the current does not change suddenly, such as in x0, but is maintained gently between 20 and 25, or a section where the current does not change suddenly, such as in x1, but is maintained gently around 60, or a section where the current does not change suddenly, such as in x2, but is maintained gently around 200, or a section where the current does not change suddenly, such as in x3, but is maintained gently around 110, or a section where the current does not change suddenly, such as in x4, but is maintained gently around 150.

[0108] In this way, multiple sections charged with different current values ​​can be derived differently depending on the state of the battery and the control strategy of the charger, and the control unit (100) can also determine the multiple sections by differentiating the graph slope.

[0109] The control unit (100) can derive the average value of sections charged with different current values ​​during the charging or discharging process. For example, the control unit (100) can determine the voltage average value of section x0 among multiple sections as 3.55, the voltage average value of section x1 as 3.62, the voltage average value of section x2 as 3.95, the voltage average value of section x3 as 3.73, and the voltage average value of section x4 as 3.83.

[0110] Through this, the control unit (100) can derive a voltage prediction value by using the average voltage value as an input value of the learning model, and can derive whether the battery is defective through the difference between the voltage prediction value and the actual value.

[0111] FIG. 6 illustrates the correlation between normal battery cells and defective battery cells derived by a battery management device according to one embodiment.

[0112] Referring to FIG. 6, both the x-axis and the y-axis may mean the voltage average value derived by the control unit (100). Specifically, the points expressed in FIG. 6 each mean a battery cell (13), and (a) may mean that the x-axis is the voltage average value of the x1 section and the y-axis is the voltage average value of the x0 section. In addition, (a-1) may mean that the x-axis is the voltage average value of the x2 section and the y-axis is the voltage average value of the x0 section, (a-2) may mean that the x-axis is the voltage average value of the x3 section and the y-axis is the voltage average value of the x0 section, and (a-3) may mean that the x-axis is the voltage average value of the x4 section and the y-axis is the voltage average value of the x0 section.

[0113] At this time, x0 and x1 to x4 generally show a linear correlation, and cases (a) to (a-3) correspond to a linear correlation and can be judged to be included in a cluster of normal cells.

[0114] On the other hand, in the case of (b) to (b-3), an outlier that deviates from the linear correlation may mean an individual data point that deviates significantly from the general pattern of the voltage average data set or has an exceptionally high or low value.

[0115] A battery management device (1) according to one embodiment can determine whether a battery is defective by deriving a diagnostic indicator using a machine learning technique based on the characteristics of a defective battery cell (13) that shows a different behavior from a normal battery cell (13) in the voltage average value data.

[0116] FIG. 7 is a graph illustrating diagnostic indicators derived by a battery management device according to one embodiment.

[0117] Referring to FIG. 7, the control unit (100) can derive a safety diagnostic index to determine whether a battery is defective. Here, the diagnostic index may refer to a sum of errors for multiple sections derived based on a learning model.

[0118] That is, the control unit (100) can determine the difference between the voltage average value (actual value) for multiple sections in which the voltage is maintained constant in the charge / discharge profile of the battery cell (13) and the voltage predicted value for multiple sections derived based on the learning model as an error for each multiple section.

[0119] For example, the control unit (100) can use the average value of five sections in which the voltage is maintained constant for a specific battery cell (13) as the actual value, and the control unit (100) can use one section among the five sections as an input value and determine four errors as the difference between the predicted value and the actual value for the remaining four sections.

[0120] Thereafter, the control unit (100) can add all four determined errors to determine a battery safety diagnostic index for a specific battery cell (13).

[0121] In this way, the diagnostic indicators determined by the control unit (100) can be predefined by the battery designer so that their absolute values ​​have a specific meaning, or a battery cell (13) having an outlier can be judged as defective based on the relative difference between each diagnostic indicator. Below, an embodiment of judging a battery's defect based on the relative difference between diagnostic indicators will be described.

[0122] The control unit (100) can visualize the diagnostic indicators by using the diagnostic indicator (Final Indicator, FI) as the y-axis and the battery pack number as the x-axis, as shown in Fig. 7. Accordingly, the control unit (100) can display points corresponding to the number of battery cells (13) included in each pack, and can derive the relative difference between the diagnostic indicators based on the characteristic that normal battery cells (13) occupy a larger number than abnormal battery cells (13).

[0123] For example, with battery cell (13) (a) and battery cell (13) (b), the control unit (100) can determine battery cell (13) (a) as a defective battery cell (13) because battery cell (13) (b) is located in a diagnostic indicator area between 0 and 2 where many are densely packed, and battery cell (13) (a) is located in an area between 4 and 6, which is somewhat outside the dense diagnostic indicator area.

[0124] In addition, the control unit (100) can determine a battery cell (13) as a defective battery cell (13) even if the diagnostic indicator of a specific battery cell (13) exceeds a preset reference value. For example, assuming that the diagnostic reference value stored in the memory (120) or input by the user is 4, the control unit (100) can determine the battery cell (13) (a) as a defective battery cell (13) because it exceeds 4, and can determine the battery cell (13) (b) as a normal battery cell (13) because it is less than 4.

[0125] Fig. 8 illustrates voltage prediction values ​​and actual values ​​derived from a single battery pack by a battery management device according to one embodiment. That is, Fig. 8 illustrates a process in which a control unit (100) derives errors for multiple battery cells (13) included in a single battery pack.

[0126] The control unit (100) can derive the values ​​of the battery cells (13) included in the battery pack as shown in Fig. 8 by using the x-axis as the voltage average value of the x0 section and the y-axis as the voltage average value of the x1 section.

[0127] In addition, the control unit (100) can derive a voltage prediction value (b) by inputting the voltage average value for a portion of the section into a learning model based on self-supervised learning. Thereafter, the control unit (100) can determine the difference between the voltage average value of each battery cell (13) and the voltage prediction value (b) as an error (c), as in battery cell (13) (a).

[0128] In Fig. 8, since the error (c) between the battery cell (13) (a) and the voltage prediction value (b) has a relatively large value compared to other battery cells (13), the battery cell (13) (a) can be determined as a defective battery cell (13) by judging it as an outlier.

[0129] FIG. 9 illustrates an error derived from one battery pack by a battery management device according to one embodiment.

[0130] Next, referring to FIG. 9, the control unit (100) can diagram whether or not a battery cell (13) is an outlier by using the number of each battery cell (13) as the x-axis and the error value as the y-axis.

[0131] According to this, the control unit (100) can determine that the battery cell (13) numbered 80 has a larger error value than other battery cells (13), and thus, the battery cell (13) numbered 80 can be determined as a defective battery cell (13).

[0132] For example, with battery cell (13) (a) and battery cell (13) (b), the control unit (100) can determine battery cell (13) (a) as a defective battery cell (13) because battery cell (13) (b) is located in an error area where a large number of errors are concentrated between 0 and 0.0015, and battery cell (13) (a) is located in an area of ​​0.0035 or more outside the concentrated error area.

[0133] In addition, the control unit (100) can determine a battery cell (13) as a defective battery cell (13) even if the error of a specific battery cell (13) exceeds a preset reference value. For example, assuming that the error reference value stored in the memory (120) or input by the user is 0.0020, the control unit (100) can determine the battery cell (13) (a) as a defective battery cell (13) because it exceeds 0.0020, and can determine the battery cell (13) (b) as a normal battery cell (13) because it is less than 0.0020.

[0134] In this way, the battery management device (1) according to one embodiment can determine whether a specific battery cell (13) is defective by directly using errors as well as diagnostic indicators, so it has the effect of enabling more diverse and precise determination of whether or not a battery cell is defective.

[0135] FIG. 10 illustrates a control flowchart of a battery management method according to one embodiment.

[0136] Referring to FIG. 10, the control unit (100) can receive battery data of multiple battery cells (13) via the communication unit (200) (1000). Thereafter, the control unit (100) can derive voltage average values ​​for multiple sections, and the multiple sections may refer to sections in which the voltage is gently maintained within a certain range during the charging and discharging process.

[0137] The control unit (100) can input some of the voltage average values ​​for multiple sections into the learning model (1020) and output voltage prediction values ​​for multiple sections (1030).

[0138] Thereafter, the control unit (100) can determine the difference between the voltage prediction value and the actual voltage value for multiple sections as an error, and derive a diagnostic index by summing up the errors for each section (1040).

[0139] The control unit (100) can determine whether the diagnostic indicator exceeds the diagnostic criteria (1050), and if the diagnostic indicator exceeds the preset diagnostic criteria (yes in 1050), the battery status can be determined as faulty (1060). Furthermore, if the diagnostic indicator is below the preset diagnostic criteria (no in 1050), the battery status can be determined as normal (1070).

[0140] At this time, the diagnostic criterion may mean an average value of diagnostic indicators of multiple battery cells (13), or may mean a reference value with an additional offset applied to the average value.

[0141] In this way, the battery management device (1) according to one embodiment can define battery safety diagnosis indicators and derive diagnosis indicators based on machine learning, thereby having the effect of more accurately diagnosing the cause of battery failure.

[0142] Meanwhile, the disclosed embodiments may be implemented in the form of a recording medium storing computer-executable instructions. The instructions may be stored in the form of program code, and when executed by a processor, may generate program modules to perform the operations of the disclosed embodiments. The recording medium may be implemented as a computer-readable recording medium.

[0143] Computer-readable storage media include all types of storage media that store instructions that can be deciphered by a computer. Examples include read-only memory (ROM), random access memory (RAM), magnetic tape, magnetic disks, flash memory, and optical data storage devices.

[0144] Additionally, a computer-readable recording medium may be provided in the form of a non-transitory storage medium. Here, the term "non-transitory storage medium" simply means a tangible device that does not contain signals (e.g., electromagnetic waves). This term does not distinguish between cases where data is permanently stored in the storage medium and cases where data is temporarily stored. For example, a "non-transitory storage medium" may include a buffer in which data is temporarily stored.

[0145] According to one embodiment, the method according to various embodiments disclosed in the present document may be provided as included in a computer program product. The computer program product may be traded as a product between a seller and a buyer. The computer program product may be distributed in the form of a machine-readable recording medium (e.g., compact disc read only memory (CD-ROM)), or may be distributed online (e.g., downloaded or uploaded) through an application store (e.g., Play Store™) or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product (e.g., a downloadable app) may be temporarily stored or temporarily generated on a machine-readable recording medium, such as a memo on a manufacturer's server, an application store's server, or an intermediary server.

[0146] Although all components constituting the embodiments disclosed in this document have been described as being combined or operating in combination as one, the embodiments disclosed in this document are not necessarily limited to such embodiments. That is, within the scope of the purpose of the embodiments disclosed in this document, all of the components may be selectively combined and operated one or more times.

[0147] Furthermore, terms such as "include," "comprise," or "have" described above, unless specifically stated otherwise, imply that the corresponding component may be present, and therefore should be interpreted to include other components rather than excluding 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 the contextual meaning of the relevant technology, and shall not be interpreted in an idealized or overly formal sense, unless explicitly defined in this document.

[0148] The above description is merely an illustrative description 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.

[0149]

[0150] [Explanation of symbols]

[0151] 1: Battery management device

[0152] 2: Voltage sensor

[0153] 3: Current sensor

[0154] 4: External devices

[0155] 10: Battery system

[0156] 12: Multiple battery modules

[0157] 13: Battery cell

[0158] 14: Sensor section

[0159] 16: Switching section

[0160] 20: Upper controller

[0161] 100: Control Unit

[0162] 110: Processor

[0163] 120: Memory

[0164] 200: Communications Department

[0165] 210: Wireless Communications Department

[0166] 220: Wired Communications Department

Claims

1. A communication unit for receiving battery data of a battery cell; and A battery management device including a control unit which derives voltage average values ​​for multiple sections included in the battery data, inputs some of the voltage average values ​​for the multiple sections into a learning model based on self-supervised learning, and detects whether the battery cell is defective using a voltage prediction value output by the learning model.

2. In claim 1, The above control unit, A battery management device that determines the average value of the voltage as the average value of each section in which the voltage is maintained constant in the charge / discharge profile of the battery cell.

3. In claim 1, The above learning model is, A battery management device comprising a plurality of linear regression models, each of which comprises a number of voltage averages excluding some of the above voltage averages.

4. In claim 3, The above control unit, A battery management device that inputs some of the voltage average values ​​into each of the plurality of linear regression models to derive multiple output values.

5. In claim 4, The above control unit, A battery management device that determines the difference between the voltage prediction value, which is the plurality of output values, and the actual value, which is a part of the voltage average value, as an error for each of the plurality of sections.

6. In claim 5, The above control unit, A battery management device that determines a diagnostic index of the battery cell by adding the errors for the multiple sections.

7. In claim 6, The above control unit, A battery management device that determines the battery cell to be defective based on the above diagnostic indicator exceeding a preset diagnostic criterion.

8. In claim 1, The above self-supervised learning-based learning model is, A battery management device for detecting whether a battery cell is defective, including a linear mixed model (LMM).

9. Receive battery data from the battery cell; Each voltage average value for multiple sections included in the above battery data is derived; Some of the voltage average values ​​for the above multiple sections are input into a learning model based on self-supervised learning; A battery management method, comprising: detecting whether the battery cell is defective using a voltage prediction value output by the learning model.

10. In claim 9, Deriving the average value of the above voltage is: A battery management method for deriving an average value of the voltage as an average value of each section in which the voltage is maintained constant in the charge / discharge profile of the battery cell.

11. In claim 9, The above learning model is, A battery management method comprising a plurality of linear regression models formed by excluding some of the above voltage average values.

12. In claim 11, A battery management method further comprising: inputting some of the voltage average values ​​into each of the plurality of linear regression models to derive a plurality of output values.

13. In claim 12, A battery management method further comprising: determining the difference between the voltage prediction value, which is the plurality of output values, and the actual value, which is a part of the voltage average value, as an error for each of the plurality of sections.

14. In claim 13, A battery management method further comprising: determining a diagnostic indicator of the battery cell by adding the errors for the plurality of sections.

15. In claim 14, Detecting whether the above battery cell is defective is as follows: A battery management method, comprising: determining the battery cell as defective based on the diagnostic indicator exceeding a preset diagnostic criterion.

16. In claim 9, The above self-supervised learning-based learning model is, A battery management method for detecting whether a battery cell is defective, including a linear mixed model (LMM).

Images