Battery management devices and methods
By extracting feature values and setting labels during the charging and discharging cycles of the battery, the problem of obtaining learning data from a limited number of batteries is solved, thereby improving the performance of the classification model and the accuracy of battery state judgment.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- LG CHEM LTD
- Filing Date
- 2021-11-26
- Publication Date
- 2026-07-17
AI Technical Summary
Existing technologies struggle to quickly obtain large amounts of learning data for classifying models from a limited number of batteries, thus limiting the improvement of model performance.
By extracting battery feature values at each charge and discharge cycle, determining battery state based on these feature values and setting labels, and learning a classification model to analyze battery state, the process includes feature extraction, state determination, labeling, and model learning.
This enables the rapid acquisition of large amounts of learning data from a limited number of batteries, improving the performance of the learning classification model and the accuracy of battery state analysis.
Smart Images
Figure CN115667960B_ABST
Abstract
Description
Technical Field
[0001] This application claims priority to Korean Patent Application No. 10-2020-0161667, filed in Korea on November 26, 2020, the disclosure of which is incorporated herein by reference.
[0002] This disclosure relates to battery management devices and methods, and more specifically to battery management devices and methods capable of using learned classification models to determine battery status. Background Technology
[0003] Recently, demand for portable electronic products such as laptops, video cameras, and mobile phones has increased dramatically, and electric vehicles, energy storage batteries, robots, and satellites are also under serious development. Therefore, high-performance batteries that allow for repeated charging and discharging are being actively researched.
[0004] Currently available batteries include nickel-cadmium batteries, nickel-metal hydride batteries, nickel-zinc batteries, and lithium-ion batteries. Among them, lithium-ion batteries have attracted much attention because they have almost no memory effect compared to nickel-based batteries, and they also have a very low self-discharge rate and high energy density.
[0005] Typically, machine learning-based or deep learning-based techniques are introduced to estimate or determine the state of a battery. For example, techniques are being developed to use learning-based models to determine the state of a battery.
[0006] Learning such a model requires a large amount of training data. Obtaining training data from batteries necessitates battery charge-discharge cycles, which can be time-consuming. Furthermore, as the expected lifespan of a battery increases, the time required for charge-discharge cycles also increases proportionally, thus increasing the time required to obtain training data.
[0007] Furthermore, the amount of learning data obtainable from a limited number of batteries is finite. Therefore, even with a limited number of batteries, it is necessary to develop techniques that can improve the performance of learning models by acquiring large amounts of learning data. Summary of the Invention
[0008] Technical issues
[0009] This disclosure is designed to address problems in the related art, and therefore, this disclosure aims to provide a battery management device and method that can obtain a large amount of learning data from a limited number of batteries for learning a classification model.
[0010] These and other objects and advantages of this disclosure will become apparent from the following detailed description and will become more fully apparent from exemplary embodiments of this disclosure. Furthermore, it will be readily understood that the objects and advantages of this disclosure can be achieved by the means shown in the appended claims and combinations thereof.
[0011] Technical solutions
[0012] A battery management device according to one aspect of this disclosure may include: a feature extraction unit configured to extract feature values of a learning battery at each charge and discharge cycle; a first state determination unit configured to determine the state of the learning battery at each charge and discharge cycle based on the feature values extracted by the feature extraction unit and preset standard values corresponding to the feature values; a labeling unit configured to set labels on the feature values of the learning battery at each charge and discharge cycle based on the state of the learning battery determined by the first state determination unit; a model learning unit configured to learn a classification model for determining and analyzing the state of the battery based on the feature values labeled by the labeling unit; and a second state determination unit configured to determine and analyze the state of the battery based on the classification model learned by the model learning unit.
[0013] The first state determination unit can be configured to determine the state of the learned battery at each charge and discharge cycle as either a normal state or a defective state.
[0014] When the state of the learning battery is determined to be normal, the marking unit can be configured to set a first mark on the feature value of the learning battery for the corresponding charge and discharge cycle.
[0015] When the state of the learning battery is determined to be a defective state, the marking unit can be configured to set a second mark on the feature value of the learning battery for the corresponding charge and discharge cycle.
[0016] The tagging unit can be configured to tag the characteristic values of the learning battery based on the state of the learning battery determined at predetermined charge and discharge cycles.
[0017] When the state of the learning battery is determined to be normal at a predetermined charge and discharge cycle, the marking unit can be configured to set a first mark on the characteristic value of the learning battery for the charge and discharge cycles up to the cycle immediately preceding the predetermined charge and discharge cycle.
[0018] When the state of the learning battery at a predetermined charge and discharge cycle is determined to be a defective state, the marking unit can be configured to set a second mark on the characteristic value of the learning battery for the charge and discharge cycles following the predetermined charge and discharge cycles.
[0019] When the state of the learning battery is determined to be defective, the marking unit can be configured to set a second mark on the feature value of the learning battery for subsequent charge and discharge cycles.
[0020] The marking unit can be configured to count the number of times the state of the learning battery is judged as a defective state, and when the counted number is equal to or greater than the standard number, the marking unit can be configured to set a second mark on the characteristic value of the learning battery for the following charge and discharge cycles: the charge and discharge cycles after the last charge and discharge cycle in which the state of the learning battery was judged as a defective state.
[0021] The feature extraction unit can be configured to extract multiple feature values of the learning cell.
[0022] The first state determination unit can be configured to determine the state of the learning battery for each of the multiple feature values at each charge and discharge cycle, based on multiple feature values of the learning battery and a preset standard value corresponding to each of the multiple feature values.
[0023] The marking unit can be configured to set a mark for each of a plurality of characteristic values at each charge and discharge cycle.
[0024] The second state determination unit can be configured to extract feature values of the analyzed battery at each charge and discharge cycle, and determine the state of the analyzed battery by inputting the extracted feature values into the learned classification model.
[0025] A battery pack according to another aspect of this disclosure may include a battery management device according to one aspect of this disclosure.
[0026] A battery management method according to another aspect of this disclosure may include: a feature extraction step, in which feature values of a learning battery are extracted at each charge and discharge cycle; a first state determination step, in which the state of the learning battery at each charge and discharge cycle is determined based on the feature values extracted in the feature extraction step and preset standard values corresponding to the feature values; a labeling step, in which a label is set on the feature values of the learning battery at each charge and discharge cycle based on the state of the learning battery determined in the first state determination step; a model learning step, in which a classification model for determining the state of the battery is learned based on the feature values for which labels are set in the labeling step; and a second state determination step, in which the state of the battery is determined based on the classification model learned in the model learning step.
[0027] Beneficial effects
[0028] According to one aspect of this disclosure, an advantage is that a large amount of learning data for learning classification models can be quickly obtained from a limited number of batteries.
[0029] The effects of this disclosure are not limited to those mentioned above, and other unmentioned effects will be clearly understood by those skilled in the art from the description of the claims. Attached Figure Description
[0030] The accompanying drawings illustrate preferred embodiments of the present disclosure and are used together with the foregoing disclosure to provide a further understanding of the technical features of the present disclosure; therefore, the present disclosure is not to be construed as limited to the drawings.
[0031] Figure 1 This is a schematic diagram illustrating a battery management device according to one aspect of this disclosure.
[0032] Figure 2 This is a schematic diagram illustrating a first embodiment in which a battery management device sets a mark according to one aspect of this disclosure.
[0033] Figure 3 This is a schematic diagram illustrating a second embodiment in which a battery management device sets a marker according to one aspect of this disclosure.
[0034] Figure 4 This is a schematic diagram illustrating a third embodiment of a battery management device with markings set according to one aspect of this disclosure.
[0035] Figure 5 This is a diagram schematically illustrating an exemplary configuration of a battery pack according to another aspect of this disclosure.
[0036] Figure 6 This is a diagram schematically illustrating a battery management method according to another aspect of this disclosure. Detailed Implementation
[0037] It should be understood that the terms used in the specification and appended claims should not be construed as limited to their general or dictionary meanings, but should be interpreted based on their meanings and concepts corresponding to the technical aspects of this disclosure, while allowing the inventors to appropriately limit the terms for the best explanation.
[0038] Therefore, the descriptions presented herein are merely preferred examples for illustrative purposes only and are not intended to limit the scope of this disclosure. It should be understood that other equivalents and modifications may be made to this disclosure without departing from its scope.
[0039] Additionally, in describing this disclosure, detailed descriptions of relevant known elements or functions are omitted herein if they are considered to obscure the key subject matter of the disclosure.
[0040] Terms including ordinal numbers such as "first" and "second" can be used to distinguish one element from another among various elements, but are not intended to restrict elements through these terms.
[0041] Throughout this specification, when a section is referred to as “containing” or “including” any element, it means that the section may also include other elements, without excluding other elements, unless otherwise specifically stated.
[0042] Furthermore, throughout the specification, when a part is referred to as "connected" to another part, it is not limited to the case where they are "directly connected," but also includes the case where another element is inserted between them and they are "indirectly connected."
[0043] In the following, preferred embodiments of the present disclosure will be described in detail with reference to the accompanying drawings.
[0044] Figure 1 This is a schematic diagram illustrating a battery management device 100 according to one aspect of this disclosure.
[0045] Reference Figure 1 The battery management device 100 may include a feature extraction unit 110, a first state determination unit 120, a labeling unit 130, a model learning unit 140, and a second state determination unit 150.
[0046] The feature extraction unit 110 can be configured to extract feature values of the learning battery at each charge and discharge cycle.
[0047] Here, "battery" refers to a physically separable, independent unit comprising a negative terminal and a positive terminal. For example, a pouch-type lithium-ion battery cell can be considered a battery.
[0048] Additionally, a learning battery is a battery used to acquire a learning dataset for learning a classification model, and it can be distinguished from an analysis battery.
[0049] For example, the feature extraction unit 110 can extract at least one feature value of the learning battery at each charge and discharge cycle. Specifically, the feature extraction unit 110 can extract feature values based on the learning battery's voltage, current, capacity, charging time, internal resistance, SOC (state of charge), SOH (state of health), or a combination thereof.
[0050] The first state determination unit 120 can be configured to determine the state of the learned battery at each charging and discharging cycle based on the feature values extracted by the feature value extraction unit 110 and the preset standard values corresponding to the feature values.
[0051] Preferably, a standard value can be preset for each characteristic value for each charge and discharge cycle.
[0052] For example, suppose that the first to Nth feature values of the learning battery are extracted at each of the first to Nth charge and discharge cycles. Here, N can be an integer of 2 or greater. In this case, a total of N standard values can be preset. That is, the first to Nth standard values can be preset to correspond to the first to Nth feature values respectively. In addition, the first state determination unit 120 can determine the state of the learning battery at the first charge and discharge cycle by comparing the first feature value and the first standard value. Similarly, the first state determination unit 120 can determine the state of the learning battery at the Nth charge and discharge cycle by comparing the Nth feature value and the Nth standard value.
[0053] Specifically, the first state determination unit 120 can be configured to determine the state of the learned battery at each charge and discharge cycle as either a normal state or a defective state.
[0054] For example, the standard value can be a threshold used to determine the state of the learning battery. The first state determination unit 120 can determine whether the state of the learning battery is normal or defective by comparing the magnitude of the standard value and the feature value.
[0055] The marking unit 130 can be configured to set a mark for the feature value of the learning battery at each charging and discharging cycle based on the state of the learning battery determined by the first state judgment unit 120.
[0056] Specifically, when the state of the learning battery is determined to be normal, the marking unit 130 can be configured to set a first mark on the feature value of the learning battery at the corresponding charging and discharging cycle.
[0057] Conversely, when the state of the learning battery is determined to be a defective state, the marking unit 130 can be configured to set a second mark on the feature value of the learning battery at the corresponding charge and discharge cycle.
[0058] That is, the marking unit 130 can set a mark on the feature value of the learning battery extracted at each charging and discharging cycle according to the state of the learning battery determined by the first state judgment unit 120.
[0059] The model learning unit 140 can be configured to learn a classification model for judging and analyzing the state of the battery based on feature values that have been labeled by the labeling unit 130.
[0060] For example, logistic regression, linear discriminant analysis, support vector machines, random forest classifiers, or artificial neural networks can be used as classification models.
[0061] The classification model learned by the model learning unit 140 can be used to determine and analyze the state of the battery.
[0062] However, since the process of learning a classification model based on its labeled feature values is a general model learning process in machine learning, its detailed description will be omitted.
[0063] The second state determination unit 150 can be configured to determine and analyze the state of the battery based on a classification model learned by the model learning unit 140.
[0064] Specifically, the second state determination unit 150 can be configured to extract and analyze the characteristic values of the battery at each charge and discharge cycle.
[0065] Here, the feature values of the analyzed battery can be of the same type as the feature values of the learned battery extracted by the feature value extraction unit 110. That is, the feature values of the analyzed battery extracted by the second state judgment unit 150 can be of the same type as the feature values used in the process of learning the classification model by the model learning unit 140.
[0066] For example, suppose the feature extraction unit 110 extracts the voltage of the learning battery as a feature value at each charge and discharge cycle. The first state determination unit 120 can determine the state of the learning battery at each charge and discharge cycle based on the voltage of the learning battery and a standard value. The labeling unit 130 can set a label for the voltage of each charge and discharge cycle based on the determined state of the learning battery at each charge and discharge cycle. The model learning unit 140 can learn a classification model based on the voltage to which the label is set. The second state determination unit 150 can extract the voltage of the analysis battery as a feature value at each charge and discharge cycle.
[0067] In addition, the second state determination unit 150 can be configured to determine the state of the battery by inputting the extracted feature values into the learned classification model.
[0068] For example, the second state determination unit 150 can set the feature values of the analyzed battery as input to the learned classification model. Furthermore, the learned classification model can output the state of the analyzed battery as output for the input. The second state determination unit 150 can acquire the output from the learned classification model and determine whether the state of the analyzed battery is normal or defective.
[0069] The battery management device 100 according to the embodiments of this disclosure has the advantage of obtaining a large amount of learning data from a limited number of batteries for learning a classification model.
[0070] That is, because the battery deteriorates as charging and discharging cycles proceed, the state of the battery may be different at each charging and discharging cycle. Taking these characteristics of the battery into account, the battery management device 100 can obtain a large amount of learning data for learning a classification model by extracting the feature values of the battery at each charging and discharging cycle.
[0071] Therefore, even with a limited number of learning batteries, a large amount of learning data can be obtained for training a classification model, and the performance of the learned classification model can be improved based on this large amount of learning data. This can thus improve the accuracy of the learning-based classification model in determining the state of the battery.
[0072] At the same time, refer to Figure 1 The battery management device 100 may also include a storage unit 160. The storage unit 160 may store data necessary for the operation and function of each component of the battery management device 100, data generated during the execution of operations or functions, etc. There are no particular limitations on the type of storage unit 160, as long as it is a known information storage device capable of recording, erasing, updating, and retrieving data. As examples, the information storage device may include RAM, flash memory, ROM, EEPROM, registers, etc. Additionally, the storage unit 160 may store program code defining processes executable by a control unit.
[0073] For example, the battery feature values extracted by the feature value extraction unit 110 at each charge and discharge cycle can be stored in the storage unit 160. Additionally, the tags set by the tagging unit 130 for each feature value can also be stored in the storage unit 160.
[0074] In addition, the feature extraction unit 110 can be configured to extract multiple feature values of the learning battery.
[0075] Specifically, the feature extraction unit 110 can extract two or more feature values for each learning battery at each charge and discharge cycle.
[0076] For example, the feature extraction unit 110 can extract the voltage and internal resistance of each learning cell at each charge and discharge cycle. If up to 500 charge and discharge cycles are performed on 10 learning cells, the total number of feature values extracted by the feature extraction unit 110 can be 10,000. That is, 5,000 voltage values and 5,000 internal resistance values can be extracted.
[0077] The first state determination unit 120 can be configured to determine the state of the learning battery for each of the multiple feature values at each charging and discharging cycle based on multiple feature values of the learning battery and a preset standard value corresponding to each of the multiple feature values.
[0078] For example, in the aforementioned embodiment, when the feature extraction unit 110 extracts a total of 10,000 feature values, the first state determination unit 120 can determine the state of the learning battery for voltage and internal resistance at each charge and discharge cycle by comparing each of the 10,000 feature values with a standard value corresponding to that feature value. That is, based on the voltage of the 10 learning batteries, the voltage-based state of the learning battery in each of the 1st to 500th cycles can be determined. Furthermore, based on the internal resistance of the 10 learning batteries, the internal resistance-based state of the learning battery in each of the 1st to 500th cycles can be determined.
[0079] The marking unit 130 can be configured to set a mark for each of a plurality of characteristic values at each charge and discharge cycle.
[0080] For example, in the aforementioned embodiment, when the first state determination unit 120 determines all states of the learning battery for 10,000 feature values, the marking unit 130 can set a mark for each of the 10,000 feature values.
[0081] That is, the battery management device 100 according to the embodiments of the present disclosure has the advantage of obtaining a large amount of learning data from a limited number of learning batteries by extracting multiple feature values of the learning batteries.
[0082] In the following text, various embodiments in which the marking unit 130 sets a mark for the feature value will be described.
[0083] Figure 2 This is a schematic diagram illustrating a first embodiment in which a battery management device 100 is marked according to one aspect of this disclosure.
[0084] The marking unit 130 can be configured to mark the feature values of the learning battery based on the state of the learning battery determined at a predetermined charge and discharge cycle.
[0085] For example, in Figure 2 In this implementation, the marking unit 130 can set a mark for the feature value of each of the first to fifth learning cells B1, B2, B3, B4, B5 based on the state of each of the first to fifth learning cells B1, B2, B3, B4, B5 determined at the target cycle TC.
[0086] Here, the target cycle TC can be the last charge and discharge cycle in which the battery state is determined by the first state determination unit 120. For example, the target cycle TC can be a cycle related to the battery's quality assurance. As a specific example, if the battery's quality assurance is set to retain 90% or more of the battery capacity at the 500th cycle, then the target cycle TC can be set to the 500th cycle. In other words, the target cycle TC can be set individually based on the battery type, the battery's purpose, the environment in which the battery is used, etc.
[0087] Specifically, when the state of the learning battery is determined to be normal at a predetermined charge and discharge cycle, the marking unit 130 can set a first mark on the characteristic value of the learning battery for the charge and discharge cycles up to the cycle immediately preceding the predetermined charge and discharge cycle.
[0088] Conversely, if the state of the learning battery at the predetermined charge and discharge cycle is determined to be a defective state, the marking unit 130 can set a second mark on the characteristic value of the learning battery at the charge and discharge cycle after the predetermined charge and discharge cycle.
[0089] exist Figure 2 In this implementation, it is assumed that the states of the first to third learning batteries B1, B2, and B3 are determined to be normal at the target cycle TC, and the states of the fourth and fifth learning batteries B4 and B5 are determined to be defective. In this case, the states of the first to third learning batteries B1, B2, and B3 before the target cycle TC will be normal, and the states of the fourth and fifth learning batteries B4 and B5 after the target cycle TC will be defective.
[0090] The tagging unit 130 can set a first tag on the feature values of the first to third learning cells B1, B2, and B3 before the target cycle TC.
[0091] Furthermore, the marking unit 130 can set a second mark on the feature values of the fourth and fifth learning batteries B4 and B5 after the target cycle TC. That is, even if the state of the fourth and fifth learning batteries B4 and B5 is not determined at the charging and discharging cycles after the target cycle TC, the marking unit 130 can quickly set a mark on the feature values.
[0092] That is, the battery management device 100 can not only set a mark for the characteristic value of the learning battery immediately before the target cycle TC, but also set a mark for the characteristic value of the learning battery at the charging and discharging cycles after the target cycle TC.
[0093] Figure 3 This is a schematic diagram illustrating a second embodiment of a battery management device 100 with markings provided, according to one aspect of this disclosure.
[0094] The marking unit 130 can be configured to set a second mark on the feature value of the learning battery for subsequent charge and discharge cycles when the state of the learning battery is determined to be a defective state.
[0095] Specifically, after a charge and discharge cycle that is determined to be in a defective state, the marking unit 130 can set a second mark on the feature value of the learning battery. That is, even if the first state determination unit 120 does not determine the state of the learning battery after a charge and discharge cycle that is determined to be in a defective state, the marking unit 130 can still set a mark on the feature value.
[0096] For example, in Figure 3 In this implementation, the state of the sixth learning battery B6 at the target cycle TC can be determined as a normal state, and the state of the seventh learning battery B7 can be determined as a defective state. The state of the eighth learning battery B8 at the third charge and discharge cycle C3 can be determined as a defective state, and the state of the ninth learning battery B9 at the second charge and discharge cycle C2 can be determined as a defective state. The state of the tenth learning battery B10 at the first charge and discharge cycle C1 can be determined as a defective state.
[0097] The tagging unit 130 can set a first tag for the feature values of the sixth learning cell B6 up to the cycle immediately preceding the target cycle TC.
[0098] Furthermore, the marking unit 130 can set a first mark for the feature values of the seventh learning battery B7 up to the cycle immediately preceding the target cycle TC, and set a second mark for the feature values of the seventh learning battery B7 after the target cycle TC. That is, even if the first state determination unit 120 fails to determine the state of the seventh learning battery B7 after the target cycle TC, the marking unit 130 can still set a second mark for the feature values of the seventh learning battery B7.
[0099] Furthermore, the marking unit 130 can set a first mark for the feature values of the eighth learning battery B8 up to the cycle immediately preceding the third charge and discharge cycle C3, and set a second mark for the feature values of the eighth learning battery B8 after the third charge and discharge cycle C3. That is, even if the first state determination unit 120 fails to determine the state of the eighth learning battery B8 after the third charge and discharge cycle C3, the marking unit 130 can still set a second mark for the feature values of the eighth learning battery B8.
[0100] Furthermore, the marking unit 130 can set a first mark for the characteristic values of the ninth learning battery B9 up to the cycle immediately preceding the second charging and discharging cycle C2, and set a second mark for the characteristic values of the ninth learning battery B9 after the second charging and discharging cycle C2. That is, even if the first state determination unit 120 fails to determine the state of the ninth learning battery B9 after the second charging and discharging cycle C2, the marking unit 130 can still set a second mark for the characteristic values of the ninth learning battery B9.
[0101] Furthermore, the marking unit 130 can set a first mark for the feature values of the tenth learning battery B10 up to the cycle immediately preceding the first charge and discharge cycle C1, and set a second mark for the feature values of the tenth learning battery B10 after the first charge and discharge cycle C1. That is, even if the first state determination unit 120 fails to determine the state of the tenth learning battery B10 after the first charge and discharge cycle C1, the marking unit 130 can still set a second mark for the feature values of the tenth learning battery B10.
[0102] According to embodiments of this disclosure, the battery management device 100 can set a second label on feature values without determining the state of the learning battery after a charging and discharging cycle that has been determined to be in a defective state. Therefore, learning data for learning a classification model can be obtained more quickly, and system resources of the battery management device 100 can be saved during the acquisition of such learning data.
[0103] Figure 4 This is a schematic diagram illustrating a third embodiment of the present disclosure, in which a battery management device 100 is marked.
[0104] The marking unit 130 can be configured to count the number of times the state of the learning cell is judged as a defective state.
[0105] For example, in Figure 4 In this implementation, at the first charge and discharge cycle C1, the eleventh learning battery B11 can be determined to be in a defective state. The marking unit 130 can set the number of times the eleventh learning battery B11 is determined to be in a defective state at the first charge and discharge cycle C1 to 1.
[0106] Furthermore, at the second charge and discharge cycle C2, the eleventh learning battery B11 can be identified as a defective state. The marking unit 130 can set the number of times the eleventh learning battery B11 is identified as a defective state at the second charge and discharge cycle C2 to 2.
[0107] Furthermore, at the third charge and discharge cycle C3, the eleventh learning battery B11 can be identified as a defective state. The marking unit 130 can set the number of times the eleventh learning battery B11 is identified as a defective state at the third charge and discharge cycle C3 to 3.
[0108] When the number of counts is equal to or greater than the standard number of counts, the marking unit 130 can be configured to set a second mark on the characteristic value of the learning battery for the following charge and discharge cycles: the charge and discharge cycles after the last charge and discharge cycle in which the state of the learning battery is judged to be a defective state.
[0109] For example, in Figure 4 In this implementation, it is assumed that the standard number of times is preset to 3. Since the number of times that will be judged as the defective state of the eleventh learning battery B11 at the third charge and discharge cycle C3 is set to 3, the marking unit 130 can set a second mark for the feature value after the third charge and discharge cycle C3.
[0110] Typically, the state of a battery deteriorates with charge and discharge cycles, but in some cases, the battery state can temporarily recover. Furthermore, during the determination of battery state, measurement errors caused by internal and external environmental issues, and temporary noise, can have an impact. Therefore, even if the state of the learning battery is not determined, the marking unit 130 can only set a second mark when the state of the learning battery has been determined to be a defective state a standard number of times or more.
[0111] Simultaneously, predetermined conditions can be added to the standard number of cycles. For example, if the state of the learning battery is continuously judged as a defective state, and the number of times it is judged as a defective state is greater than or equal to the standard number of cycles, the marking unit 130 can be configured to set a second mark on the characteristic value of the learning battery for the following charge and discharge cycles: the charge and discharge cycles after the last charge and discharge cycle in which the state of the learning battery is judged as a defective state.
[0112] Taking into account the type of battery, its intended use, and the environment in which it is used, users can preset the standard number of cycles and add conditions to the standard number of cycles.
[0113] Meanwhile, the feature extraction unit 110 can be configured to generate additional feature values based on the feature values extracted at each charge and discharge cycle.
[0114] For example, suppose there are a total of 10 learning cells, and the feature extraction unit 110 has extracted feature values for each learning cell from the first charge and discharge cycle C1 to the cycle immediately preceding the 500th charge and discharge cycle. In this case, a total of 5000 feature values can be extracted.
[0115] The feature extraction unit 110 can generate additional feature values by combining the feature values extracted at each charge and discharge cycle to further obtain learning data for learning the classification model.
[0116] For example, the feature extraction unit 110 can generate multiple groups for multiple feature values extracted at the same charge and discharge cycles, and calculate an average feature value for each generated group.
[0117] In addition, the feature extraction unit 110 can be configured to generate groups of features of the same type.
[0118] For example, suppose that voltage and internal resistance are extracted as feature values by feature value extraction unit 110. In this case, a group for voltage can be generated based on the feature values for voltage, and a group for internal resistance can be generated based on the feature values for internal resistance. That is, a mixed group for voltage and internal resistance can be avoided.
[0119] In the aforementioned embodiment, the feature extraction unit 110 can generate groups for 2 to 10 feature values out of 10 feature values to be extracted at the same charge and discharge cycles, and can also generate an average feature value for each group. Here, the total number of groups generated based on the 10 feature values can be 1013.
[0120] Specifically, when the feature value extraction unit 110 generates groups by combining 2, 3, 4, 5, 6, 7, 8, 9, and 10 feature values from 10 feature values, the number of generated groups can be 45, 120, 210, 252, 210, 120, 45, 10, and 1, respectively. Therefore, the total number of generated groups can be 1013. The feature value extraction unit 110 can generate 1013 additional feature values at each charge and discharge cycle by calculating the average feature value of each of the 1013 generated groups.
[0121] The first state determination unit 120 can determine the state of the virtual learning battery based on each of the additional feature values generated by the feature extraction unit 110. Additionally, the labeling unit 130 can set a label for each additional feature value based on the determined state of the virtual learning battery. The model learning unit 140 can learn a classification model using both the feature values of the learning battery and the feature values of the virtual learning battery.
[0122] That is, the battery management device 110 has the advantage of obtaining a large amount of learning data based on a limited number of learning batteries by generating additional feature values of virtual learning batteries.
[0123] Simultaneously, the feature extraction unit 110 can be configured to further consider the state of the learning cell determined by the first state determination unit 120 during the group generation process. That is, the feature extraction unit 110 can be configured to generate groups by combining learning cells having the same state determined by the first state determination unit 120.
[0124] For example, at the Nth charge and discharge cycle, suppose that 6 out of 10 learning cells are classified as normal and 4 are classified as defective. The feature value extraction unit 110 can generate multiple groups by combining the feature values of the 6 learning cells classified as normal. In this case, a total of 57 groups can be generated. Alternatively, the feature value extraction unit 110 can generate multiple groups by combining the feature values of the 4 learning cells classified as defective. In this case, a total of 11 groups can be generated. Furthermore, the feature value extraction unit 110 can generate a total of 68 additional feature values by calculating the average feature value of each generated group.
[0125] The battery management device 100 according to this disclosure can be applied to a BMS (Battery Management System). That is, the BMS according to this disclosure may include the battery management device 100 described above. In this configuration, at least some components of the battery management device 100 can be implemented by supplementing or adding functionality to a configuration included in a conventional BMS. For example, the feature extraction unit 110, the first state determination unit 120, the tagging unit 130, the model learning unit 140, and the second state determination unit 150 can be implemented as components of the BMS.
[0126] Figure 5 This is a diagram schematically illustrating an exemplary configuration of battery pack 1 according to another aspect of this disclosure.
[0127] Reference Figure 5 The battery management device 100 according to this disclosure can be disposed in the battery pack 1. That is, the battery pack 1 according to this disclosure may include the battery management device 100 described above and one or more battery modules 10. Here, the battery module 10 may include one or more battery cells connected in series and / or in parallel. In addition, the battery pack 1 may also include electrical equipment (relays, fuses, etc.) and a housing.
[0128] For example, in Figure 5In this embodiment, the feature value extraction unit 110 can be connected to the battery module 10 and the current measurement unit. The feature value extraction unit 110 can measure the voltage, current, temperature, etc. of each battery cell included in the battery module 10 and extract them as feature values. In addition, the feature value extraction unit 110 can extract the internal resistance, SOC, SOH, etc. of each battery cell as feature values based on the measured values.
[0129] Figure 6 This is a diagram schematically illustrating a battery management method according to another aspect of this disclosure.
[0130] Preferably, each step of the battery management method can be performed by the battery management device 100. In the following, content overlapping with the previously described content will be omitted or briefly described.
[0131] The feature extraction step (S100) is a step of extracting feature values of the learning battery at each charge and discharge cycle, and can be performed by the feature extraction unit 110.
[0132] The first state determination step (S200) is a step of determining the state of the learned battery at each charging and discharging cycle based on the feature values extracted in the feature value extraction step (S100) and the preset standard values corresponding to the feature values, and can be executed by the first state determination unit 120.
[0133] For example, suppose that the first to Nth feature values of the learning battery are extracted at each of the first to Nth charge and discharge cycles. Here, N can be an integer of 2 or greater. In this case, a total of N standard values can be preset. That is, the first to Nth standard values can be preset to correspond to the first to Nth feature values respectively. In addition, the first state determination unit 120 can determine the state of the learning battery at the first charge and discharge cycle by comparing the first feature value with the first standard value. Similarly, the first state determination unit 120 can determine the state of the learning battery at the Nth charge and discharge cycle by comparing the Nth feature value with the Nth standard value.
[0134] The marking step (S300) is a step of setting a mark on the characteristic value of the learning battery at each charging and discharging cycle based on the state of the learning battery determined at the first state determination step (S200), and can be executed by the marking unit 130.
[0135] Specifically, when the state of the learning battery is determined to be normal, the marking unit 130 can be configured to set a first mark on the characteristic value of the learning battery for the corresponding charging and discharging cycle.
[0136] Conversely, when the state of the learning battery is determined to be a defective state, the marking unit 130 can be configured to set a second mark on the characteristic value of the learning battery for the corresponding charge and discharge cycle.
[0137] The model learning step (S400) is a step of learning a classification model for judging and analyzing the state of the battery based on the feature values that have been labeled in the labeling step (S300), and can be executed by the model learning unit 140.
[0138] The second state determination step (S500) is a step to determine and analyze the state of the battery based on the classification model learned in the model learning step (S400), and can be executed by the second state determination unit 150.
[0139] In other words, the battery management method according to the embodiments of this disclosure has the advantage of obtaining a large amount of learning data from a limited number of batteries for learning a classification model. Therefore, the performance of the learned classification model can be improved. Consequently, the accuracy of battery state determination based on the learned classification model according to one embodiment of this disclosure can be improved.
[0140] The embodiments of this disclosure described above can be implemented not only by devices and methods, but also by a program that implements functions corresponding to the configuration of the embodiments of this disclosure, or a recording medium on which the program is recorded. Based on the above description of the embodiments, those skilled in the art can readily implement the program or recording medium.
[0141] The present disclosure has been described in detail. However, it should be understood that while the detailed description and specific examples indicate preferred embodiments of the present disclosure, they are given by way of illustration only, as various changes and modifications within the scope of the present disclosure will become apparent to those skilled in the art based on this detailed description.
[0142] Furthermore, without departing from the technical aspects of this disclosure, those skilled in the art can make many substitutions, modifications and changes to the disclosure described above, and this disclosure is not limited to the embodiments and drawings described above, and each embodiment can be selectively combined in part or in whole to allow for various modifications.
[0143] (See attached image labels)
[0144] 1: Battery pack
[0145] 10: Battery Module
[0146] 100: Battery Management Device
[0147] 110: Feature Value Extraction Unit
[0148] 120: First State Judgment Unit
[0149] 130: Marker unit
[0150] 140: Model Learning Unit
[0151] 150: Second State Judgment Unit
[0152] 160: Storage unit
Claims
1. A battery management device, comprising: The feature extraction unit is configured to extract feature values of the learning battery at each charge and discharge cycle; The first state determination unit is configured to determine the state of the learning battery at each charging and discharging cycle based on the feature value extracted by the feature value extraction unit and a preset standard value corresponding to the feature value. The marking unit is configured to set a mark for the characteristic value of the learning battery at each charging and discharging cycle based on the state of the learning battery determined by the first state judgment unit. The model learning unit is configured to learn a classification model for determining and analyzing the state of the battery based on the feature values to which the labeling unit has set the label. as well as The second state determination unit is configured to determine the state of the analytical battery based on the classification model learned by the model learning unit. Specifically, when the learning battery is determined to be in a normal state, the marking unit is configured to set a first mark for the corresponding charge and discharge cycle to indicate the normal state of the learning battery. When the state of the learning battery is determined to be defective, the marking unit is configured to set a second mark for the corresponding charge and discharge cycle to indicate the defective state of the learning battery, and to set the second mark for each subsequent charge and discharge cycle after the corresponding charge and discharge cycle, without determining the state of the learning battery.
2. The battery management device according to claim 1, in, The first state determination unit is configured to determine whether the state of the learning battery at each charge and discharge cycle is the normal state or the defective state.
3. The battery management device according to claim 1, in, The marking unit is configured to mark the characteristic value of the learning battery based on the state of the learning battery determined at predetermined charge and discharge cycles.
4. The battery management device according to claim 3, in, When the state of the learning battery is determined to be the normal state at the predetermined charge and discharge cycle, the marking unit is configured to set the first mark on the characteristic value of the learning battery for the charge and discharge cycles up to the cycle immediately preceding the predetermined charge and discharge cycle. and Wherein, when the state of the learning battery is determined to be the defective state at the predetermined charge and discharge cycle, the marking unit is configured to set the second mark on the characteristic value of the learning battery for the charge and discharge cycle after the predetermined charge and discharge cycle.
5. The battery management device according to claim 1, in, The feature extraction unit is configured to extract multiple feature values from the learning battery, and The first state determination unit is configured to determine the state of the learning battery for each of the plurality of feature values at each charging and discharging cycle based on the plurality of feature values of the learning battery and a preset standard value corresponding to each of the plurality of feature values.
6. The battery management device according to claim 5, in, The marking unit is configured to set the mark for each of the plurality of feature values at each charge and discharge cycle.
7. The battery management device according to claim 1, in, The second state determination unit is configured to extract feature values of the analyzed battery at each charge and discharge cycle, and determine the state of the analyzed battery by inputting the extracted feature values into a learned classification model.
8. A battery management device, comprising: The feature extraction unit is configured to extract feature values of the learning battery at each charge and discharge cycle; The first state determination unit is configured to determine the state of the learning battery at each charging and discharging cycle based on the feature value extracted by the feature value extraction unit and a preset standard value corresponding to the feature value. The marking unit is configured to set a mark for the characteristic value of the learning battery at each charging and discharging cycle based on the state of the learning battery determined by the first state judgment unit. The model learning unit is configured to learn a classification model for determining and analyzing the state of the battery based on the feature values to which the labeling unit has set the label. as well as The second state determination unit is configured to determine the state of the analytical battery based on the classification model learned by the model learning unit. Specifically, when the learning battery is determined to be in a normal state, the marking unit is configured to set a first mark for the corresponding charging and discharging cycle to indicate the normal state of the learning battery. Specifically, when the state of the learning battery is determined to be defective, the marking unit is configured to set a second mark for the corresponding charge and discharge cycle to indicate the defective state of the learning battery. The marking unit is configured to count the number of times the state of the learning battery is determined to be the defective state, and when the counted number is equal to or greater than a standard number, the marking unit is configured to set the second mark for each subsequent charge and discharge cycle after the last charge and discharge cycle in which the state of the learning battery is determined to be the defective state, without determining the state of the learning battery.
9. The battery management device according to claim 8, in, The first state determination unit is configured to determine whether the state of the learning battery at each charge and discharge cycle is the normal state or the defective state.
10. The battery management device according to claim 8, in, The marking unit is configured to mark the characteristic value of the learning battery based on the state of the learning battery determined at predetermined charge and discharge cycles.
11. The battery management device according to claim 10, in, When the state of the learning battery is determined to be the normal state at the predetermined charge and discharge cycle, the marking unit is configured to set the first mark on the characteristic value of the learning battery for the charge and discharge cycles up to the cycle immediately preceding the predetermined charge and discharge cycle. and Wherein, when the state of the learning battery is determined to be the defective state at the predetermined charge and discharge cycle, the marking unit is configured to set the second mark on the characteristic value of the learning battery for the charge and discharge cycle after the predetermined charge and discharge cycle.
12. The battery management device according to claim 8, in, The feature extraction unit is configured to extract multiple feature values from the learning battery, and The first state determination unit is configured to determine the state of the learning battery for each of the plurality of feature values at each charging and discharging cycle based on the plurality of feature values of the learning battery and a preset standard value corresponding to each of the plurality of feature values.
13. The battery management device according to claim 12, in, The marking unit is configured to set the mark for each of the plurality of feature values at each charge and discharge cycle.
14. The battery management device according to claim 8, in, The second state determination unit is configured to extract feature values of the analyzed battery at each charge and discharge cycle, and determine the state of the analyzed battery by inputting the extracted feature values into a learned classification model.
15. A battery pack comprising a battery management device according to any one of claims 1 to 14.
16. A battery management method, comprising: The feature extraction step extracts the feature values of the learning battery at each charge and discharge cycle; The first state determination step determines the state of the learning battery at each charging and discharging cycle based on the feature values extracted in the feature value extraction step and the preset standard values corresponding to the feature values. The marking step involves setting a mark for the characteristic value of the learning battery at each charging and discharging cycle based on the state of the learning battery determined in the first state determination step. The model learning step learns a classification model for judging and analyzing the state of the battery based on the feature values of the labels set in the labeling step. as well as The second state determination step determines the state of the analytical battery based on the classification model learned in the model learning step. Wherein, when the state of the learning battery is determined to be in a normal state, the marking step includes a setting step: setting a first mark for the corresponding charging and discharging cycle to indicate the normal state of the learning battery, and Wherein, when the state of the learning battery is determined to be a defective state, the marking step includes a setting step: setting a second mark for the corresponding charging and discharging cycle to indicate the defective state of the learning battery, and setting the second mark for each subsequent charging and discharging cycle after the corresponding charging and discharging cycle, without determining the state of the learning battery.
17. A battery management method, comprising: The feature extraction step extracts the feature values of the learning battery at each charge and discharge cycle; The first state determination step determines the state of the learning battery at each charging and discharging cycle based on the feature values extracted in the feature value extraction step and the preset standard values corresponding to the feature values. The marking step involves setting a mark for the characteristic value of the learning battery at each charging and discharging cycle based on the state of the learning battery determined in the first state determination step. The model learning step learns a classification model for judging and analyzing the state of the battery based on the feature values of the labels set in the labeling step. as well as The second state determination step determines the state of the analytical battery based on the classification model learned in the model learning step. Wherein, when the state of the learning battery is determined to be in a normal state, the marking step includes a setting step: setting a first mark for the corresponding charging and discharging cycle to indicate the normal state of the learning battery, and Wherein, when the state of the learning battery is determined to be defective, the marking step includes a setting step: setting a second mark for the corresponding charge and discharge cycles to indicate the defective state of the learning battery, and The marking step includes: A counting step that counts the number of times the state of the learning battery is judged to be the defective state, and When the counted number is equal to or greater than the standard number, a setting step is taken to set the second flag for each subsequent charge and discharge cycle after the last charge and discharge cycle in which the state of the learning battery is determined to be the defective state, without determining the state of the learning battery.