Battery charge and discharge curve analysis method and battery charge and discharge curve analysis device

Abstract

A method and apparatus for battery charge / discharge curve analysis are provided. The method includes training a machine learning model using a plurality of training charge / discharge curves as a training data set, wherein each training charge / discharge curve includes training section classification information, and the training section classification information is a data set in which an identification number of any one of a plurality of charge / discharge control sections sequentially performed in an activation process is assigned to each time index; inputting a target charge / discharge curve acquired through an activation process of a battery cell to the machine learning model; and acquiring target section classification information for the input target charge / discharge curve from the machine learning model. The target section classification information is a data set in which an identification number of any one of the plurality of charge / discharge control sections is assigned to each time index of the target charge / discharge curve.

Description

[0001] This application claims the benefit of Korean Patent Application No. 10-2021-0136161, filed with the Korean Intellectual Property Office on October 13, 2021, the disclosure of which is incorporated herein by reference in its entirety. Technical Field

[0002] This disclosure relates to battery charge / discharge curve analysis, and more specifically, to a technique for analyzing charge / discharge curves monitored from individual cells during the activation process based on the unique characteristics of the activation process steps and automatically assigning segment classification information corresponding to the activation process steps to the charge / discharge curves. Background Technology

[0003] Recently, the demand for portable electronic products such as laptops, cameras and mobile phones has increased dramatically, and with the widespread development of electric vehicles, energy storage batteries, robots and satellites, much research has been conducted on high-performance batteries that can be repeatedly recharged.

[0004] Currently, commercially available batteries include nickel-cadmium batteries, nickel-metal hydride batteries, nickel-zinc batteries, and lithium batteries. Among them, lithium batteries have little or no memory effect. Therefore, due to their advantages of easy recharging, low self-discharge rate, and high energy density, they receive more attention than nickel-based batteries.

[0005] Battery cells assembled into finished products on the production line are finally shipped after undergoing an activation process. During activation, a pre-arranged sequence of charge / discharge control segments is executed sequentially by charging / discharging equipment, and the capacity and performance of the battery cells are set to design specifications accordingly.

[0006] Furthermore, the activation process is the final step in acquiring battery testing data used to filter out defective battery cells before product shipment. Time-series data of various battery parameters (e.g., voltage, current, capacity) acquired from the battery cells during the charge / discharge control phase of the activation process can be compared with preset reference conditions to correlate them with certain charge / discharge control phases to test for battery cell faults.

[0007] Therefore, traditional charging / discharging devices may not be able to properly perform the segmentation function, which involves repeatedly assigning the identification number of the corresponding charging / discharging control segment to the battery parameters during each charging / discharging control segment of the activation process. Even when identification numbers are assigned, more precise segmentation settings may be required to use physical identification segments to determine whether the product is good or bad.

[0008] As mentioned above, it is difficult to extract features from time series data and diagnose and analyze individual battery cells (charge / discharge curves) when there is a lack of segment information in data analysis or when appropriate segment information is not stored for data analysis even if it exists. Summary of the Invention

[0009] Technical issues

[0010] This disclosure is designed to solve the above-mentioned problems, and therefore this disclosure is directed to providing a method and apparatus for automatically assigning an identifier number to each charge / discharge control segment of the activation process for each sample value (data point) in the time series data of the charge / discharge curve obtained by monitoring the battery parameters of the battery cells during the activation process.

[0011] These and other objects and advantages of this disclosure will become apparent from the following description and from embodiments thereof. Furthermore, it will be readily understood that the objects and advantages of this disclosure can be achieved by the means and combinations thereof set forth in the appended claims.

[0012] Technical solution

[0013] The battery charge / discharge curve analysis method according to this disclosure includes training a machine learning model using multiple training charge / discharge curves as a training dataset, wherein each training charge / discharge curve includes training segment classification information, and the training segment classification information is a dataset in which the identifier of any one of a plurality of charge / discharge control segments executed sequentially during the activation process is assigned to each time index; inputting a target charge / discharge curve obtained through the activation process of a battery cell into the machine learning model; and obtaining target segment classification information for the input target charge / discharge curve from the machine learning model. The target segment classification information is a dataset in which the identifier of any one of a plurality of charge / discharge control segments is assigned to each time index of the target charge / discharge curve.

[0014] The target charge / discharge curve may include: time-series voltage data, which indicates the time-related changes in the voltage of a single battery cell according to the time index of the target charge / discharge curve; and time-series current data, which indicates the time-related changes in the charge / discharge current of a single battery cell according to the time index of the target charge / discharge curve.

[0015] Machine learning models can be decision trees.

[0016] The battery charging / discharging curve analysis method may also include: determining whether the target charging / discharging curve is abnormal by comparing the identifier numbers of the time indices assigned to the target segment classification information according to the order of the time indices of the target segment classification information.

[0017] The identifier of the previous charging / discharging control segment among any two of the multiple charging / discharging control segments is less than the identifier of the subsequent charging / discharging control segment.

[0018] The step of determining whether a target charging / discharging curve is abnormal may include determining that the target charging / discharging curve is abnormal when the identifier assigned to the previous time index is greater than the identifier assigned to the subsequent time index in any two time indices of the target segment classification information.

[0019] The step of determining whether a target charge / discharge curve is abnormal may include determining that the target charge / discharge curve is abnormal when any of the multiple identifiers of multiple charge / discharge control segments has a value between any two identifiers of any two adjacent time indices assigned to the classification information of the target segment.

[0020] The step of determining whether a target charge / discharge curve is abnormal may include determining that the target charge / discharge curve is abnormal when at least one of a plurality of charge / discharge control segments has not been assigned a time index to the target segment classification information.

[0021] According to another aspect of the present invention, a battery charge / discharge curve analysis apparatus includes a data acquisition unit configured to store a plurality of training charge / discharge curves, wherein each training charge / discharge curve includes training segment classification information, and the training segment classification information is a dataset in which an identifier of any one of a plurality of charge / discharge control segments executed sequentially during the activation process is assigned to each time index; and a data processing unit configured to train a machine learning model using the plurality of training charge / discharge curves as a training dataset. The data processing unit is configured to acquire target segment classification information of a target charge / discharge curve acquired through the activation process of a battery cell by inputting a target charge / discharge curve into the machine learning model. The target segment classification information is a dataset in which an identifier of any one of a plurality of charge / discharge control segments is assigned to each time index of the target charge / discharge curve.

[0022] Machine learning models can be decision trees.

[0023] The data processing unit can be configured to determine whether the target charging / discharging curve is abnormal by comparing the identifiers of the time indices assigned to the target segment classification information according to the order of the time indices of the target segment classification information.

[0024] According to another aspect of this disclosure, a battery activation system includes a battery charge / discharge curve analysis device.

[0025] Beneficial effects

[0026] According to at least one embodiment of this disclosure, an identifier for each charge / discharge control segment of the activation process can be automatically assigned to each sample value (data point) in the time-series data of the charge / discharge curve obtained by monitoring the battery parameters of individual cells during the activation process. Therefore, even if the charging / discharge device lacks segmentation functionality or the segmentation function fails (an error occurs), the entire charge / discharge curve can be segmented according to the order of the charge / discharge control segments of the activation process, thereby improving the efficiency of feature extraction, diagnosis, and analysis of individual cells.

[0027] The effects of this disclosure are not limited to those described above, and those skilled in the art will clearly understand these and other effects from the appended claims. Attached Figure Description

[0028] The accompanying drawings illustrate exemplary embodiments of the present disclosure and, together with the following detailed description of the present disclosure, are intended to provide a further understanding of the technical aspects of the present disclosure, and therefore the present disclosure should not be construed as limited to the drawings.

[0029] Figure 1 This is a schematic diagram illustrating, by way of example, the architecture of a battery activation system according to an embodiment of the present invention.

[0030] Figure 2 This is an exemplary graph showing the relationship between the charge / discharge curves obtainable from a single battery cell during the activation process and the charge / discharge control segment of the activation process.

[0031] Figure 3 This is a flowchart illustrating an exemplary method for training a machine learning model.

[0032] Figure 4 In describing as through Figure 3 The example decision tree is used as a reference when training a machine learning model using this method.

[0033] Figures 5 to 8 The segmented results of the target charge / discharge curve obtained from the trained machine learning model are shown.

[0034] Figure 9 This is an exemplary flowchart illustrating a battery charge / discharge curve analysis method according to an embodiment of the present disclosure. Detailed Implementation

[0035] In the following, exemplary embodiments of the present disclosure will be described in detail with reference to the accompanying drawings. Before the description, it should be understood that the terms or words used in the specification and appended claims should not be construed as limited to their general or dictionary meanings, but rather as being interpreted based on their meanings and concepts corresponding to the technical aspects of the present disclosure, on the basis of allowing the inventors to appropriately define the terms to obtain the best interpretation.

[0036] Therefore, the embodiments described herein and the illustrations shown in the accompanying drawings are merely exemplary embodiments of this disclosure and are not intended to fully describe the technical aspects of this disclosure. It should be understood that various other equivalent substitutions and modifications may be made thereto in this application.

[0037] Ordinal terms such as “first” and “second” can be used to distinguish one element from another among various elements, but are not intended to limit the elements.

[0038] Unless the context clearly indicates otherwise, it will be understood that, as used herein, the term "comprising" specifies the presence of the said element, but does not exclude the presence or addition of one or more other elements. Additionally, as used herein, the term "unit" refers to at least one processing unit of function or operation, and can be implemented by hardware or software alone or by a combination of hardware and software.

[0039] Furthermore, as will be further understood throughout the specification, when an element is referred to as being “connected to” another element, it may be directly connected to the other element, or there may be an intermediate element present.

[0040] Figure 1 This is a schematic diagram illustrating the architecture of a battery activation system 1 according to an embodiment of the present disclosure.

[0041] refer to Figure 1 The battery activation system 1 includes a charging / discharging device 100 and a battery charging / discharging curve analysis device 200.

[0042] For the activation process of battery cells BC assembled through the production line, a charging / discharging device 100 is provided to sequentially execute a sequence of charging / discharging control sections included in the activation process according to a preset activation schedule.

[0043] The charging / discharging device 100 has charging, discharging, and pausing functions, and is configured to selectively execute one of these functions according to an activation schedule. For each charging / discharging control segment, it charges and discharges the battery cell BC based on preset control parameters (e.g., charging voltage, charging current, discharging voltage, discharging current, etc.). The battery cell BC is not limited to a specific type and can include any battery cell that can be repeatedly recharged, such as a lithium-ion cell.

[0044] The charging / discharging device 100 includes a charger / discharger 110 and a process controller 120.

[0045] The charger / discharger 110 includes a power supply unit 111 and a charging / discharging unit 112.

[0046] The power supply unit 111 is configured to convert power supplied from an AC power source and / or a DC power source into DC power having a predetermined voltage level suitable for the input specifications of the charging / discharging unit 112. The power supply unit 111 may include at least one of an AC-DC converter or a DC-DC converter.

[0047] The charging / discharging unit 112 may include a pair of charging / discharging terminals connected to the positive and negative terminals of the battery cell BC, and may charge or discharge the battery cell BC or stop charging and discharging in response to a command from the process controller 120. The charging / discharging unit 112 may include at least one of a constant current circuit or a constant voltage circuit.

[0048] The process controller 120 pre-records the activation schedule in an embedded memory. In response to user input, the process controller 120 initiates the activation process according to the activation schedule and controls the charger / discharger 110 to sequentially execute each charging / discharging control segment of the activation process according to a preset sequence.

[0049] The battery charge / discharge curve analysis device 200 includes a data acquisition unit 210 and a data processing unit 220. The battery charge / discharge curve analysis device 200 may also include an information output unit 230. The operation of the battery charge / discharge curve analysis device 200 can be activated in response to the start of the activation process of the charging / discharging device 100.

[0050] The data acquisition unit 210 is configured to periodically detect sample values ​​of each battery parameter at predetermined time intervals during the activation process of the battery cell BC. The battery parameters include the voltage and current of the battery cell BC, and the data acquisition unit 210 includes a voltage detector 211 and a current detector 212.

[0051] Voltage detector 211 is connected to the positive and negative terminals of battery cell BC to detect the voltage across battery cell BC and generate (output) a signal indicating a sample value of the detected voltage.

[0052] A current detector 212 is mounted at the charging / discharging path connecting the battery cell BC to the charging / discharging unit 112 to detect the charging / discharging current flowing through the battery cell BC during the activation process and to generate (output) a signal indicating a sample value of the detected current. For example, the current detector 212 may include known current sensing instruments such as a shunt resistor and / or a Hall sensor. Voltage detector 211 and current detector 212 may be integrated into a single chip.

[0053] An information output unit 230 is provided to output various information related to the activation process in a user-recognizable format. In an example, the information output unit 230 may include a monitor, a touch screen, a speaker, and / or a vibrator.

[0054] The data processing unit 220 is operatively coupled to the data acquisition unit 210, the information output unit 230, and the charging / discharging device 100. Operable coupling means directly or indirectly connecting to transmit and receive signals in one or both directions.

[0055] The data processing unit 220 can be implemented in hardware using at least one of the following: application-specific integrated circuit (ASIC), digital signal processor (DSP), digital signal processing instrument (DSPD), programmable logic instrument (PLD), field-programmable gate array (FPGA), microprocessor or electrical unit for performing other functions.

[0056] The data processing unit 220 may have embedded memory. The memory may include at least one type of storage medium, such as flash memory, hard disk, solid-state drive (SSD), silicon disk drive (SDD), multimedia card micro, random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), or programmable read-only memory (PROM). The memory may store data and programs required for subsequent operations of the data processing unit 220. The memory may also store data instructing the results of operations performed by the data processing unit 220.

[0057] Figure 2 This is an exemplary graph showing the relationship between the charge / discharge curves obtainable from a single battery cell during the activation process and the charge / discharge control segment of the activation process.

[0058] refer to Figure 2 The charge / discharge curve includes time-series voltage data and time-series current data.

[0059] Curve C1 shows the time-series voltage data. The time-series voltage data is a collection of sample values ​​of the voltage across cell BC, periodically detected during the activation process, arranged according to the time index of the detection time.

[0060] Curve C2 shows the time-series current data. The time-series current data is a collection of sample values ​​of the charging / discharging current of the battery cell BC, periodically detected during the activation process, arranged according to the time index of the detection time.

[0061] Figure 2 The execution of a total of 6 charge / discharge control segments, namely the first charge / discharge control segment to the sixth charge / discharge control segment N1 to N6, is illustrated in the manner of illustration.

[0062] The first charge / discharge control section N1 is a standby section before charging and is used for the electrical and chemical internal stabilization of the battery cell BC before the second charge / discharge control section N2 is executed on the assembled battery cell BC. The battery cell BC is placed alone without charging / discharging for a predetermined first rest period. In the first charge / discharge control section N1, the battery voltage is maintained constant and the battery current is also maintained constant at 0A.

[0063] The second charge / discharge control segment N2 begins immediately after the end of the first charge / discharge control segment N1, during which battery cell BC is charged at a constant current by the charger / discharger 110. The second charge / discharge control segment N2 may end in response to the battery cell BC reaching a predetermined cutoff voltage or after a predetermined time elapsed from the start time of the second charge / discharge control segment N2. During the second charge / discharge control segment N2, the voltage of battery cell BC continuously increases, and the charging current of battery cell BC remains constant at a preset current rate.

[0064] The third charge / discharge control section N3 begins immediately after the end of the second charge / discharge control section N2, during which battery cell BC is charged at a constant voltage by the charger / discharger 110. The voltage level of the constant current charging can be equal to the cutoff voltage at the second charge / discharge control section N2. The third charge / discharge control section N3 can end in response to the charging current of battery cell BC reaching a predetermined cutoff current or after a predetermined time elapsed from the start time of the third charge / discharge control section N3. In the third charge / discharge control section N3, the voltage of battery cell BC increases more slowly than in the second charge / discharge control section N2. In the third charge / discharge control section N3, the difference between the constant voltage charging voltage level and the voltage of battery cell BC gradually decreases, and the charging current of battery cell BC spontaneously decreases towards 0A.

[0065] The fourth charge / discharge control section N4 is the standby section after charging. To mitigate polarization caused by charging in the second and third charge / discharge control sections, battery cell BC is placed separately and not charged / discharged for a predetermined second pause. In the fourth charge / discharge control section N4, the voltage of battery cell BC gradually decreases from the cutoff voltage to a slightly lower voltage to gradually stabilize the voltage, and in the same manner as in the first charge / discharge control section N1, the current of battery cell BC is kept constant at 0A.

[0066] The fifth charge / discharge control section N5 begins immediately after the end of the fourth charge / discharge control section N4, in which battery cell BC discharges at a constant current through charger / discharger 110. The fifth charge / discharge control section N5 can end in response to the battery cell BC reaching a predetermined discharge end voltage or after a predetermined time has elapsed from the start time of the fifth charge / discharge control section N5. In the fifth charge / discharge control section N5, the voltage of battery cell BC continuously decreases, and the discharge current of battery cell BC remains constant at a preset current rate.

[0067] The sixth charge / discharge control section N6 is the standby section after discharge. To mitigate the polarization caused by discharge through the fifth charge / discharge control section N5, battery cell BC is placed separately and not charged / discharged for a predetermined third pause. At the beginning of the sixth charge / discharge control section N6, the voltage of battery cell BC recovers from the discharge end voltage by the voltage drop calculated by multiplying the constant current discharge rate by the internal resistance of battery cell BC, and then gradually stabilizes. In the same manner as the first and fourth charge / discharge control sections N1 and N4, the current of battery cell BC is kept constant at 0A.

[0068] It should be noted that in each of the first to sixth charge / discharge control sections N1 to N6, sample values ​​of battery parameters uniquely distinguishable from each of the remaining charge / discharge control sections are obtained. In the example, the fifth charge / discharge control section N5 is distinguished from the other charge / discharge control sections because it is the section in which the discharge current flows through the battery cell BC. In another example, the first, fourth, and sixth charge / discharge control sections differ from the second, third, and fifth charge / discharge control sections in that the current is 0A. Furthermore, the first, fourth, and sixth charge / discharge control sections are identical in terms of the 0A current, but the voltage of the battery cell BC in the three sections is set in three non-overlapping voltage sections.

[0069] Because sample values ​​of battery parameters with unique characteristics that distinguish different charge / discharge control segments are acquired in each charge / discharge control segment of the activation process, the battery charge / discharge curve analysis device 200 can provide segmentation functionality to the charge / discharge curve after the activation process is completed by assigning identification numbers to charge / discharge control segments, which are specified by a sample value of one battery parameter or a combination of sample values ​​of two or more battery parameters, arranged by the time index of the charge / discharge curve, and associating them with the time index of the corresponding sample value (or the combination of sample values ​​of two battery parameters). The identification numbers can have an ascending or descending order depending on the order of the charge / discharge control segments. In the example, the identification number of a previous charge / discharge control segment (e.g., N1) can be less than the identification number of a subsequent charge / discharge control segment (e.g., N2). Figure 2 The illustration shows that the difference between the two identifiers of two adjacent charge / discharge control sections is 1. Hereinafter, for ease of explanation, each charge / discharge control section and its identifier will be labeled with the same symbol.

[0070] The segment classification information is a dataset of identifiers arranged according to time indices set within a time range from the start time to the end time of the activation process. The battery charge / discharge curve analysis device 200 can activate the segmentation function by executing a machine learning model trained using multiple training charge / discharge curves as training datasets. Each training charge / discharge curve includes training segment classification information, which is a dataset that assigns the identifier of any one of the multiple charge / discharge control segments executed sequentially during the activation process to each time index. The training segment classification information can be manually entered by the user.

[0071] Figure 3 This is a flowchart illustrating an exemplary method for training a machine learning model, and Figure 4 In describing as a basis Figure 3 The method of training a machine learning model is used as an example of a decision tree referenced in the graph. The battery charge / discharge curve analysis device 200 can perform this operation. Figure 3 This method allows any model capable of multi-class classification, other than decision trees, to be applied as a machine learning model.

[0072] refer to Figure 3 and Figure 4 In step S310, the data processing unit 220 inputs multiple training charge / discharge curves into the machine learning model 400.

[0073] In step S320, the data processing unit 220 obtains the observation segment classification information corresponding to each input training charge / discharge curve from the machine learning model 400. The classification information for each observation segment refers to the time series of the identifier assigned to the dataset classified by the machine learning model 400, i.e., the time index of each training charge / discharge curve.

[0074] In step S330, the data processing unit 220 determines whether at least one observation segment classification information is abnormal by comparing the observation segment classification information with the corresponding training segment classification information. A value of "No" in step S330 indicates that training is complete because the impurities of the machine learning model are less than the reference value. A value of "Yes" in step S330 indicates that additional training is required because the impurities of the machine learning model are equal to or greater than the reference value. When the value of step S330 is "Yes", step S340 is executed.

[0075] In step S340, the data processing unit 220 inputs the training charging / discharging curves corresponding to the classification information of the abnormal observation segment from the multiple training charging / discharging curves into the machine learning model 400.

[0076] Figure 4 The diagram shows a decision tree 400 trained for the first to sixth charge / discharge control segments N1 to N6. The decision tree 400 includes a root node 410, first to fourth intermediate nodes 421 to 424, and first to sixth terminal nodes. The root node 410 and the intermediate nodes can be collectively referred to as internal nodes, each with a unique classification criterion. As a result of classifying the data for each time index (voltage sample value + current sample value), the first to sixth terminal nodes 431 to 436 are associated with the identifiers of the first to sixth charge / discharge control segments N1 to N6, respectively.

[0077] Root node 410 provides stronger classification criteria than the first to fourth intermediate nodes 421-424. Figure 4 In order to classify the current sample value S for each time index into the fifth charge / discharge control segment N5 and the remaining charge / discharge control segments, in the root node 410, the current sample value S for each time index is... C The discharge current value H1 associated with the fifth charge / discharge control section N5 is compared. The discharge current value H1 is set through the training process described above. The identifier N5 associated with the fifth terminal node 435 is assigned to the data of each time index that meets the classification criteria of the root node 410. The data of each time index that does not meet the classification criteria of the root node 410 is input to the first intermediate node 421.

[0078] The first intermediate node 421 provides stronger classification criteria than the second to fourth intermediate nodes 422-424. Figure 4 In order to classify the current sampling value S of each time index into the second charge / discharge control segment N2 and the remaining charge / discharge control segments, in the first intermediate node 421, the current sampling value S of each time index is... C The charging current value H2 associated with the second charging / discharging control section N2 is compared. The charging current value H2 is set through the training process described above. The identifier N2 associated with the second terminal node 432 is assigned to the data of each time index that meets the classification criteria of the first intermediate node 421. The data of each time index that does not meet the classification criteria of the first intermediate node 421 is input to the second intermediate node 422.

[0079] The second intermediate node 422 provides stronger classification criteria than the third and fourth intermediate nodes 423 and 424. Figure 4 In order to classify the sixth charge / discharge control segment N6 and the remaining charge / discharge control segments, at the second intermediate node 422, the voltage sample value S of each time index is... V The voltage value H3 associated with the sixth charge / discharge control segment N6 is compared with the voltage value H3 associated with the sixth charge / discharge control segment N6. The voltage value H3 associated with the sixth charge / discharge control segment N6 is set through the above training process. The identifier N6 associated with the sixth terminal node 436 is assigned to the data of each time index that meets the classification criteria of the second intermediate node 422. The data of each time index that does not meet the classification criteria of the second intermediate node 422 is input to the third intermediate node 423.

[0080] The third intermediate node 423 provides stronger classification criteria than the fourth intermediate node 424. Figure 4 In order to classify the third charge / discharge control segment N3 and the remaining charge / discharge control segments, at the third intermediate node 423, the voltage sample value S of each time index is... V The voltage value H4 associated with the third charge / discharge control section N3 is compared with the voltage value H4 associated with the third charge / discharge control section N3. The voltage value H4 associated with the third charge / discharge control section N3 is set through the above training process. The identifier N3 associated with the third terminal node 433 is assigned to the data of each time index that meets the classification criteria of the third intermediate node 423. The data of each time index that does not meet the classification criteria of the third intermediate node 423 is input to the fourth intermediate node 424.

[0081] The fourth intermediate node, 424, is the node that provides the final classification criteria. Figure 4 In order to classify the first charge / discharge control segment N1 and the fourth charge / discharge control segment N4, at the fourth intermediate node 424, the voltage sample value S of each time index is... VThe voltage value H5 associated with the first charge / discharge control segment N1 is compared with the voltage value H5 associated with the first charge / discharge control segment N1. The voltage value H5 associated with the first charge / discharge control segment N1 is set through the above training process. The identifier N1 associated with the first terminal node 431 is assigned to data for each time index that meets the classification criteria of the fourth intermediate node 424. The identifier N4 associated with the fourth terminal node 434 is assigned to data for each time index that does not meet the classification criteria of the fourth intermediate node 424.

[0082] The data processing unit 220 can assign target segment classification information to the target charge / discharge curve by inputting the target charge / discharge curve into the trained machine learning model 400. The target charge / discharge curve refers to the charge / discharge curve obtained through the actual activation process of assembling the battery cell BC. The target segment classification information refers to the segment classification information assigned to the target charge / discharge curve.

[0083] Figures 5 to 8 The segmentation results of the target charge / discharge curve obtained from the trained machine learning model 400 are shown. The data processing unit 220 can determine whether the target charge / discharge curve is abnormal by comparing the identifiers of the time indices assigned to the target segment classification information, based on the order of the time indices of the target segment classification information. Figures 5 to 8 In time indexing, the symbols K, L, M, N, and O represent natural numbers with the relationship 1 < K < L < M < N < O.

[0084] first, Figure 5 This shows an example of typical target segment classification information assigned to a target charge / discharge curve. (Reference) Figure 3 The identifier of the first charging / discharging control segment N1 is assigned to each time index from time t1 to time t2; the identifier of the second charging / discharging control segment N2 is assigned to each time index from time t2 to time t3; the identifier of the third charging / discharging control segment N3 is assigned to each time index from time t3 to time t4; the identifier of the fourth charging / discharging control segment N4 is assigned to each time index from time t4 to time t5; the identifier of the fifth charging / discharging control segment N5 is assigned to each time index from time t5 to time t6; and the identifier of the sixth charging / discharging control segment N6 is assigned to each time index from time t6 to time t7.

[0085] Subsequently, Figure 6 This example illustrates the classification information for abnormal target segments in the assigned target charge / discharge curve. (See reference) Figure 6 In relation to Figure 5During the comparison, the identifier of the fourth charge / discharge control segment N4, executed after the third charge / discharge control segment N3, is assigned to at least one of the time indices from time t3 to time t4. As shown in Figure 6, when the identifier (e.g., N4) assigned to the previous time index (e.g., M-2) is greater than the identifier (e.g., N3) assigned to the subsequent time index (e.g., M-1) in any two time indices of the target segment classification information, the data processing unit 220 can determine that the target charge / discharge curve is abnormal.

[0086] Subsequently, Figure 7 This shows another example of abnormal target segment classification information for the assigned target charge / discharge curve. (Reference) Figure 7 In relation to Figure 5 During comparison, the identifier of the first charge / discharge control segment N1 is typically assigned to a time index from time t1 to time t2, while the identifier of the fourth charge / discharge control segment N4, instead of the second charge / discharge control segment N2, is assigned to some time index between time t2 and time t3. As shown in Figure 7, when there is an identifier (e.g., N3) with a value between two identifiers (e.g., N2, N4) assigned to any two adjacent time indices (e.g., K, K+1) in the target segment, the data processing unit 220 can determine that the target charge / discharge curve is abnormal.

[0087] Subsequently, Figure 8 This is yet another example showing the classification information of abnormal target segments assigned to the target's charge / discharge curve. (See reference) Figure 8 In relation to Figure 5 During comparison, the identifier of the fourth charge / discharge control segment N4 is assigned to all time indices from time t5 to time t6. For example... Figure 8 As shown, when at least one of the multiple charging / discharging control sections N1 to N6 (e.g., N5) is not assigned a time index to the target section classification information, the data processing unit 220 can determine that the target charging / discharging curve is abnormal.

[0088] Figure 9 This is a flowchart illustrating an exemplary method for analyzing battery charge / discharge curves according to an embodiment of this disclosure.

[0089] refer to Figure 9 In step S910, the data processing unit 220 uses multiple training charge / discharge curves stored in the data acquisition unit 210 as a training dataset (see...). Figure 3 and Figure 4 ) to train machine learning model 400.

[0090] In step S920, the data processing unit 220 inputs the target charge / discharge curve obtained through the activation process of the battery cell BC into the machine learning model 400.

[0091] In step S930, the data processing unit 220 obtains target segment classification information for the input target charge / discharge curve from the machine learning model 400. The target segment classification information can be recorded in a memory or sent to the user via the information output unit 230.

[0092] Figure 9 The method may also include step S940. In step S940, the data processing unit 220 processes the data according to the time index of the target segment classification information (see...). Figures 5 to 8 By comparing the identifier number of the time index assigned to the classification information of the target segment, it is determined whether the target charge-discharge curve is abnormal. When the value of step S940 is "yes", step S950 can be executed.

[0093] In step S950, the data processing unit 220 may output a notification signal, which includes at least one of (i) an anomaly type in the target segment classification information or (i) an identifier being incorrectly assigned to a time index. The notification signal may be recorded in memory or sent to the user via the information output unit 230.

[0094] The embodiments of this disclosure described above are not only implemented by apparatus and methods, but also by a program that performs functions corresponding to the configuration of the embodiments of this disclosure or a recording medium on which the program is recorded. Those skilled in the art can easily implement such implementations through the disclosure of the above embodiments.

[0095] While this disclosure has been described above with respect to a limited number of embodiments and accompanying drawings, this disclosure is not limited thereto and it will be apparent to those skilled in the art that various modifications and variations can be made to it in terms of its technical aspects and in accordance with the appended claims and their equivalents.

[0096] Furthermore, those skilled in the art can make many substitutions, modifications and variations to the above-described disclosure without departing from the technical aspects of this disclosure. Therefore, this disclosure is not limited to the above embodiments and drawings, and all or some embodiments can be selectively combined to allow for various modifications.

[0097] (Refer to the description of the figures)

[0098] 1: Battery Activation System BC: Battery Cell

[0099] 100: Charging / discharging equipment; 110: Charger / discharger

[0100] 111: Power supply unit; 112: Charging / discharging unit

[0101] 120: Controller

[0102] 200: Battery Charge / Discharge Curve Analysis Device

[0103] 210: Data acquisition unit; 211: Voltage detector

[0104] 212: Current Detector

[0105] 220: Data processing unit; 230: Information output unit

Claims

1. A method for analyzing battery charge / discharge curves, comprising: The machine learning model is trained using multiple training charge / discharge curves as training datasets, wherein each training charge / discharge curve includes training segment classification information, and the training segment classification information is a dataset in which the identifier of any one of a plurality of charge / discharge control segments executed sequentially during the activation process is assigned to each time index. The target charge / discharge curve obtained through the activation process of the battery cell is input into the machine learning model; Obtain target segment classification information for the input target charge / discharge curve from the machine learning model, wherein the target segment classification information is a dataset in which the identifier of any one of the plurality of charge / discharge control segments is assigned to each time index of the target charge / discharge curve; and Based on the order of the time indexes in the target segment classification information, the identification number assigned to the time indexes in the target segment classification information is compared to determine whether the target charging / discharging curve is abnormal.

2. The battery charge / discharge curve analysis method according to claim 1, wherein, The target charge / discharge curve includes: Time-series voltage data, indicating the time-related changes in the voltage of the individual battery cells according to the time index of the target charge / discharge curve; and Time-series current data, which indicates the time-related changes in the charging / discharging current of the battery cell according to the time index of the target charging / discharging curve.

3. The battery charge / discharge curve analysis method according to claim 1, wherein, The machine learning model is a decision tree.

4. The battery charge / discharge curve analysis method according to claim 1, wherein, The identifier of the preceding charge / discharge control segment in any two of the plurality of charge / discharge control segments is less than the identifier of the subsequent charge / discharge control segment.

5. The battery charge / discharge curve analysis method according to claim 4, wherein, When the identifier assigned to the previous time index in any two time indices of the target segment classification information is greater than the identifier assigned to the subsequent time index, the target charging / discharging curve is determined to be abnormal.

6. The battery charge / discharge curve analysis method according to claim 1, wherein, The target charge / discharge curve is determined to be abnormal when any one of the multiple identifiers of the plurality of charge / discharge control segments has a value between two identifiers of any two adjacent time indices assigned to the classification information of the target segment.

7. The battery charge / discharge curve analysis method according to claim 1, wherein, The target charge / discharge curve is determined to be abnormal when the identifier of at least one of the plurality of charge / discharge control segments is not assigned to the time index of the target segment classification information.

8. A battery charge / discharge curve analysis device, comprising: A data acquisition unit is configured to store multiple training charge / discharge curves, wherein each training charge / discharge curve includes training segment classification information, and the training segment classification information is a dataset in which an identifier of any one of multiple charge / discharge control segments executed sequentially during activation is assigned to each time index; and A data processing unit is configured to use the plurality of training charge / discharge curves as a training dataset to train a machine learning model. The data processing unit is configured to obtain target segment classification information of the target charge / discharge curve obtained through the activation process of the battery cell by inputting the target charge / discharge curve into the machine learning model. Wherein, the target segment classification information is the dataset in which the identifier of any one of the plurality of charging / discharging control segments is assigned to each of the time indices of the target charging / discharging curve, and The data processing unit is configured to determine whether the target charging / discharging curve is abnormal by comparing the identifiers assigned to the time indices of the target segment classification information according to the order of the time indices of the target segment classification information.

9. The battery charge / discharge curve analysis device according to claim 8, wherein, The machine learning model is a decision tree.

10. A battery activation system comprising a battery charge / discharge curve analysis device according to any one of claims 8 to 9.

Images