Diagnostic model generation apparatus and method, and battery diagnostic device

Abstract

A diagnostic model generation apparatus according to an embodiment of the disclosure includes a battery information obtaining unit configured to obtain an index value for a plurality of indices of each of a plurality of batteries in each of a plurality of cycles, and obtain a target value of the plurality of batteries in a target cycle, and a model training unit configured to calculate a correlation coefficient of the target value of the plurality of batteries and each of the plurality of indices in each of the plurality of cycles, determine a target index among the plurality of indices based on the plurality of correlation coefficients calculated for each of the plurality of indices, and train a diagnostic model using the target value of the plurality of batteries and the index value corresponding to the target index of each of the plurality of batteries.

Description

Technical Field

[0001] This application is based on and claims priority to Korean Patent Application No. 10-2024-0143309, filed with the Korean Intellectual Property Office on October 18, 2024, the disclosure of which is incorporated herein by reference in its entirety.

[0002] This disclosure relates to a diagnostic model generation apparatus and method for generating diagnostic models, and a battery diagnostic device for using diagnostic models to diagnose the state of a battery. Background Technology

[0003] Recently, demand for portable electronic products such as laptops, cameras, and mobile phones has increased dramatically, and electric vehicles, energy storage batteries, robots, and satellites have seen significant development. Therefore, high-performance batteries that allow for repeated charging and discharging are being actively researched.

[0004] Currently available batteries include nickel-cadmium (NiCd), nickel-metal hydride (NiMH), nickel-zinc (NiZn), 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 also have a very low self-discharge rate and high energy density.

[0005] While much research is underway to increase the capacity and density of these batteries, improving their lifespan and safety is also crucial. To enhance battery safety, technology is needed to accurately diagnose the current state of the battery. Summary of the Invention

[0006] Technical issues

[0007] This disclosure aims to provide a diagnostic model generation apparatus and method for generating a diagnostic model for estimating the state of a battery. Furthermore, this disclosure aims to provide a battery diagnostic device that uses a trained diagnostic model to estimate the state of a battery.

[0008] These and other objects and advantages of this disclosure will be understood from the following detailed description and will become even more apparent from the exemplary embodiments of this disclosure. Moreover, 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.

[0009] Technical solution

[0010] A diagnostic model generation apparatus according to one aspect of this disclosure may include: a battery information acquisition unit configured to acquire, in each of a plurality of cycles, an index value for each of a plurality of batteries against a plurality of preset indicators, and to acquire a target value for the plurality of batteries in a target cycle; and a model training unit configured to calculate, in each of the plurality of cycles, a correlation coefficient between the target value of the plurality of batteries and each of the plurality of indicators, determine a target indicator among the plurality of indicators based on the plurality of correlation coefficients calculated for each of the plurality of indicators, and train a diagnostic model using the target value of the plurality of batteries and the index value of each of the plurality of batteries corresponding to the target indicator.

[0011] The model training unit can be configured to calculate the correlation coefficient of each of the multiple metrics by considering the linear relationship between the target values ​​of multiple batteries and the metric values ​​of each of the multiple metrics.

[0012] The model training unit can be configured to determine a standardized coefficient among multiple correlation coefficients corresponding to each of multiple indicators, and to compare the determined standardized coefficients with each other to determine the target indicator.

[0013] The model training unit can be configured to compare the magnitudes of multiple standard coefficients and determine at least one target metric based on the comparison results.

[0014] The model training unit can be configured to determine the largest standardized coefficient among multiple standardized coefficients and to determine the index corresponding to the largest standardized coefficient as the target index.

[0015] The model training unit can be configured to determine multiple standard coefficients that are greater than or equal to a preset reference value and to determine the index corresponding to the determined multiple standard coefficients as the target index.

[0016] The model training unit can be configured to determine the largest correlation coefficient among multiple correlation coefficients as the standard coefficient.

[0017] The target loop can be set to a loop following multiple loops.

[0018] A battery diagnostic device according to another aspect of this disclosure may include: a memory configured to store a diagnostic model trained by a diagnostic model generation apparatus according to one aspect of this disclosure; and a processor configured to obtain diagnostic index values ​​of target indicators of the diagnostic battery and diagnose the state of the target cycle of the diagnostic battery by inputting the diagnostic index values ​​into the trained diagnostic model.

[0019] The processor can be configured to estimate target values ​​for the target cycle of the diagnostic battery.

[0020] According to another aspect of this disclosure, the battery pack includes the battery diagnostic device.

[0021] According to another aspect of this disclosure, the vehicle includes the battery diagnostic device.

[0022] A diagnostic model generation method according to another aspect of this disclosure may include: a battery information acquisition step, acquiring the index value of each of the multiple batteries in each of the multiple cycles for a plurality of preset indicators, and acquiring the target value of the multiple batteries in a target cycle; a correlation coefficient calculation step, calculating the correlation coefficient between the target value of the multiple batteries and each of the multiple indicators in each of the multiple cycles; a target indicator determination step, determining a target indicator among the multiple indicators based on the multiple correlation coefficients calculated for each of the multiple indicators; and a training step, training a diagnostic model using the target values ​​of the multiple batteries and the index values ​​of each of the multiple batteries corresponding to the target indicator.

[0023] A computer-readable recording medium storing a computer program for performing a diagnostic model generation method, the diagnostic model generation method comprising: a battery information acquisition step, acquiring indicator values ​​for each of a plurality of batteries against a plurality of preset indicators in each of a plurality of cycles, and acquiring target values ​​for the plurality of batteries in a target cycle; a correlation coefficient calculation step, calculating correlation coefficients between the target values ​​of the plurality of batteries and each of the plurality of indicators in each of the plurality of cycles; a target indicator determination step, determining a target indicator among the plurality of indicators based on the plurality of correlation coefficients calculated for each of the plurality of indicators; and a training step, training a diagnostic model using the target values ​​of the plurality of batteries and the indicator values ​​of each of the plurality of batteries corresponding to the target indicator.

[0024] Beneficial effects

[0025] According to one aspect of this disclosure, the diagnostic model generation apparatus has the advantage of being able to generate a trained diagnostic model that can more accurately diagnose the state of a battery in a target cycle.

[0026] Furthermore, according to one aspect of this disclosure, since the state of the diagnostic battery in the target cycle is diagnosed based on a trained diagnostic model, the future state of the diagnostic battery can be predicted more accurately.

[0027] The effects of this disclosure are not limited to those described above, and other effects not mentioned will be clearly understood by those skilled in the art based on the description of the claims. Attached Figure Description

[0028] The accompanying drawings illustrate preferred embodiments of the present disclosure and, together with the foregoing disclosure, serve 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.

[0029] Figure 1 This is a diagram schematically illustrating a diagnostic model generation apparatus according to an embodiment of the present disclosure.

[0030] Figure 2 This is a schematic diagram illustrating the correlation coefficients according to embodiments of the present disclosure.

[0031] Figures 3 to 5 This is a schematic diagram showing the correlation coefficients for each cycle of the first to third indicators.

[0032] Figure 6 This is a schematic diagram illustrating a battery diagnostic device according to another embodiment of the present disclosure.

[0033] Figure 7 This is a schematic diagram illustrating a battery pack according to yet another embodiment of the present disclosure.

[0034] Figure 8 This is a schematic diagram illustrating a vehicle according to yet another embodiment of the present disclosure.

[0035] Figure 9 This is a diagram schematically illustrating a diagnostic model generation method according to yet another embodiment of the present disclosure. Detailed Implementation

[0036] Before the description, it should be understood that the terms used in the specification and appended claims should not be construed as limited to their general and dictionary meanings, but rather as being interpreted based on the meanings and concepts corresponding to the technical aspects of this disclosure, in accordance with the principle that the inventors are permitted to define the terms appropriately for the best interpretation.

[0037] Therefore, the description presented herein is for illustrative purposes only and is merely a preferred example, and is not intended to limit the scope of this disclosure. It should be understood that other equivalents and modifications may be made thereto without departing from the scope of this disclosure.

[0038] Furthermore, in interpreting this disclosure, if a detailed description of a relevant known structure or function is deemed likely to obscure the essential points of this disclosure, such detailed description will be omitted.

[0039] 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 these elements by terminology.

[0040] Throughout this specification, when a section is referred to as “comprising” or “including” any element, it means that the section may further include other elements without excluding them, unless otherwise specifically stated.

[0041] Furthermore, throughout the specification, when one 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 they are "indirectly connected" when another element is inserted between them.

[0042] Preferred embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings.

[0043] Figure 1 This is a schematic diagram illustrating a diagnostic model generation apparatus 100 according to an embodiment of the present disclosure.

[0044] refer to Figure 1 The diagnostic model generation device 100 may include a battery information acquisition unit 110 and a model training unit 120.

[0045] The battery information acquisition unit 110 can be configured to acquire the index value of each of the multiple batteries for each of the multiple preset indexes in each of the multiple cycles.

[0046] Here, a battery refers to a physically separable, individual cell with a negative and a positive terminal. For example, a lithium-ion cell or a lithium polymer cell can be considered a battery. Additionally, batteries can be cylindrical, prismatic, or pouch-shaped. Furthermore, a battery can also refer to a battery bank, battery module, or battery pack comprising multiple cells connected in series and / or parallel. In the following text, for ease of explanation, a battery is described as referring to a single, independent cell.

[0047] The battery information acquisition unit 110 can communicate with the outside via a wired and / or wireless connection. Furthermore, the battery information acquisition unit 110 can acquire battery information from multiple batteries from the outside.

[0048] The battery information acquisition unit 110 can acquire various indicator values ​​of multiple batteries in each cycle. For example, if the number of batteries is n and m indicators are preset, the battery information acquisition unit 110 can acquire n×m indicator values ​​in each cycle. Here, if the number of cycles is k, the battery information acquisition unit 110 can acquire n×m×k indicator values.

[0049] In addition, multiple metrics can be preset with items that can be obtained from the battery's charging / discharging data. For example, multiple metrics can be preset as follows.

[0050] (1) Charging start voltage: the voltage at the beginning of charging.

[0051] (2) Charging termination voltage: The voltage at the end of charging.

[0052] (3) Discharge initiation voltage: the voltage at the start of discharge.

[0053] (4) Discharge termination voltage: The voltage at the end of the discharge.

[0054] (5) Charging termination resistor: The resistor at the end of charging.

[0055] (6) Discharge termination resistor: The resistance at the end of the discharge.

[0056] (7) Charge / discharge capacity ratio: the ratio of charge amount to discharge amount

[0057] (8) Capacity: The amount of battery charge or discharge.

[0058] (9) SOH (State of Health): Degree of battery degradation

[0059] (10) CC (constant current) charging capacity: the amount of charge during CC charging.

[0060] (11) CC charging capacity ratio: the ratio of CC charging capacity to total charging capacity.

[0061] (12) CC charging time: The time required for CC charging

[0062] (13) CV (constant voltage) charging capacity: the amount of charge generated by CV charging.

[0063] (14) CV charging capacity ratio: the ratio of CV charging capacity to total charging capacity.

[0064] (15) CV charging time: The time required for CV charging.

[0065] (16) Peak voltage or differential capacitance: The voltage or differential capacitance of the peak included in the differential capacitance curve that represents the correspondence between voltage (V) and differential capacitance (dQdV).

[0066] Here, the differential capacitance curve may include one or more peaks. If the differential capacitance curve includes multiple peaks, the voltage or differential capacitance of each of the multiple peaks can be set as an indicator.

[0067] (17) Peak capacity or differential voltage: The capacity or differential voltage of the peak included in the differential voltage curve that represents the correspondence between capacity (Q) and differential voltage (dVdQ).

[0068] Here, the differential voltage curve may include one or more peaks. If the differential voltage curve includes multiple peaks, the capacity or differential voltage of each of the multiple peaks can be set as an indicator.

[0069] (18) The amount or rate of change of each of the indicators (1) to (17): The amount or rate of change of the value of indicator (1) to (17) compared with the reference value of indicator (1) to (17) (e.g., the value of the initial cycle).

[0070] The battery information acquisition unit 110 can be configured to acquire target values ​​for multiple batteries in a target cycle.

[0071] Specifically, the target cycle can be set to a cycle following multiple cycles. For example, the battery information acquisition unit 110 can acquire battery information of multiple batteries from the first to the 100th cycle. Furthermore, the battery information acquisition unit 110 can acquire target values ​​of multiple batteries in the 300th cycle corresponding to the target cycle.

[0072] Here, the target value can be set as a value that represents the state of the battery in the target cycle. Specifically, the target value can be set as a value that directly or indirectly represents the battery's lifespan or degree of degradation in the target cycle. For example, the target value can be set as the battery's SOH, SOHQ (State of Health Quality), SOHC (State of Health Capacity), SOHR (State of Health Resistance), or SOHP (State of Health Power) in the target cycle. Although a limited number of examples of target values ​​have been described above, any value that can represent the battery's performance indicators in the target cycle can be applied as a target value.

[0073] As described above, the battery information acquisition unit 110 can be connected to communicate with the outside. Furthermore, the battery information acquisition unit 110 can obtain multiple target values ​​of the battery in the target cycle from the outside.

[0074] The battery information acquisition unit 110 can be connected to communicate with the model training unit 120 via a wired and / or wireless connection. Furthermore, the battery information acquisition unit 110 can send various acquired information to the model training unit 120.

[0075] The model training unit 120 can be configured to calculate the correlation coefficient between the target values ​​of multiple batteries and each of multiple metrics in each of multiple cycles.

[0076] Specifically, the model training unit 120 can calculate the correlation coefficients of multiple batteries in each cycle. Here, the correlation coefficient is a value corresponding to each metric and can be calculated as a value representing the correlation between metrics of multiple batteries for a cycle.

[0077] Figure 2 This is a schematic diagram illustrating the correlation coefficients according to embodiments of the present disclosure.

[0078] For example, in Figure 2In this embodiment, it is assumed that the charging termination voltages of the first to fourth batteries B1, B2, B3, and B4 are measured for the first to fourth cycles. The model training unit 120 can calculate the correlation coefficient of the first cycle as C1 by considering the correlation between the voltages V11, V12, V13, and V14 of the first batteries B1, B2, B3, and B4 in the first cycle and the target values ​​of the first batteries B1, B2, B3, and B4 in the target cycle. The model training unit 120 can calculate the correlation coefficient of the second cycle as C2 by considering the correlation between the voltages V21, V22, V23, and V24 of the first batteries B1, B2, B3, and B4 in the second cycle and the target values ​​of the first batteries B1, B2, B3, and B4 in the target cycle. The model training unit 120 can calculate the correlation coefficient as C3 for the third cycle by considering the correlation between the voltages V31, V32, V33, and V34 of the third cycle of the first to fourth batteries B1, B2, B3, and B4 and the target values ​​of the target cycles of the first to fourth batteries B1, B2, B3, and B4. Similarly, the model training unit 120 can calculate the correlation coefficient as C4 for the fourth cycle by considering the correlation between the voltages V41, V42, V43, and V44 of the fourth cycle of the first to fourth batteries B1, B2, B3, and B4 and the target values ​​of the target cycles of the first to fourth batteries B1, B2, B3, and B4. In other words, the model training unit 120 can calculate the correlation coefficient for each indicator in each cycle.

[0079] The model training unit 120 can be configured to determine the target indicator among multiple indicators based on multiple correlation coefficients calculated for each of the multiple indicators.

[0080] For example, the target metric can be identified as one of multiple metrics. The model training unit 120 can determine the target metric that best reflects the state of the battery by considering multiple correlation coefficients calculated for multiple metrics.

[0081] As another example, the target metric can be identified as at least one of multiple metrics. The model training unit 120 can determine one or more target metrics that reflect the battery state to a specific level or higher by considering multiple correlation coefficients calculated for multiple metrics.

[0082] First, the model training unit 120 can be configured to determine a standardized coefficient among multiple correlation coefficients corresponding to each of the multiple indicators. Here, a standardized coefficient refers to a single coefficient value determined for each of the multiple indicators. That is, any one of the multiple correlation coefficients corresponding to any one indicator can be determined as the standardized coefficient representing the corresponding indicator.

[0083] For example, assuming there are k cycles, k correlation coefficients can be calculated corresponding to the P-index. This is because a correlation coefficient corresponding to the P-index is determined for each cycle. Furthermore, the model training unit 120 can determine any one of the k correlation coefficients as the standard coefficient representing the P-index.

[0084] Specifically, the model training unit 120 can be configured to determine the largest correlation coefficient among multiple correlation coefficients as the standard coefficient. For example, the model training unit 120 can determine the maximum value among multiple correlation coefficients corresponding to multiple cycles as the standard coefficient.

[0085] Figures 3 to 5 This is a schematic diagram showing the correlation coefficients for each cycle of the first to third indicators. Specifically, Figure 3 This is a diagram illustrating multiple correlation coefficients for the first indicator. Figure 4 This is a diagram illustrating multiple correlation coefficients for the second indicator. Figure 5 This is a diagram illustrating multiple correlation coefficients for the third indicator. Figures 3 to 5 In the embodiments, the number of cycles is the same.

[0086] For example, in Figure 3 In one embodiment, the model training unit 120 can determine t1, which corresponds to the maximum value among multiple correlation coefficients of the first indicator, as the standard coefficient of the first indicator. Figure 4 In one embodiment, the model training unit 120 can determine t2, corresponding to the maximum value among multiple correlation coefficients of the second indicator, as the standard coefficient of the second indicator. Figure 5 In one embodiment, the model training unit 120 can determine t3, which corresponds to the maximum value among multiple correlation coefficients of the third index, as the standard coefficient of the third index.

[0087] Furthermore, the model training unit 120 can be configured to compare multiple standard coefficients with each other to determine the target metric.

[0088] Specifically, the model training unit 120 can compare the determined standard coefficients with each other and determine the target indicator among the multiple indicators based on the comparison results. For example, the model training unit 120 can determine at least one indicator among the multiple indicators as the target indicator.

[0089] exist Figures 3 to 5In this embodiment, the standard coefficient of the first indicator is t1, the standard coefficient of the second indicator is t2, and the standard coefficient of the third indicator is t3. The model training unit 120 can compare t1, t2, and t3 with each other and determine the target indicator among the first to third indicators based on the comparison results. For example, the model training unit 120 can determine the second indicator with the largest corresponding standard coefficient as the target indicator.

[0090] The model training unit 120 can be configured to train a diagnostic model using target values ​​for multiple batteries and indicator values ​​for each of the multiple batteries corresponding to the target indicator.

[0091] Specifically, the target metric determined by the model training unit 120 is the metric most closely related to the state of the target cycle among multiple batteries. Therefore, the model training unit 120 can use the metric values ​​corresponding to the target metric among multiple batteries to train the diagnostic model.

[0092] Here, the process of training the diagnostic model can employ traditional model training methods. For example, model training unit 120 can classify some batteries from a plurality of batteries as training batteries and the remaining batteries as validation batteries. Model training unit 120 can input the target value and index value of the training batteries into the diagnostic model and compare the output value of the diagnostic model with the corresponding target value. Based on the comparison result, model training unit 120 can train the diagnostic model by modifying various parameters or conditions of the diagnostic model. Furthermore, model training unit 120 can input the target value and index value of the validation batteries into the trained diagnostic model and obtain validation output values ​​from the diagnostic model. Finally, model training unit 120 can validate the trained diagnostic model by comparing the obtained validation output values ​​with the target values. By repeating this process, model training unit 120 can train the diagnostic model to achieve validation accuracy exceeding a certain level.

[0093] The diagnostic model generation apparatus 100 according to embodiments of the present disclosure has the advantage of being able to generate a trained diagnostic model that can more accurately estimate the state of a battery in a target cycle.

[0094] Meanwhile, the battery information acquisition unit 110 and / or model training unit 120 included in the diagnostic model generation apparatus 100 may optionally include processors, application-specific integrated circuits (ASICs), other chipsets, logic circuits, registers, communication modems, data processing devices, etc., known in the art, to execute the various control logics performed in this disclosure. Furthermore, when the control logic is implemented as software, the battery information acquisition unit 110 and / or model training unit 120 may be implemented as a set of program modules.

[0095] Storage unit 130 can store data required for the operation and function of each component of the diagnostic model generation apparatus 100, data generated during the execution of operations or functions, etc. Storage unit 130 is not particularly limited in type, as long as it is a known information storage device capable of recording, erasing, updating, and retrieving data. As examples, information storage devices may include RAM, flash memory, ROM, EEPROM, registers, etc. Furthermore, storage unit 130 can store program code that defines procedures executable by each component of the diagnostic model generation apparatus 100.

[0096] The model training unit 120 can be configured to calculate the correlation coefficient of each of the multiple metrics by considering the linear relationship between the target values ​​of the multiple batteries and the metric values ​​of each of the multiple metrics.

[0097] Specifically, the model training unit 120 can calculate the correlation coefficient for each cycle by considering the linear relationship between the target value and the indicator value of multiple batteries in the corresponding cycle. For example, the closer the correlation coefficient calculated by the model training unit 120 is to 1, the stronger the correlation between the target value and the indicator value; conversely, the closer it is to 0, the weaker the correlation. In other words, the model training unit 120 can quantify the correlation between the target value and the indicator value by calculating the linear correlation coefficient.

[0098] Preferably, the model training unit 120 can calculate the correlation coefficient of each cycle of multiple indicators by calculating the Pearson correlation coefficient between the target value and the indicator value of multiple batteries.

[0099] Here, the Pearson correlation coefficient can be calculated in the range of -1 to 1. Specifically, when there is a strong positive correlation between two variables (i.e., when one variable increases, the other variable also increases), the Pearson correlation coefficient is calculated to be close to 1; when there is no correlation between the two variables, the Pearson correlation coefficient is calculated to be close to 0; and when there is a strong negative correlation between the two variables (i.e., when one variable increases, the other variable decreases), the Pearson correlation coefficient is calculated to be close to -1. The model training unit 120 can calculate the correlation coefficient corresponding to each cycle by obtaining the absolute value of the calculated Pearson correlation coefficient.

[0100] For example, in Figure 2In this embodiment, it is assumed that the target cycle SOH of the first battery B1 is SOH1, the target cycle SOH of the second battery B2 is SOH2, the target cycle SOH of the third battery B3 is SOH3, and the target cycle SOH of the fourth battery B4 is SOH4. The model training unit 120 can calculate the correlation coefficient of the first cycle as C1 by considering the correspondence between the target values ​​SOH1, SOH2, SOH3, and SOH4 of the target cycles of the first to fourth batteries B1, B2, B3, and B4 and the index values ​​V11, V12, V13, and V14 of the first cycle, namely SOH1-V11, SOH2-V12, SOH3-V13, and SOH4-V14. Furthermore, the model training unit 120 can calculate the correlation coefficient of the second cycle as C2 by considering the correspondence between the target values ​​SOH1, SOH2, SOH3, and SOH4 of the target cycles of the first batteries B1, B2, B3, and B4 and the index values ​​V21, V22, V23, and V24 of the second cycle, namely SOH1-V21, SOH2-V22, SOH3-V23, and SOH4-V24. Similarly, the model training unit 120 can calculate the correlation coefficient of the third cycle as C3 by considering the correspondence between the target values ​​SOH1, SOH2, SOH3, and SOH4 of the target cycles of the first to fourth batteries B1, B2, B3, and B4 and the index values ​​V31, V32, V33, and V34 of the third cycle, namely SOH1-V31, SOH2-V32, SOH3-V33, and SOH4-V34. Furthermore, the model training unit 120 can calculate the correlation coefficient of the fourth cycle as C4 by considering the correspondence between the target values ​​SOH1, SOH2, SOH3, and SOH4 of the target cycles of the first to fourth batteries B1, B2, B3, and B4 and the index values ​​V41, V42, V43, and V44 of the fourth cycle, namely SOH1-V41, SOH2-V42, SOH3-V43, and SOH4-V44. Here, C1, C2, C3, and C4 are the absolute values ​​of the Pearson correlation coefficients of the corresponding cycles.

[0101] Since the diagnostic model generation device 100 determines the target indicator among multiple indicators by considering only the correlation between the target value and the indicator value without considering the directionality (positive or negative correlation) of the correlation between the target value and the indicator value, indicators that are highly correlated with the target value can be identified as target indicators. Furthermore, the diagnostic model generation device 100 can generate a trained diagnostic model that can more accurately diagnose the state of the battery in the target cycle by training the diagnostic model using the target indicator.

[0102] The following describes various embodiments in which the model training unit 120 determines the target metric among multiple metrics.

[0103] Specifically, the model training unit 120 can be configured to compare the magnitudes of multiple standard coefficients and determine at least one target metric based on the comparison results. In other words, the model training unit 120 can determine the target metric among multiple metrics by comparing the values ​​of the standard coefficients representing each of the multiple metrics.

[0104] In one embodiment, the model training unit 120 can be configured to determine the largest standardized coefficient among a plurality of standardized coefficients, and to determine the index corresponding to the largest standardized coefficient as the target index.

[0105] The standard coefficient indicates the correlation between a battery's corresponding metric and a target value. Therefore, the model training unit 120 can identify the metric with the highest correlation to the target value—that is, the metric with the largest standard coefficient—as the target metric. For example, in... Figures 3 to 5 In the model, the standard coefficient of the first indicator is t1, the standard coefficient of the second indicator is t2, and the standard coefficient of the third indicator is t3. The model training unit 120 can determine the second indicator with the largest standard coefficient among the first to third indicators as the target indicator.

[0106] In another embodiment, the model training unit 120 can be configured to determine a set of standard coefficients that are greater than or equal to a preset reference value, and to determine the index corresponding to the determined standard coefficients as the target index. Here, the reference value can be preset as a criterion for determining the corresponding index as the target index.

[0107] For example, reference values ​​can be preset through experiments and / or theory. Figures 3 to 5 In this embodiment, it is assumed that the reference value is set to a value greater than t3 and less than t1. In this case, the model training unit 120 can determine the first and second indicators, whose standard coefficients are greater than or equal to the reference value, as the target indicators.

[0108] As another example, the model training unit 120 can obtain a coefficient range including multiple standard coefficients and set the values ​​in the coefficient range corresponding to a preset ratio (e.g., 95%) as reference values. For example, assuming the preset ratio is α, the maximum value among the multiple standard coefficients is Cmax and the minimum value is Cmin. The model training unit 120 can set the reference values ​​using the calculation formula "Cmin + {(Cmax - Cmin) × α}" or "Cmax - {(Cmax - Cmin) × (1 - α)}". Furthermore, the model training unit 120 can identify the indicators among the multiple indicators whose standard coefficient values ​​are greater than or equal to the reference values ​​as target indicators.

[0109] As another example, the model training unit 120 can determine a preset number of standard coefficients in descending order of their values, and set the smallest standard coefficient among the determined standard coefficients as a reference value. For example, in Figures 3 to 5 In this embodiment, the preset quantity is assumed to be 2. The model training unit 120 can determine two standard coefficients (t1 and t2) among a plurality of standard coefficients (t1, t2, t3). In addition, the model training unit 120 can set the smallest standard coefficient (t1) among the determined standard coefficients (t1 and t2) as a reference value.

[0110] The diagnostic model generation apparatus 100 according to embodiments of the present disclosure has the advantage of being able to determine, in various ways, a target indicator that is highly correlated with a target value among multiple indicators.

[0111] In one embodiment, the model training unit 120 can determine the final target metric by considering a combination of target metrics determined in various ways.

[0112] For example, the model training unit 120 may select only the overlapping metrics among the target metrics determined in various ways as the final target metrics.

[0113] As another example, the model training unit 120 can determine all target metrics identified in various ways as the final target metric.

[0114] The diagnostic model generation device 100 has the advantage of being able to train a diagnostic model to determine the final target index by synthesizing target indices determined in various ways and using a final target index that is more relevant to the target value of the battery's target cycle.

[0115] Figure 6 This is a schematic diagram illustrating a battery diagnostic device 200 according to another embodiment of the present disclosure.

[0116] refer to Figure 6 The battery diagnostic device 200 may include a memory 210 and a processor 220.

[0117] The memory 210 can be configured to store diagnostic models trained by the diagnostic model generation device 100. Furthermore, the processor 220 can be configured to access the memory 210 and use the trained diagnostic models.

[0118] The processor 220 can be configured to obtain diagnostic indicator values ​​for target indicators of the diagnostic battery.

[0119] For example, processor 220 can be connected to communicate with the outside via wired and / or wireless connections. Furthermore, processor 220 can obtain diagnostic indicator values ​​for target indicators of the battery from the outside.

[0120] The diagnostic battery is the same type of battery as the multiple batteries, but it is an independent battery not included in the multiple batteries. In other words, the multiple batteries are reference batteries used by the diagnostic model generation device 100 to train the diagnostic model, and the diagnostic battery is the battery targeted for state diagnosis.

[0121] Preferably, the diagnostic metric values ​​obtained by the processor 220 can be values ​​of target metrics obtained over multiple cycles. For example, the processor 220 can obtain diagnostic metric values ​​for target metrics in the first to 100th cycles of diagnosing the battery.

[0122] The processor 220 can be configured to diagnose the state of a target cycle of a diagnostic battery by inputting diagnostic metric values ​​into a trained diagnostic model.

[0123] Specifically, when the target indicator value is input, a diagnostic model is trained based on the input data to output the target value of the target cycle. Therefore, the processor 220 can input the diagnostic indicator value into the trained diagnostic model and estimate the target value of the target cycle of the diagnostic battery based on the results output from the trained diagnostic model.

[0124] Meanwhile, the diagnostic metric values ​​obtained by processor 220 are values ​​corresponding to multiple cycles, and these multiple cycles are cycles preceding the target cycle. For example, if the multiple cycles are the first through the 100th cycle, then the target cycle is the 101st cycle or a later cycle. In other words, processor 220 can diagnose the battery state of future cycles based on the diagnostic metric values ​​of the target metric from past cycles.

[0125] For example, assuming the target value is the State of Health (SOH) of a target cycle, multiple diagnostic indicator values ​​are obtained for cycles 1 through 100, and the target cycle is the 300th cycle. The processor 220 can input the multiple diagnostic indicator values ​​into a trained diagnostic model and estimate the SOH of the diagnostic battery for the 300th cycle based on the input results.

[0126] The battery diagnostic device 200 according to embodiments of this disclosure can estimate the future state of a battery using a trained diagnostic model. According to the battery diagnostic device 200, because the future state of the battery is predicted, appropriate charge / discharge control can be performed on the battery to extend its lifespan. This can significantly increase the expected lifespan of the battery.

[0127] The processor 220 disposed in the battery diagnostic device 200 may optionally include application-specific integrated circuits (ASICs), other chipsets, logic circuits, registers, communication modems, data processing devices, etc., known in the art, to execute the various control logics performed in this disclosure. Furthermore, when the control logic is implemented in software, the processor 220 may be implemented as a set of program modules.

[0128] The memory 210 can store data required for the operation and function of each component of the battery diagnostic device 200, data generated during the execution of operations or functions, etc. The memory 210 is not particularly limited in type, 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. Furthermore, the memory 210 can store program code that defines procedures that can be executed by each component of the battery diagnostic device 200.

[0129] The battery diagnostic device 200 according to this disclosure can be applied to a battery management system (BMS). That is, a BMS according to this disclosure may include the aforementioned battery diagnostic device 200. In this configuration, at least some components of the battery diagnostic device 200 can be implemented by supplementing or adding the functionality of components included in a conventional BMS. For example, the memory 210 and processor 220 of the battery diagnostic device 200 can be implemented as components of the BMS.

[0130] In one embodiment, the battery diagnostic device 200 can control the charging and / or discharging of the battery based on the diagnostic results. That is, the processor 220 can control the charging and / or discharging of the battery, enabling the optimization of charging and / or discharging for the battery by utilizing the diagnostic results.

[0131] In another embodiment, the battery diagnostic device 200 can modify various state conditions set for the battery based on the battery's diagnostic results to prevent further battery degradation. For example, the processor 220 can set at least one of the following: upper limit of charge SOC, upper limit of charge voltage, upper limit of charge rate (C), lower limit of discharge SOC, lower limit of discharge voltage, upper limit of discharge rate (C), and upper limit of discharge temperature. For ease of explanation, the state conditions that can be set by the battery diagnostic device 200 have been listed above; however, it should be noted that any state conditions that can be set using the battery's diagnostic results can be applied without limitation to delay battery degradation.

[0132] Furthermore, the battery diagnostic device 200 according to this disclosure can be equipped in a battery pack. That is, a battery pack according to this disclosure may include the aforementioned battery diagnostic device 200 and at least one battery cell. In addition, the battery pack may also include electrical components (relays, fuses, etc.) and a housing, etc.

[0133] Figure 7 This is a schematic diagram illustrating a battery pack 10 according to another embodiment of the present disclosure.

[0134] Furthermore, the positive terminal of battery 11 can be connected to the positive terminal P+ of battery pack 10, and the negative terminal of battery 11 can be connected to the negative terminal P- of battery pack 10. Figure 7 In some embodiments, battery 11 is shown as a single battery cell, but battery 11 may include multiple battery cells connected in series and / or in parallel.

[0135] The measuring unit 12 is electrically connected to the battery 11 and can measure the voltage of the battery 11.

[0136] Furthermore, the measuring unit 12 can be connected to the current measuring unit A. For example, the current measuring unit A can be an ammeter or a shunt resistor capable of measuring the charging current and discharging current of the battery 11. The measuring unit 12 can measure the charging current of the battery 11 and calculate the amount of charge of the battery 11 via the current measuring unit A. In addition, the measuring unit 12 can measure the discharging current of the battery 11 and calculate the amount of discharge of the battery 11 via the current measuring unit A.

[0137] Processor 220 can be connected to communicate with measurement unit 12. Furthermore, processor 220 can obtain diagnostic indicator values ​​of battery 11 for target metrics from measurement unit 12. Processor 220 can use the diagnostic indicator values ​​of battery 11 and a trained diagnostic model to estimate the target value of battery 11 in the target cycle.

[0138] An external device can be connected to the positive terminal P+ and the negative terminal P- of the battery pack 10. For example, the external device can be a charging device or a load. In addition, the positive terminal of the battery 11, the positive terminal P+ of the battery pack 10, the external device, the negative terminal P- of the battery pack 10, and the negative terminal of the battery 11 can be electrically connected.

[0139] Figure 8 The figure schematically illustrates a vehicle 800 according to yet another embodiment of the present disclosure.

[0140] refer to Figure 8The battery pack 10 according to embodiments of this disclosure can be included in a vehicle 800, such as an electric vehicle (EV) or a hybrid vehicle (HV). Additionally, the battery pack 10 can supply power to a motor within the vehicle 800 to drive the vehicle 800. Here, the battery pack 10 may include a battery diagnostic device 200. That is, the battery diagnostic device 200 can be included in the vehicle 800. In this case, the battery diagnostic device 200 can be an on-board device included in the vehicle 800.

[0141] Figure 9 This is a diagram schematically illustrating a diagnostic model generation method according to yet another embodiment of the present disclosure.

[0142] refer to Figure 9 The diagnostic model generation method may include a battery information acquisition step (S100), a correlation coefficient calculation step (S200), a target index determination step (S300), and a training step (S400).

[0143] Preferably, each step of the diagnostic model generation method can be performed by the diagnostic model generation device 100. In the following text, for ease of explanation, content overlapping with the previously described content will be briefly described or omitted.

[0144] The battery information acquisition step (S100) is a step of acquiring the index value of each of the multiple batteries in each of the multiple cycles for a set of preset indicators and acquiring the target value of the multiple batteries in the target cycle, and can be executed by the battery information acquisition unit 110.

[0145] The battery information acquisition unit 110 can acquire battery information for multiple cycles and battery information for a target cycle of multiple batteries from an external device connected for communication. Here, the target cycle is the cycle following the multiple cycles. For example, the multiple cycles can be the first to the 100th cycles, and the target cycle can be the 300th cycle.

[0146] The correlation coefficient calculation step (S200) is a step of calculating the correlation coefficient between each of the multiple indicators and the target value of the multiple batteries for each of the multiple cycles, and can be performed by the model training unit 120.

[0147] The model training unit 120 can calculate the correlation coefficient between the target values ​​of multiple batteries and each metric for each of the multiple loops. For example, the model training unit 120 can calculate the Pearson correlation coefficient between the target values ​​of multiple batteries and the metric values ​​of each metric for each of the multiple loops.

[0148] The target indicator determination step (S300) is a step to determine the target indicator among multiple indicators based on multiple correlation coefficients calculated for each of the multiple indicators, and can be executed by the model training unit 120.

[0149] The model training unit 120 can determine the standardized coefficient among multiple correlation coefficients corresponding to each of the multiple indicators. For example, the model training unit 120 can determine the largest correlation coefficient among the multiple correlation coefficients for each indicator as the standardized coefficient.

[0150] Furthermore, the model training unit 120 can determine the target metric by comparing the determined multiple standard coefficients with each other. For example, the model training unit 120 can determine at least one of the multiple metrics as the target metric.

[0151] The training step (S400) is a step of training a diagnostic model using target values ​​of multiple batteries and index values ​​of each battery in the multiple batteries corresponding to the target index, and can be executed by the model training unit 120.

[0152] The model training unit 120 can generate a trained diagnostic model by training the diagnostic model using target values ​​of multiple batteries and indicator values ​​corresponding to the target indicators of each of the multiple batteries. The trained diagnostic model can be used to predict the state of the target cycle of the diagnostic battery.

[0153] The embodiments of this disclosure described above can be implemented not only by apparatus 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. Those skilled in the art can readily implement the program or recording medium based on the above description of the embodiments.

[0154] Another embodiment of this disclosure may provide a computer-readable storage medium having programs recorded thereon for executing the various embodiments described above on a computer.

[0155] The program can be implemented as hardware components, software components, and / or a combination of hardware and software components. The program can be executed by any system capable of executing computer-readable instructions.

[0156] Software may include computer programs, code, instructions, or combinations thereof, which may configure processing equipment to perform desired operations or may independently or jointly command processing equipment.

[0157] Software can be implemented as a computer program that includes instructions stored on a computer-readable storage medium. Examples of computer-readable storage media include magnetic storage media (e.g., read-only memory (ROM), random access memory (RAM), floppy disks, hard disks, etc.) and optically readable media (e.g., CD-ROM, DVD: digital multifunction disc). Computer-readable storage media can be distributed across network-connected computer systems, allowing computer-readable code to be stored and executed in a distributed manner. The storage medium can be read by a computer, stored in memory, and executed by a processor.

[0158] Computer-readable storage media may be provided in the form of non-transitory storage media. Here, the term "non-transitory storage media" simply means that it is a tangible device and does not contain signals (e.g., electromagnetic waves), and the term does not distinguish between cases where data is stored semi-permanently on the storage medium and cases where data is temporarily stored. For example, a "non-transitory storage medium" may include a buffer in which data is temporarily stored.

[0159] In addition, the program can be provided as part of a computer program product. The computer program product can be traded as a commodity between a seller and a buyer.

[0160] A computer program product may include a software program and a computer-readable storage medium storing the software program. For example, a computer program product may include a product in the form of a software program (e.g., a downloadable application) distributed electronically by a manufacturer of an electronic device or through an electronic marketplace. For electronic distribution, at least a portion of the software program may be stored on the storage medium or temporarily generated. In this case, the storage medium may be the storage medium of a server belonging to the manufacturer of the electronic device, a server of an electronic marketplace, or a relay server temporarily storing the software program.

[0161] This disclosure has been described in detail. However, it should be understood that while the detailed description and specific examples indicate preferred embodiments of this disclosure, they are given by way of illustration only, as various changes and modifications within the scope of this disclosure will become apparent to those skilled in the art from the detailed description.

[0162] Furthermore, without departing from the technical aspects of this disclosure, those skilled in the art can make many substitutions, modifications and changes to this disclosure, and this disclosure is not limited to the above embodiments and drawings, and each embodiment can be selectively combined in part or in whole to allow various modifications.

[0163] [Explanation of reference numerals in the attached figures]

[0164] 10: Battery Pack

[0165] 11: Battery

[0166] 12: Measurement Unit

[0167] 100: Diagnostic model generation device

[0168] 110: Battery Information Acquisition Unit

[0169] 120: Model Training Unit

[0170] 130: Storage unit

[0171] 200: Battery diagnostic equipment

[0172] 210: Memory

[0173] 220: Processor

[0174] 800: Vehicles

Claims

1. A diagnostic model generation device, comprising: A battery information acquisition unit is configured to acquire an index value for each of a plurality of batteries in each of a plurality of cycles for a plurality of preset indicators, and to acquire a target value for the plurality of batteries in a target cycle. as well as A model training unit is configured to calculate the correlation coefficient between the target value of the plurality of batteries and each of the plurality of indicators in each of the plurality of loops, determine a target indicator among the plurality of indicators based on the plurality of correlation coefficients calculated for each of the plurality of indicators, and train a diagnostic model using the target value of the plurality of batteries and the indicator value of each of the plurality of batteries corresponding to the target indicator.

2. The diagnostic model generation device according to claim 1, in, The model training unit is configured to calculate the correlation coefficient of each of the plurality of indicators by considering the linear relationship between the target value of the plurality of batteries and the indicator value of each of the plurality of indicators.

3. The diagnostic model generation device according to claim 1, in, The model training unit is configured to determine a standard coefficient among the plurality of correlation coefficients corresponding to each of the plurality of indicators, and to compare the determined plurality of standard coefficients with each other to determine the target indicator.

4. The diagnostic model generation device according to claim 3, in, The model training unit is configured to compare the magnitudes of the plurality of standard coefficients and determine at least one target metric based on the comparison results.

5. The diagnostic model generation device according to claim 4, in, The model training unit is configured to determine the largest standardized coefficient among the plurality of standardized coefficients and to determine the index corresponding to the largest standardized coefficient as the target index.

6. The diagnostic model generation device according to claim 4, in, The model training unit is configured to determine a plurality of standard coefficients that are greater than or equal to a preset reference value and to determine the index corresponding to the determined plurality of standard coefficients as the target index.

7. The diagnostic model generation device according to claim 3, in, The model training unit is configured to determine the largest correlation coefficient among the plurality of correlation coefficients as the standard coefficient.

8. The diagnostic model generation device according to claim 1, in, The target loop is set as the loop following the plurality of loops.

9. A battery diagnostic device, comprising: A memory configured to store a diagnostic model trained by the diagnostic model generation apparatus according to any one of claims 1 to 8; as well as A processor configured to obtain diagnostic index values ​​of the target index of the diagnostic battery and diagnose the state of the target cycle of the diagnostic battery by inputting the diagnostic index values ​​into a trained diagnostic model.

10. The battery diagnostic device according to claim 9, in, The processor is configured to estimate the target value for the target cycle of the diagnostic battery.

11. A battery pack comprising the battery diagnostic device according to claim 9.

12. A vehicle comprising the battery diagnostic device according to claim 9.

13. A method for generating a diagnostic model, comprising: The battery information acquisition step involves obtaining the index value of each of the multiple batteries for a set of preset indicators in each of the multiple cycles, and obtaining the target value of the multiple batteries in the target cycle. The correlation coefficient calculation step involves calculating the correlation coefficient between the target value of the plurality of batteries and each of the plurality of indicators in each of the plurality of cycles. The target indicator determination step involves determining the target indicator among the plurality of indicators based on the plurality of correlation coefficients calculated for each of the plurality of indicators. as well as The training step involves training a diagnostic model using the target value of the plurality of batteries and the indicator value of each of the plurality of batteries corresponding to the target indicator.

14. A computer-readable recording medium storing a computer program for performing a diagnostic model generation method, the diagnostic model generation method comprising: The battery information acquisition step involves obtaining the index value of each of the multiple batteries for a set of preset indicators in each of the multiple cycles, and obtaining the target value of the multiple batteries in the target cycle. The correlation coefficient calculation step involves calculating the correlation coefficient between the target value of the plurality of batteries and each of the plurality of indicators in each of the plurality of cycles. The target indicator determination step involves determining the target indicator among the plurality of indicators based on the plurality of correlation coefficients calculated for each of the plurality of indicators. as well as The training step involves training a diagnostic model using the target value of the plurality of batteries and the indicator value of each of the plurality of batteries corresponding to the target indicator.

Images