Apparatus and method for detecting abnormal cells in a battery pack

Abstract

The present invention discloses an apparatus and method for detecting a cell exhibiting an abnormality in a battery pack. In the apparatus according to the present invention, while the battery pack is being charged according to a charging profile having multiple charging intervals, a control unit acquires first cell voltage time series data (measured values) for each of first through Nth cells in the first half of each charging interval. The control unit then applies a deep learning model to the first cell voltage time series data to determine predicted cell voltage time series data for the second half of the charging interval. The control unit also acquires second cell voltage time series data (measured values) in the second half and determines an error between the second cell voltage time series data and the predicted cell voltage time series data. The control unit then detects a cell exhibiting an abnormality that has a relatively larger error than other cells in at least one charging interval.

Description

[Technical field]

[0001] The present invention relates to an apparatus and method for detecting a cell showing an abnormality in a battery pack, and more particularly to an apparatus and method for detecting a cell showing an abnormality in advance by statistically analyzing the difference between measured and predicted values ​​for the voltage behavior of the cell during charging of the battery pack.

[0002] This application claims priority to Korean Patent Application No. 10-2021-0142202, filed on October 22, 2021, the contents of which are the basis of the priority application are incorporated herein by reference. [Background technology]

[0003] Rechargeable secondary batteries, unlike non-rechargeable primary batteries, are widely used in a variety of fields, including small, cutting-edge electronic devices such as smartphones, laptops, and tablet PCs, as well as electric vehicles and energy storage systems (ESS).

[0004] Although secondary batteries have the efficiency and convenience to be used in such a wide range of fields, they have the characteristic of having a high energy density and therefore have the risk of explosion or fire if they overheat for any reason.

[0005] In particular, the recent occurrence of fire accidents due to battery explosions has raised concerns in the secondary battery market itself, and ensuring the safety of secondary batteries has been cited as a major issue. Summary of the Invention [Problem to be solved by the invention]

[0006] The present invention has been devised against the background of the conventional technology as described above, and has an object to provide an apparatus and method capable of improving the safety of a battery pack by detecting in advance any cell in a battery pack that shows signs of abnormality while the battery pack is being charged. [Means for solving the problem]

[0007] To achieve the above technical objective, an apparatus for detecting a cell showing an abnormality in a battery pack according to one aspect of the present invention includes a voltage measuring unit, a current measuring unit, and a temperature measuring unit that measure the voltage, current, and temperature of first to Nth cells included in a battery pack while the battery pack is being charged, and a control unit operably coupled to the voltage measuring unit, the current measuring unit, and the temperature measuring unit.

[0008] Preferably, the control unit may be configured to acquire first cell voltage time series data by the voltage measurement unit in a first half of a charging interval for each of the first to Nth cells in each charging interval while the battery pack is being charged according to a charging profile having a plurality of charging intervals. The control unit may also be configured to apply a deep learning model to the first cell voltage time series data to determine predicted cell voltage time series data in a second half of the charging interval. The control unit may also be configured to acquire second cell voltage time series data by the voltage measurement unit in the second half, and determine an error between the second cell voltage time series data and the predicted cell voltage time series data. The control unit may also be configured to detect a cell having a relatively larger error than other cells in at least one charging interval as an abnormality indication cell.

[0009] According to one aspect, the control unit may be configured to determine, for each of the first to Nth cells, a maximum difference between the second cell voltage time series data and the predicted cell voltage time series data for each charging interval as the error.

[0010] The control unit may also be configured to determine a first average and a first standard deviation of the errors of the first to Nth cells for each charging interval, determine "(error-first average) / first standard deviation" corresponding to a first standardized value of the errors of each of the first to Nth cells for each charging interval, and detect a cell whose first standardized value is greater than a first threshold value in at least one charging interval as a cell showing an abnormality.

[0011] Preferably, the deep learning model is trained in advance using first cell voltage time series data and second cell voltage time series data measured in the first and second halves of each charging interval for the first to mth learning cells, and may be trained in advance to receive input of the first cell voltage time series data and output predicted cell voltage time series data in which the error with the second cell voltage time series data is minimized.

[0012] According to another aspect, the control unit may be configured to determine, for each of the first to Nth cells, a maximum difference between the second cell voltage time series data and the predicted cell voltage time series data for each charging interval as an error, determine "(error-first average) / first standard deviation" corresponding to a first standardized value of the error for each of the first to Nth cells for each charging interval, and detect a cell in which the first standardized value is greater than a first threshold value in at least one charging interval as a cell indicating an abnormality. Here, the first average and the first standard deviation may be values ​​determined in advance during a learning process of the deep learning model.

[0013] Alternatively, the control unit may be configured to determine a second average and a second standard deviation of the errors of each of the cells included in each of the first to pth modules constituting the battery pack, determine "(error-second average) / second standard deviation" corresponding to a second standardized value of the errors of each of the first to nth cells for each charging interval, and detect a battery cell in which the first standardized value is greater than a first threshold value and the second standardized value is greater than a second threshold value in at least one charging interval as a cell showing an abnormality.

[0014] Alternatively, the control unit may be configured to monitor, for each of the first to Nth cells, whether the relative change behavior between the second cell voltage time series data and the predicted cell voltage time series data corresponds to a change behavior pattern predefined for each type of abnormal sign as the charging interval shifts, and to finally determine the type of abnormal sign for a cell in which a predefined change behavior pattern is confirmed a reference number of times or more.

[0015] Preferably, the predefined change behavior pattern is such that the second cell voltage time series data increases faster than the predicted cell voltage time series data in a first half charging interval and the predicted cell voltage time series data increases faster than the second cell voltage time series data in a second half charging interval, and the type of the abnormal sign may be lithium deposition at the negative electrode.

[0016] The device according to the present invention may further include a recording storage medium in which data, predefined parameters, programs, or a combination thereof are stored, and a display. The control unit may be configured to record identification information regarding the detected abnormal symptom cell in the recording medium, output a message indicating that an abnormal symptom cell has been detected in the battery pack via the display, or transmit the identification information of the abnormal symptom cell to an external device via communication.

[0017] The technical object of the present invention can be achieved by a battery management system including the above-mentioned device for detecting an abnormal symptom cell in a battery pack.

[0018] In order to achieve the above technical objective, a method for detecting a cell showing an abnormality in a battery pack according to another aspect of the present invention includes: While the battery pack is being charged according to a charging profile having a plurality of charging intervals, for each of the first to Nth cells in each charging interval, the method may include: (a) acquiring first cell voltage time series data in a first half of the charging interval; (b) applying a deep learning model to the first cell voltage time series data to determine predicted cell voltage time series data in a second half of the charging interval; (c) acquiring second cell voltage time series data in the second half; (d) determining an error between the second cell voltage time series data and the predicted cell voltage time series data; and (e) detecting a cell having a relatively larger error than other cells in at least one charging interval as a cell showing an abnormality. Effect of the Invention

[0019] According to the present invention, a cell having an abnormality symptom can be easily detected by dividing the charging profile of a battery pack into a plurality of charging sections and statistically comparing and analyzing the behavior of the measured voltage and the predicted voltage for each charging section. Therefore, it is possible to prevent a personal injury accident by early detecting an abnormality symptom directly related to a fire or explosion accident, particularly a serious symptom such as lithium deposition at the negative electrode, and warning the user. In addition, the present invention can effectively deal with other abnormality symptoms by detecting not only lithium deposition at the negative electrode but also voltage change behavior caused by swelling and micro-short circuit phenomena.

[0020] The following drawings attached to this specification are intended to illustrate one embodiment of the present invention and, together with the detailed description of the invention described below, serve to further understand the technical concept of the present invention. Therefore, the present invention should not be interpreted as being limited only to the matters depicted in such drawings. [Brief description of the drawings]

[0021] [Figure 1] 1 is a block diagram showing a schematic configuration of a device for detecting an abnormality symptom cell in a battery pack according to an embodiment of the present invention; [Diagram 2]4 is a graph illustrating an example charging profile according to an embodiment of the present invention. [Diagram 3] 3 is a graph showing the voltage change behavior of a cell while the battery pack is being charged according to the charging profile illustrated in FIG. 2. [Figure 4] 1 is a graph showing an example in which a first half and a second half of a charging section are specifically set in a charging profile according to an embodiment of the present invention; [Diagram 5] 4 is a graph showing an example of first and second cell voltage time series data measured in each charging interval while first to m-th learning cells are charged according to a charging profile according to an embodiment of the present invention. [Figure 6] 13 is a graph showing the change in second cell voltage time series data Vk,i(j) measured in the latter half of charging section (2) of the charging profile for a specific cell in which abnormal signs were detected, the predicted cell voltage time series data V*k,i(j) predicted by a deep learning model, and the difference between them, “Vk,i(j)-V*k,i(j).” [Figure 7] 1 is a graph showing the change in second cell voltage time series data Vk,i(j) measured in the latter half of charging sections (1) to (5) of the charging profile for a specific cell in which lithium has been deposited on the negative electrode, the predicted cell voltage time series data V*k,i(j) predicted by a deep learning model, and the difference between them, “Vk,i(j)-V*k,i(j).” [Figure 8] 4 is a flowchart showing the flow of a method for detecting an abnormality symptom cell in a battery pack according to an embodiment of the present invention. [Figure 9] 4 is a flowchart showing the flow of a method for detecting an abnormality symptom cell in a battery pack according to an embodiment of the present invention. [Figure 10] 4 is a flowchart showing the flow of a method for detecting an abnormality symptom cell in a battery pack according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0022] Hereinafter, a preferred embodiment of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, the terms and words used in this specification and claims should not be interpreted as being limited to their ordinary or dictionary meanings, but should be interpreted as being in accordance with the meaning and concept of the technical idea of ​​the present invention, in accordance with the principle that the inventor himself can appropriately define the concept of the term in order to best describe the invention. Therefore, it should be understood that the embodiment described in this specification and the configuration shown in the drawings are merely the most preferred embodiment of the present invention, and do not represent the entire technical idea of ​​the present invention, and therefore there may be various equivalents and modifications that can be substituted for them at the time of this application.

[0023] FIG. 1 is a block diagram showing a schematic configuration of an apparatus for detecting an abnormality symptom cell in a battery pack according to one embodiment of the present invention.

[0024] Referring to FIG. 1, an apparatus 10 according to an embodiment of the present invention is an apparatus for detecting a cell in a battery pack 20 that shows signs of abnormality while the battery pack 20 is being charged by a charging device 30 according to a charging profile having multiple charging intervals.

[0025] An abnormal cell refers to a cell that exhibits an abnormal voltage change behavior different from that of a normal cell. For example, in the case of a lithium polymer cell, if lithium is precipitated at the negative electrode, a micro-short circuit occurs inside the cell, or swelling occurs, the voltage change behavior exhibited during charging of the cell will be different from that of a normal cell.

[0026] The battery pack 20 includes first to p-th modules 21. The first to p-th modules 21 may be connected in series and / or in parallel to each other. The a-th module is a Includes cell 22 of n a is the total number of cells included in the a-th module. a The cells 22 may be connected in series and / or in parallel with each other. The number of cells included in each module may be the same or different. The total number of cells included in the battery pack 20 is The file is TIFF0007605416000001.tif5157. Hereinafter, the total number of cells included in the battery pack 20 is defined as N, and all the cells in the battery pack 20 may be referred to as the first to Nth cells.

[0027] In one embodiment, the first to Nth cells may be pouch-type lithium polymer cells. However, it is obvious that the present invention is not limited to the type of cell or the type of packaging material. Therefore, the present invention can be applied to other types of secondary battery cells such as lithium-sulfur batteries and natium batteries. In addition, the present invention can be applied to cells having structures such as cylindrical cells and square cells.

[0028] The charging device 30 is a device that applies a charging current to the battery pack 20 according to a charging profile having a plurality of charging intervals. When the battery pack 20 is mounted on an electric vehicle, the charging device 30 may be a charging station. In another example, when the battery pack 20 is mounted on a power storage device, the charging device 30 may be a power converting system (PCS) provided between the power storage device and a power grid. The PCS is a system that controls charging and discharging of the power storage device.

[0029] In the present invention, a charging profile is a protocol that defines how the magnitude of the charging current supplied to the battery pack 20 varies over time.

[0030] FIG. 2 is a graph illustrating an example charging profile 40 according to an embodiment of the present invention.

[0031] 2, the charging profile 40 has a number of charging sections ((1) to (7)). Each charging section has a different magnitude and duration of the charging current. The duration is a time period during which each charging section is maintained.

[0032] In one embodiment, charging profile 40 is a step charging profile in which the magnitude of the charging current is decreased stepwise until the voltage of the cells included in battery pack 20 reaches a cutoff voltage, after which the magnitude of the charging current is adjusted according to a constant voltage charging mode.

[0033] 2, charging sections (1) to (5) are sections in which the magnitude of the charging current is gradually decreased to the cutoff voltage, charging section (6) is a section in which a constant voltage charging mode is applied, and charging section (7) is a section in which a small charging current substantially close to zero is applied until the battery cell is fully charged. In charging profile 40, the substantial charging sections are charging sections (1) to (6).

[0034] FIG. 3 is a graph illustrating the voltage change behavior of the cells while the battery pack 20 is being charged according to the charging profile 40 illustrated in FIG.

[0035] The cells are lithium polymer cells that operate at 3.2 V to 4.2 V. The total number of cells is 100, and since the cells are not the same in terms of degradation, the voltage change behavior of the cells also shows differences.

[0036] Meanwhile, the present invention is not limited by a specific change pattern of the charging profile 40. Therefore, the charging profile 40 in Fig. 2 is merely an example. It should be understood that if a certain charging profile has a plurality of charging sections, the profile corresponds to the charging profile according to the present invention.

[0037] Referring again to FIG. 1, the device 10 according to the present invention includes a voltage measuring unit 11, a current measuring unit 12, a temperature measuring unit 13 and a control unit 14.

[0038] The voltage measurement unit 11 measures the voltages of the first to Nth cells included in the battery pack 20 at regular time intervals while the battery pack 20 is being charged according to a charging profile 40 having a plurality of charging sections, and outputs the cell voltage measurement values ​​to the control unit 14.

[0039] The voltage measurement unit 11 may include a voltage measurement circuit known in the art, and as voltage measurement circuits are widely known, a detailed description thereof will be omitted.

[0040] The current measuring unit 12 measures the magnitude of the charging current at regular time intervals while the battery pack 20 is being charged according to a charging profile 40 having a plurality of charging sections, and outputs the current measurement value to the control unit 14.

[0041] The current measuring unit 12 may be a Hall sensor or a sense resistor that outputs a voltage value corresponding to the magnitude of the current. The voltage value can be converted into a current value according to Ohm's law. When the cells of the first to pth modules 21 are connected in series and the first to pth modules are also connected in series, the current measuring unit 12 may be installed on a line through which the charging current flows. In such an example, the current measurement value measured by the current measuring unit 12 corresponds to the cell current values ​​of all the cells included in the battery pack 20. Of course, it is obvious to those skilled in the art that when there are cells or modules connected in parallel, the current measuring unit 12 can be further installed at an appropriate point on the line through which the charging current flows in order to measure the cell current value.

[0042] The temperature measurement unit 13 measures the temperatures of the first to Nth cells at regular time intervals while the battery pack 20 is being charged according to a charging profile 40 having a plurality of charging sections, and outputs the cell temperature measurement values ​​to the control unit 14.

[0043] The temperature measurement unit 13 may be a thermocouple or a temperature measurement element that outputs a voltage value corresponding to the temperature. The voltage value can be converted to a temperature value using a voltage-temperature conversion lookup table (function).

[0044] The temperature measuring unit 13 may be attached to each of the first to pth modules 21. In this case, the temperature of each module can be regarded as the temperature of the cell contained in the module. Needless to say, the temperature measuring unit 13 is not necessarily provided for each cell.

[0045] The control unit 14 periodically receives cell voltage values ​​of the first to Nth cells from the voltage measurement unit in each charging interval while the battery pack 20 is being charged according to the charging profile 40, and acquires first and second cell voltage time series data for each cell. Here, the cell voltage time series data is a set of cell voltage data measured continuously at multiple points in time.

[0046] In one embodiment, the first cell voltage time series data is a set of cell voltage data measured during the first half of each charging interval, and the second cell voltage time series data is a set of cell voltage data measured during the second half of each charging interval.

[0047] The boundary between the first and second halves of the charging section can be set arbitrarily. For example, when the duration of the i-th charging section is Ti, the first half of the charging section is 0.3T from the start of the charging section. i The latter half of the charging section is 0.3T. i From time T i In this example, the duration of the first half is 0.3T i The duration of the latter half is 0.7T. i It is.

[0048] FIG. 4 is a graph showing an example in which a first half and a second half of a charging section are specifically set in a charging profile 40 according to an embodiment of the present invention.

[0049] In the figure, the vertical solid lines indicate the boundaries between the charging sections, and the dotted lines indicate the boundaries between the first and second halves of the charging sections. The durations of the first and second halves of charging section (1) are 0.3T (1) and 0.7T (1)The duration of the first and second halves of the charging section (2) is 0.3T (2) and 0.7T (2) The duration of the first and second halves of the charging section (3) is 0.3T (3) and 0.7T (3) The duration of the first and second halves of the charging section (4) is 0.3T (4) and 0.7T (4) The duration of the first and second halves of the charging section (5) is 0.3T (5) and 0.7T (5) The duration of the first and second halves of the charging section (6) is 0.3T (6) and 0.7T (6) The duration of the first and second halves of the charging section (7) is 0.3T (7) and 0.7T (7) It is.

[0050] The duration of the charging section may be different for each charging section. The duration of the first and second halves of the charging section may be the same for the entire charging section or may be different for each charging section. That is, the boundary between the first and second halves of the charging section may be set at any time within the duration of the charging section.

[0051] In addition to acquiring the first and second cell voltage time series data for the first to Nth cells by the voltage measurement unit 11 in each charging interval, the control unit 14 can apply a deep learning model to the first cell voltage time series data in each charging interval to determine predicted cell voltage time series data in the latter half of the charging interval.

[0052] The deep learning model is a pre-trained model that is capable of receiving first cell voltage time series data measured in the first half of each charging interval for each learning cell while the first to mth learning cells, which have the same specifications as the cells in the battery pack 20, are being charged according to the above-mentioned charging profile 40, and outputting predicted cell voltage time series data in which the error with respect to the second cell voltage time series data measured in the second half of the charging interval is minimized.

[0053] FIG. 5 is a graph showing an example of first and second cell voltage time series data measured in each charging interval while the first to m-th learning cells are charged according to the charging profile 40 in accordance with an embodiment of the present invention.

[0054] The upper graph in FIG. 5 illustrates an example of voltage curves for 100 of the m learning cells superimposed on one another, and the lower graph in FIG. 5 illustrates an example of first cell voltage time series data and second cell voltage time series data measured in charging sections (1) to (7).

[0055] Preferably, when training a deep learning model, thousands to tens of thousands of training cells each having a different degree of degeneracy may be used to improve the accuracy and reliability of the model.

[0056] 5, the deep learning model may be trained in advance to receive first cell voltage time series data X measured on the learning cell in the first half of the charging interval (1) and output predicted cell voltage time series data in which an error between the first cell voltage time series data X measured on the learning cell in the second half of the charging interval (1) is minimized. The deep learning model may also be trained in advance to receive first cell voltage time series data X measured on the learning cell in the first half of the charging interval (2) and output predicted cell voltage time series data in which an error between the second cell voltage time series data Y measured on the learning cell in the second half of the charging interval (2) is minimized. The deep learning model may also be trained in advance to receive first cell voltage time series data X measured on the learning cell in the first half of the charging interval (3) and output predicted cell voltage time series data in which an error between the second cell voltage time series data Y measured on the learning cell in the second half of the charging interval (3) is minimized. Also, the battery may be pre-trained to receive first cell voltage time series data X measured on the learning cell in the first half of charging interval (4) and output predicted cell voltage time series data in which an error between the first cell voltage time series data X measured on the learning cell in the second half of charging interval (4) is minimized. Also, the battery may be pre-trained to receive first cell voltage time series data X measured on the learning cell in the first half of charging interval (5) and output predicted cell voltage time series data in which an error between the second cell voltage time series data Y measured on the learning cell in the second half of charging interval (5) is minimized. Also, the battery may be pre-trained to receive first cell voltage time series data X measured on the learning cell in the first half of charging interval (6) and output predicted cell voltage time series data in which an error between the second cell voltage time series data Y measured on the learning cell in the second half of charging interval (6) is minimized. In addition, the first cell voltage time series data X measured on the learning cell in the first half of the charging section 7 may be input, and the first cell voltage time series data X may be pre-learned so as to output predicted cell voltage time series data in which the error between the first cell voltage time series data X measured on the learning cell in the second half of the charging section 7 and the second cell voltage time series data Y measured on the learning cell in the second half of the charging section 7 is minimized.

[0057] More preferably, the data used for training the deep learning model may include not only the first cell voltage time series data measured in the first half of the charging interval, but also time series data on the cell current and cell temperature measured in the first half of the charging interval. In this case, the deep learning model may be trained to receive the first cell voltage time series data, the cell current time series data, and the cell temperature time series data measured in the first half of the charging interval, and to output predicted cell voltage time series data in which an error with the second cell voltage time series data measured in the second half of the charging interval is minimized.

[0058] Preferably, the deep learning model can be any model based on an artificial neural network suitable for predicting the time series behavior of the cell voltage. For example, the artificial neural network may be a recurrent neural network (RNN), a convolution neural network (CNN), etc. However, the present invention is not limited by the type of the artificial neural network.

[0059] The control unit 14 can acquire second cell voltage time series data and predicted cell voltage time series data in the latter half of each charging interval for each of the first to Nth cells in the battery pack 20, and then determine the error between the second cell voltage time series data (measured value) and the predicted cell voltage time series data (predicted value).

[0060] In addition, the control unit 14 monitors the error between the second cell voltage time series data determined in each charging interval and the predicted cell voltage time series data for each of the first to Nth cells, and can detect a cell with a relatively large error in at least one charging interval as a cell showing signs of abnormality.

[0061] Specifically, the control unit 14 calculates the second cell voltage time series data V for each charging section by using the following Equation 1 for the first cell: k、1 (j) and predicted cell voltage time series data V k、1* Error E between (j) k、1 can be calculated.

[0062] <Formula 1> E k,i =max(|V k,i (j)-V * k,i (j)|), 1≦j≦Num measure k: charging section index E k,i : Error in the kth charging interval for the ith battery cell V k,i (j): the jth measured second cell voltage in the latter half of the kth charging interval for the ith battery cell V * k,i( j): The jth predicted cell voltage by the deep learning model in the latter half of the kth charging interval for the ith battery cell Num measure The total number of times that the cell voltage was measured in the latter half of the kth charging interval, or the total number of times that the cell voltage was predicted using the deep learning model in the latter half of the kth charging interval max() : A function that returns the maximum value among multiple input values.

[0063] In addition, the control unit 14 also calculates an error E between the second cell voltage time series data and the predicted cell voltage time series data for each charging section for the second to Nth cells. k,2 , E k,3 , ..., E k,n can be calculated.

[0064] FIG. 6 shows the second cell voltage time series data V measured in the latter half of the charging section (2) of the charging profile 40 for a specific cell in which an abnormality symptom has been detected. k,i (j) Predicted cell voltage time series data V predicted by the deep learning model * k,i (j) and the difference between them, "V k,i (j)-V * k,i13 is a graph showing the change in the temperature (j) of the filter.

[0065] The solid black line is V k,i (j) and the black dotted line is V * k,i As shown in Figure 6, the cell showing the abnormality is "V k,i (j)-V * k,i (j)” increases to a significant level. Therefore, the error E k,i can be used as a parameter for detecting abnormality indication cells. For reference, the second cell voltage time series data of a cell without abnormality indication and the predicted cell voltage time series data are shown by gray lines, which are consistent with each other and cannot be clearly distinguished.

[0066] In addition, the control unit 14 calculates the error E of the first to Nth cells for each charging section by using the following Equation 2: k,i First average Avr k and the first standard deviation σ k may be determined.

[0067] <Formula 2> Avr k =mean(E k,i ), 1≦i≦N σ k =std(E k,i ), 1≦i≦N i: The index of the cell N: Number of cells in the battery pack k: charging section index E k,i : Error in the kth charging interval for the ith battery cell mean() : A function that returns the average of multiple input values. std() : A function that returns the standard deviation of multiple input values.

[0068] In addition, the control unit 14 calculates the error E of each of the first to Nth cells for each charging section by using the following Equation 3: k,i The first standardized value Std_Value for k,i The number of charging intervals may be determined as Numcharge Then, the first standardized value Std_Value for each cell is k,i The number of is Num charge There are 10 pieces.

[0069] <Formula 3> Std_Value k,i =(E k,i -A.V.r. k ) / σ k ,1≦k≦Num charge i: The index of the cell k: charging section index Num charge : Number of charging sections AVR k : The first average of the kth charging interval σ k : The first standard deviation of the kth charging interval

[0070] In addition, the control unit 14 calculates the first standardized value Std_Value k,i A cell having a value larger than a preset first threshold value can be detected as a cell indicating an abnormality.

[0071] First standardized value Std_Value k,i teeth 、 The error E of the i-th cell determined in the k-th charging interval k,i is the standard deviation of the error, σ k When using is as the standard, the average error AVr k This is a factor that indicates how far the

[0072] For example, Std_Value k,i If is 2, then E k,i The difference between the error average and the error standard deviation is twice the error standard deviation. Therefore, in the k-th charging interval, the first standardized value Std_Value k,iA cell having a relatively larger Vcc than other cells can be regarded as a cell that may exhibit an abnormality symptom because the error between the measured cell voltage and the predicted cell voltage is relatively large, since the deviation between the voltage change behavior of the abnormality symptom cell (the change behavior of the second cell voltage) and the voltage change behavior of a normal cell (the change behavior of the predicted cell voltage) is large.

[0073] Preferably, the first threshold may be set to 3 or more, more preferably 4 or more, and even more preferably 4.5 or more.

[0074] Meanwhile, in Equation 3, the first average AVr of each charging section k and the first standard deviation σ k may be determined in advance during the learning process of the deep learning model. That is, after the learning of the deep learning model is completed, the first to mth learning cells are charged according to the charging profile 40 having a plurality of charging intervals, while collecting the first cell voltage time series data, the second cell voltage time series data, and the predicted cell voltage time series data, and then the error E k,i In addition, the error E calculated using Equation 2 for each charging section may be determined. k,i The average and standard deviation of the first average AVr k and the first standard deviation σ k may be preset as

[0075] Alternatively, the control unit 14 may calculate an error E k,i The second average AVr k、 a and second standard deviation σ k、a Preferably, the second average AVr can be determined. k, a and second standard deviation σ k,a is determined for each charging section.

[0076] <Formula 4> A vrk,a =mean(E k,i@a ), 1≦i≦n a ,1≦k≦Num charge σk,a =std(E k,i ), 1≦i≦n a ,1≦k≦Num charge a: The module index i: The index of the cell n a :The total number of cells in the a-th module k: charging section index Num charge : Total number of charging sections E k,i@a : Error in the kth charging interval for the i-th cell in the a-th module mean() : A function that returns the average of multiple input values. std() : A function that returns the standard deviation of multiple input values.

[0077] In addition, the control unit 14 calculates the error E of each of the first to Nth cells for each charging section by using the following Equation 5: k,i The second standardized value Std_Value in the module for * k,i The second standardized value Std_Value in the module for each cell can be determined. * k,i The number of is Num charge There are 10.

[0078] <Formula 5> Std_Value * k,i =(E k,i -A.V.r. k,a ) / σ k,a ,1≦i≦N,1≦a≦p,1≦k≦Num charge i: Battery cell index N: Total number of cells a: The module index p: total number of modules k: charging section index Num charge : Total number of charging sections AVR k,a : The second mean defined by the module containing the cell σ k,a : the second standard deviation defined by the module that contains the cell

[0079] In addition, the control unit 14 calculates the first standardized value Std_Value k,i is greater than the first threshold and the second standardized value Std_Value * k,i may be configured to detect a cell having a second threshold value or more as a symptom of an abnormality.

[0080] Second standardized value Std_Value * k,i teeth 、 The cell error E determined in the kth charging interval k,i is the standard deviation of the error in the module σ k,a When the standard is used, the average value of the error within the module, AVr k,a This is a factor that indicates how far the

[0081] For example, Std_Value * k,i If is 2, then E k,i The difference between the mean error within a module and the mean error within a module is twice the standard deviation of the error within a module. Therefore, the second standardized value Std_Value * k,i A cell having a larger V than other cells can be considered a cell that may exhibit an abnormal symptom because the error between the measured cell voltage and the predicted cell voltage is relatively larger than other cells in the module.

[0082] Preferably, the second threshold value may be smaller than the first threshold value, for example, the second threshold value may be set to be preferably 3.0 or less, more preferably 2.5 or less.

[0083] Optionally, the control unit 14 may further execute logic for detecting a type of abnormal symptom of the battery cell.

[0084] Specifically, the control unit 14 may monitor whether the relative change behavior between the second cell voltage time series data of the first to Nth cells measured in the latter half of each charging interval and the predicted cell voltage time series data predicted in the latter half of each charging interval shows a change behavior pattern predefined for each abnormality indication type as the charging interval shifts. Also, when a predefined change behavior is detected in a same cell a reference number of times or more while a plurality of charging cycles are performed, the control unit 14 may identify the cell as an abnormality indication cell and determine the type of the abnormality indication.

[0085] In one example, when lithium deposition occurs at the negative electrode of a lithium polymer cell, the second cell voltage time series data increases faster than the predicted cell voltage time series data in the first half of the charging period. This is because when lithium deposition occurs at the negative electrode, the potential of the negative electrode increases in the first half of the charging period. Since the cell voltage corresponds to the difference between the positive electrode potential and the negative electrode potential, the slope of the change in the cell voltage increases as the negative electrode potential increases. As a result, the second cell voltage increases faster than the predicted cell voltage in the first half of the charging period. This is because the predicted cell voltage is a voltage predicted by a deep learning model, and therefore shows a voltage change behavior close to that of a normal cell in the first half of the charging period, so the increase in the cell voltage is not rapid.

[0086] In contrast, in the charging section in the latter half of the charging, the predicted cell voltage time series data increases faster than the second cell voltage time series data. This is because as charging progresses, the potential of the negative electrode gradually decreases, and when lithium is precipitated from the negative electrode, the amount of lithium involved in the electrochemical reaction decreases, so the decrease in the potential of the negative electrode is offset. If the decrease in the potential of the negative electrode is offset, the increase in the cell voltage is also offset accordingly. As a result, in the latter charging section, the predicted cell voltage increases faster than the second cell voltage. This is because the predicted cell voltage is a voltage predicted by the deep learning model, and therefore the voltage change behavior is close to that of a normal cell even in the latter half of the charging, so the decrease in the potential of the negative electrode is not offset.

[0087] FIG. 7 shows the second cell voltage time series data V measured in the latter half of the charging sections (1) to (5) of the charging profile 40 for a particular cell in which lithium was actually deposited on the negative electrode. k,i (j) Predicted cell voltage time series data V predicted by the deep learning model * k,i (j), and the difference between them, "V k,i (j)-V * k,i 10(j)" changes in the voltage (milli-volts) in the horizontal axis.

[0088] The black solid line is V k,i (j) and the black dotted line is V * k,i As shown in FIG. 7, in the cell where lithium deposition occurred, the second cell voltage time series data increases faster than the predicted cell voltage time series data in the first half of the charging interval (1). In contrast, in charging intervals (3), (4), and (5), the predicted cell voltage time series data increases faster than the second cell voltage time series data. The second cell voltage time series data and the predicted cell voltage time series data of the cell without abnormality signs are shown by gray lines and are consistent with each other and cannot be clearly distinguished.

[0089] In one embodiment, when the control unit 14 confirms that a cell among the first to Nth cells exhibits a behavior in which the second cell voltage time series data increases faster than the predicted cell voltage time series data in the first half of the charging interval and the predicted cell voltage time series data increases faster than the second cell voltage time series data in the second half of the charging interval, the control unit 14 may identify that there is an abnormal sign of lithium deposition in the negative electrode of the cell.

[0090] Alternatively, while charging the battery pack 20 according to the charging profile 40 including a plurality of charging intervals, the control unit 14 may increase the abnormality sign count for a cell by one each time a cell is identified among the first to Nth cells in which the second cell voltage time series data increases faster than the predicted cell voltage time series data in the first half of the charging interval and the predicted cell voltage time series data increases faster than the second cell voltage time series data in the second half of the charging interval.

[0091] Furthermore, when the abnormality indication count is equal to or greater than a reference number, the control unit 14 may finally determine that there is an abnormality indication that lithium has been deposited on the negative electrode of the cell.

[0092] Referring again to FIG. 1, the device 10 according to the present invention may further include a storage medium 15 on which data, predefined parameters, programs or combinations thereof are stored, and a display 16.

[0093] When the control unit 14 detects an abnormality indication cell according to the above-described embodiment, the control unit 14 may record identification information of the abnormality indication cell and / or information regarding the abnormality indication type in the recording medium 15 together with a time stamp.

[0094] The identification information of the abnormal symptom cell includes a model code of the battery pack 20, a module code including the abnormal symptom cell, a production lot number of the abnormal symptom cell, or a combination thereof. The information regarding the type of abnormal symptom may include a diagnostic code indicating lithium deposition in the negative electrode.

[0095] In addition, when the control unit 14 detects an abnormality indicating cell according to the above-described embodiment, the control unit 14 may be configured to output a message via the display 16 indicating that an abnormality indicating cell has been detected in the battery pack 20.

[0096] When the battery pack 20 is installed in an electric vehicle, the display 16 may be an instrument panel or a vehicle integrated control display of the electric vehicle. In another example, when the battery pack 20 is installed in a power storage device, the display 16 may be a display included in an integrated control computer of the power storage device. However, the present invention is not limited by the type of display.

[0097] Meanwhile, the device 10 according to the present invention may further include a communication interface 17. In this case, the control unit 14 can transmit identification information regarding the abnormal symptom cell and / or information regarding the type of the abnormal symptom to an external device via the communication interface 17.

[0098] The communication interface 17 supports wired or wireless communication. The communication interface 17 can support data transmission and reception by a controller area network (CAN), a daisy chain, RS-232, or the like. The communication interface 17 can also support data transmission and reception via short-range wireless communication such as Wi-Fi (registered trademark), Bluetooth (registered trademark), or Zigbee (registered trademark). The communication interface 17 can also support wide-area data transmission and reception via wired or wireless Internet, base station communication, or satellite communication.

[0099] The external device may be a charging device. In another example, the external device may be a cloud server that collects status information of the battery pack 20. In yet another example, the external device may be a diagnostic device that checks the performance of the battery pack 20.

[0100] In the present invention, the control unit 14 may selectively include a processor, an application-specific integrated circuit (ASIC), other chipsets, logic circuits, registers, communication modems, data processing devices, etc., as known in the art, for executing various control logic.

[0101] There is no particular limitation on the type of recording medium 15, so long as it is a medium capable of recording and erasing information. For example, recording medium 15 may be a hard disk, RAM, ROM, EEPROM, register, or flash memory. Recording medium 15 can store and / or update and / or erase and / or transmit a program including a control logic executed by control unit 14, and / or data generated when the control logic is executed, as well as predefined lookup tables, functions, parameters, chemical / physical / electrical constants, and the like.

[0102] At least one control logic of the control unit 14 can be combined, and the combined control logic can be created as a computer-readable code set and recorded on the recording medium 15. The code set may be distributed, stored, and executed in computers connected via a network. In addition, functional programs, codes, and code segments for realizing the combined control logic can be easily inferred by a programmer skilled in the art to which the present invention pertains.

[0103] The device 10 according to the present invention may be included in a battery management system or a battery diagnostic system. The battery management system is a system that controls the overall operation of the battery pack 20. Such a battery management system may be an integrated control system included in a load device in which the battery pack 20 is mounted, such as an electric vehicle or a power storage device. The device 10 according to the present invention may be included as a part of another device or system, as necessary, in addition to the battery management system or the battery diagnostic system.

[0104] Hereinafter, a method for detecting a cell showing an abnormality in a battery pack according to an embodiment of the present invention will be described with reference to FIGS.

[0105] 8, first, in step S10, the charging device 30 applies a charging current to the battery pack 20 according to a charging profile 40 including a plurality of charging intervals. The charging device 30 may be a charging station for an electric vehicle or a PCS for a power storage device.

[0106] The application of the charging current may be started by a request from the control unit 14. That is, the control unit 14 can recognize that the charging cable is connected to the high potential line and the low potential line of the battery pack 20 and can request the charging device to start charging. Alternatively, the application of the charging current may be started automatically when the charging device 30 is connected to the battery pack 20.

[0107] Next, in step S20, while the battery pack 20 is being charged according to the charging profile 40, the control unit 14 periodically receives cell voltage values ​​from the voltage measurement unit for the first to Nth cells in the battery pack 20 in each charging section, and acquires first and second cell voltage time series data.

[0108] Next, in step S30, in addition to acquiring the first and second cell voltage time series data by the voltage measurement unit 11 in each charging interval, the control unit 140 applies a deep learning model previously trained on the first cell voltage time series data in each charging interval for the first to Nth cells to determine predicted cell voltage time series data in the latter half of the charging interval.

[0109] Next, in step S40, the control unit 14 calculates an error E between the second cell voltage time series data and the predicted cell voltage time series data in each charging section for the first to Nth cells using Equation 1. k,i In step S40, the number of charging sections is calculated as Num charge Then, the error E k,iThe total number of charge *N pieces.

[0110] Next, in step S50, the control unit 14 calculates the error E of the first to Nth cells for each charging section using Equation 2. k,i First average Avr k and the first standard deviation σ k Determine.

[0111] Next, in step S60, the control unit 14 calculates the error E of each of the first to Nth cells for each charging section using Equation 3. k,i The first standardized value Std_Value for k,i Determine the number of charging sections. charge Then, the first standardized value Std_Value k,i The total number of charge *N pieces.

[0112] Next, in step S70, the control unit 14 calculates the first standardized value Std_Value k,i A cell having a value larger than a preset first threshold value can be detected as a cell indicating an abnormality.

[0113] After step S70, step S80 or step S90 may be selectively performed.

[0114] In step S80, when the control unit 14 detects the abnormal symptom cell, the control unit 14 may record identification information of the abnormal symptom cell and / or information on the abnormal symptom type together with a time stamp in the recording medium 15. The identification information of the abnormal symptom cell includes a model code of the battery pack 20, a module code to which the abnormal symptom cell belongs, a production lot number of the abnormal symptom cell, or a combination thereof. The information on the type of abnormal symptom may include a diagnostic code indicating lithium deposition in the negative electrode.

[0115] Furthermore, in step S90, when the control unit 14 detects an abnormality indicating cell, the control unit 14 may output, via the display 16, a message indicating that an abnormality indicating cell has been detected in the battery pack 20. When the message is output, the user can stop using the battery pack 20 and replace the battery pack 20, or can request a detailed inspection of the battery pack 20 at a repair shop, an AS center, or the like.

[0116] Note that step S50 may be omitted. In this case, the first average AVr k and the first standard deviation σ k may be determined in advance during the learning process of the deep learning model. That is, after the learning of the deep learning model is completed, the first to mth learning cells are charged according to the charging profile 40, and the error E k,i In addition, the error E for each charging section may be determined using Equation 2. k,i Calculate the average and standard deviation of the first average AVr k and the first standard deviation σ k The first average AVr set during the learning process of the deep learning model may be set in advance as k and the second standard deviation σ k may be stored in advance in recording medium 15, and may be referred to by control unit 14 when step S60 is performed.

[0117] In a preferred embodiment, the steps after step S60 can be modified as shown in FIG.

[0118] That is, in step S100, the control unit 14 calculates an error E k,i The second average AVr k、a and the second standard deviation σ k、a may be determined.

[0119] Next, in step S110, the control unit 14 calculates the error E of each of the first to Nth cells for each charging section using Equation 5.k,i The second standardized value Std_Value in the module for * k,i Determine the number of charging sections. charge Then, the second standardized value Std_Value in the module for each cell is * k,i The number of is Num charge There are 10 pieces.

[0120] Next, in step S120, the control unit 14 calculates the first standardized value Std_Value k,i is greater than the first threshold and at the same time the second standardized value Std_Value * k,i A cell having a value greater than the second threshold can be detected as a cell indicating an abnormality.

[0121] Preferably, the second threshold value may be smaller than the first threshold value. For example, the first threshold value may be 3.0 or more, preferably 4.0 or more, more preferably 4.5 or more. Also, the second threshold value may be set to 3.0 or less, more preferably 2.5 or less.

[0122] After step S120, step S80 or step S90 may be performed substantially similarly to the above-described embodiment.

[0123] That is, when the control unit 14 detects an abnormal symptom cell in the battery pack 20, the control unit 14 may record identification information of the abnormal symptom cell and / or information on the abnormal symptom type together with a time stamp in the recording medium 15. The identification information of the abnormal symptom cell includes a model code of the battery pack 20, a module code to which the abnormal symptom cell belongs, a production lot number of the abnormal symptom cell, or a combination thereof. The information on the type of abnormal symptom may include a diagnostic code indicating lithium deposition in the negative electrode.

[0124] Furthermore, when the control unit 14 detects an abnormality indicating cell in the battery pack 20, the control unit 14 may output, via the display 16, a message indicating that an abnormality indicating cell has been detected in the battery pack 20.

[0125] The method for detecting an abnormal symptom cell in a battery pack according to the present invention may further include identifying a type of an abnormal symptom of the battery cell.

[0126] As shown in FIG. 10, in step S130, the control unit 14 may monitor whether the relative change behavior between the second cell voltage time series data of the first to Nth cells measured in the latter half of each charging interval and the predicted cell voltage time series data predicted in the latter half of each charging interval shows a change behavior pattern predefined for each type of abnormal symptom as the charging interval shifts.

[0127] In one embodiment, the control unit 14 may monitor whether or not there is a battery cell among the first to Nth cells that exhibits a behavior pattern in which the second cell voltage time series data increases faster than the predicted cell voltage time series data in the first half of the charging interval, and the predicted cell voltage time series data increases faster than the second cell voltage time series data in the second half of the charging interval.

[0128] Next, in step S140, the control unit 14 determines whether or not the relative change behavior between the second cell voltage time series data and the predicted cell voltage time series data corresponds to a change behavior pattern defined in advance according to the type of abnormality sign.

[0129] If the result of step S140 is YES, the process proceeds to step S150. Conversely, if the result of step S140 is NO, the process returns to step S130.

[0130] If the result of step S140 is YES, the control unit 140 identifies a type of abnormality sign corresponding to the predefined changing behavior pattern in step S150. In one embodiment, when a battery cell is identified among the first to Nth cells that shows a behavior pattern in which the second cell voltage time series data increases faster than the predicted cell voltage time series data in the first half of the charging interval and the predicted cell voltage time series data increases faster than the second cell voltage time series data in the second half of the charging interval, the control unit 14 can identify that there is an abnormality sign of lithium deposition in the negative electrode of the cell.

[0131] Next, in step S160, the control unit 14 increments the abnormality symptom count by 1 for the cell in which the type of abnormality symptom (for example, lithium deposition in the negative electrode) has been identified.

[0132] Next, in step S170, the control unit 14 determines whether the abnormality sign count exceeds a reference number.

[0133] If the determination in step S170 is YES, the process proceeds to step S180. Conversely, if the determination in S170 is NO, the process returns to step S130.

[0134] If the determination in step S170 is YES, the control unit 14 makes a final decision in step S180 on the type of abnormality sign (for example, lithium deposition in the negative electrode) for the cell whose abnormality sign count exceeds the reference number of times.

[0135] After step S180 is performed, step S80 or step S90 may be performed in the same manner as in the above-described embodiment.

[0136] That is, when the control unit 14 identifies an abnormal symptom for a specific cell in the battery pack 20 and finally determines the type of the abnormal symptom, it can record identification information of the abnormal symptom cell and information on the type of the abnormal symptom together with a time stamp in the recording medium 15. The identification information of the abnormal symptom cell includes a model code of the battery pack 20, a module code to which the abnormal symptom cell belongs, a production lot number of the abnormal symptom cell, or a combination thereof. The information on the type of the abnormal symptom may include a diagnostic code indicating lithium deposition in the negative electrode.

[0137] In addition, when an abnormal symptom cell is identified within the battery pack 20 and the type of the abnormal symptom is finally determined, the control unit 14 may output a message indicating that an abnormal symptom cell has been detected within the battery pack 20 via the display 16, together with information regarding the type of the abnormal symptom.

[0138] Meanwhile, the control unit 14 can transmit identification information regarding the abnormal symptom cell and / or information regarding the type of the abnormal symptom to an external device via the communication interface 17. The communication interface 17 can support wired communication or wireless communication.

[0139] The external device may be a charging device. In another example, the external device may be a cloud server that collects status information of the battery pack 20. In yet another example, the external device may be a diagnostic device that checks the performance of the battery pack 20.

[0140] In the present invention, the abnormal symptom type may include, in addition to lithium deposition at the negative electrode, other abnormal symptom types such as cell swelling, internal micro-short circuit, etc. For each abnormal symptom type, it is obvious to a person skilled in the art to which the present invention pertains that it is possible to easily confirm by experiment what relative change behavior pattern the second cell voltage time series data and the predicted cell voltage time series data show as the charging interval shifts.

[0141] According to the present invention, the charging profile of the battery pack 20 is divided into a plurality of charging sections, and the behavior of the actual voltage and the predicted voltage is statistically compared and analyzed for each charging section, thereby easily detecting cells with abnormal symptoms. Therefore, it is possible to prevent accidents involving personal injury by early detection of abnormal symptoms directly related to fires and explosions, particularly serious symptoms such as lithium deposition at the negative electrode, and warning the user. In addition, the present invention can effectively deal with other abnormal symptoms by detecting not only lithium deposition at the negative electrode, but also voltage change behaviors caused by swelling, micro-short circuit phenomena, etc.

[0142] Although the present invention has been described above using limited embodiments and drawings, the present invention is not limited thereto, and it goes without saying that various modifications and variations can be made by a person having ordinary knowledge in the technical field to which the present invention pertains within the scope of the technical concept of the present invention and the scope of the claims.

Claims

1. a voltage measuring unit, a current measuring unit and a temperature measuring unit for measuring a voltage, a current and a temperature of a first to an Nth cell included in the battery pack while the battery pack is being charged; a controller operatively coupled to the voltage measurement unit, the current measurement unit, and the temperature measurement unit; While the battery pack is being charged according to a charging profile having a plurality of charging intervals, the control unit acquires first cell voltage time series data for each of the first to Nth cells in each charging interval in a first half of the charging interval, applies a deep learning model to the first cell voltage time series data to determine predicted cell voltage time series data in a second half of the charging interval, acquires second cell voltage time series data by the voltage measurement unit in the second half of the charging interval, and determines an error between the second cell voltage time series data and the predicted cell voltage time series data; The control unit detects a cell having a relatively larger error than other cells in at least one charging section as a cell exhibiting an abnormality.

2. The control unit is determining a maximum difference between the second cell voltage time series data and the predicted cell voltage time series data for each of the first to Nth cells as an error for each charging interval; determining a first average and a first standard deviation of the errors of the first to Nth cells for each charging interval; and determining "(error-first average) / first standard deviation" corresponding to a first standardized value of the errors of the first to Nth cells for each charging interval; The device for detecting an abnormality symptom cell in a battery pack according to claim 1 , further comprising: detecting a cell in which the first normalized value is greater than a first threshold value in at least one charging period as an abnormality symptom cell.

3. The deep learning model is The learning is performed in advance using the first cell voltage time series data and the second cell voltage time series data measured in the first to mth learning cells in the first and second half of each charging interval, 2. The device for detecting an abnormal cell in a battery pack as described in claim 1, which is pre-trained to receive first cell voltage time series data and output predicted cell voltage time series data in which an error with respect to the second cell voltage time series data is minimized.

4. The control unit is determining a maximum difference between the second cell voltage time series data and the predicted cell voltage time series data for each of the first to Nth cells as an error for each charging interval, and determining "(error-first average) / first standard deviation" corresponding to a first standardized value of the error for each of the first to Nth cells for each charging interval; a cell having the first normalized value greater than a first threshold value in at least one charging section is detected as a cell showing a malfunction, The device for detecting an abnormality symptom cell in a battery pack according to claim 3 , wherein the first average and the first standard deviation are values ​​determined in advance during a learning process of the deep learning model.

5. The battery pack includes first to pth modules connected in series and / or in parallel, The control unit is determining, for each of the first through pth modules, a second mean and a second standard deviation of the errors of each of the cells contained within the module; determining "(error-second average) / second standard deviation" corresponding to a second normalized value of the error of each of the first to Nth cells for each charging section; 5. The device for detecting an abnormality indication cell in a battery pack according to claim 2 or 4, further comprising: detecting a battery cell in which the first normalized value is greater than a first threshold value and the second normalized value is greater than a second threshold value in at least one charging section as an abnormality indication cell.

6. The control unit monitors whether a relative change behavior between the second cell voltage time series data and the predicted cell voltage time series data corresponds to a change behavior pattern predefined for each type of abnormality sign as a charging interval shifts for each of the first to Nth cells; The device for detecting an abnormal symptom cell in a battery pack according to claim 1 , further comprising: a final determination of a type of abnormal symptom of a cell in which a predefined change behavior pattern is confirmed a reference number of times or more.

7. the predefined change behavior pattern is such that the second cell voltage time series data increases faster than the predicted cell voltage time series data in a first half charging section, and the predicted cell voltage time series data increases faster than the second cell voltage time series data in a second half charging section; The device for detecting an abnormal symptom cell in a battery pack according to claim 6 , wherein the type of the abnormal symptom is lithium deposition in a negative electrode.

8. Further comprising a recording medium on which data, predefined parameters, programs or combinations thereof are stored, and a display; 5. The device for detecting an abnormality symptom cell in a battery pack according to claim 1, wherein the control unit records identification information regarding the detected abnormality symptom cell in the recording medium, or outputs a message indicating that an abnormality symptom cell has been detected in the battery pack via the display, or transmits the identification information of the abnormality symptom cell to an external device via communication.

9. A battery management system comprising the device for detecting an abnormality symptom cell in a battery pack according to any one of claims 1 to 4.

10. While the battery pack is being charged according to a charging profile having a plurality of charging intervals, for each of the first through Nth cells in each charging interval, (a) acquiring first cell voltage time series data in a first half of a charging period; (b) applying a deep learning model to the first cell voltage time series data to determine predicted cell voltage time series data for a second half of the charging interval; (c) acquiring second cell voltage time series data in the latter half part; (d) determining an error between the second cell voltage time series data and the predicted cell voltage time series data; (e) detecting a cell having a relatively larger error than other cells in at least one charging section as a cell showing an abnormality; A method for detecting a malfunctioning cell in a battery pack, comprising:

11. Step (d) is a step of determining a maximum difference between the second cell voltage time series data and the predicted cell voltage time series data for each charging section as an error for each of the first to Nth cells; The step (e) comprises: (e1) determining a first average and a first standard deviation of errors of the first to Nth cells for each charging interval; (e2) determining "(error-first average) / first standard deviation" corresponding to a first normalized value of the error of each of the first to Nth cells for each charging section; and (e3) detecting a cell having the first normalized value greater than a first threshold value in at least one charging interval as a cell exhibiting an abnormality.

12. 11. The method for detecting an abnormality symptom cell in a battery pack according to claim 10, wherein the deep learning model is pre-trained using first cell voltage time series data and second cell voltage time series data measured in the first and second halves of each charging interval for the first to mth learning cells, and is pre-trained to receive input of the first cell voltage time series data and output predicted cell voltage time series data in which an error with the second cell voltage time series data is minimized.

13. The step (d) is a step of determining a maximum difference between the second cell voltage time series data and the predicted cell voltage time series data for each charging section for each of the first to Nth cells as an error; The step (e) comprises: (e1) determining "(error-first average) / first standard deviation" corresponding to a first normalized value of the error of each of the first to Nth cells for each charging section; (e2) detecting a cell having the first normalized value greater than a first threshold value in at least one charging interval as a cell showing an abnormality; The method of claim 12 , wherein the first average and the first standard deviation are values ​​that are predetermined during a learning process of the deep learning model.

14. The battery pack includes first to pth modules connected in series and / or in parallel, determining, for each of the first through pth modules, a second mean and a second standard deviation of the errors of each of the plurality of cells contained within the module; determining "(error-second average) / second standard deviation" corresponding to a second normalized value of the error of each of the first to Nth cells for each charging section; 14. The method for detecting an abnormality symptom cell in a battery pack according to claim 11 or 13, further comprising: detecting a cell in which the first normalized value is greater than a first threshold value and the second normalized value is greater than a second threshold value as an abnormality symptom cell.

15. monitoring whether a relative change behavior between the second cell voltage time series data and the predicted cell voltage time series data corresponds to a change behavior pattern predefined for each type of abnormality symptom with a shift in a charging interval for each of the first to Nth cells; determining a type of abnormality symptom of a cell in which a predefined change behavior pattern is confirmed a reference number of times or more; The method for detecting a cell showing an abnormality in a battery pack according to claim 10 , further comprising:

16. the predefined change behavior pattern is such that the second cell voltage time series data increases faster than the predicted cell voltage time series data in a first half charging section, and the predicted cell voltage time series data increases faster than the second cell voltage time series data in a second half charging section; The method for detecting a cell showing an abnormality in a battery pack according to claim 15, wherein the type of the abnormality is lithium deposition at a negative electrode.

17. a step of recording identification information regarding the detected abnormality symptom cell on a recording medium; outputting a message via a display indicating that a malfunctioning cell has been detected within the battery pack; or The method for detecting an abnormality indicating cell in a battery pack according to claim 10 , further comprising the step of transmitting identification information of the abnormality indicating cell to an external device side via communication.

Images