Electronic device for detecting battery abnormality and operating method thereof

Abstract

An electronic device according to one embodiment disclosed herein can: obtain input data by a detection circuit; obtaining normalized data of the input data by normalizing each of the column vectors of the input data; determining reference data based on the standardized data obtaining; and determining whether each of the M battery cells has a first state abnormality and / or a second state abnormality based on values represented by the items of the row vector for which the reference data is determined. The presence of the first state anomaly can be determined according to a learning-based model, and the presence of the second state anomaly can be determined according to a method other than the learning-based model.

Description

Technical Field

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] This application claims priority to and the benefit of Korean Patent Application No. 10-2022-0113110 filed in the Korean Intellectual Property Office on September 6, 2022, the entire contents of which are incorporated herein by reference. Technical Field

[0004] Embodiments disclosed herein relate to an electronic device for detecting battery abnormality and an operating method of the electronic device. Background Art

[0005] Secondary batteries, which are easily applicable depending on product groups and have electrical characteristics such as high energy density, are generally used not only for portable devices but also for electric vehicles (EVs), hybrid electric vehicles (HEVs), and the like driven by an electric driving source.

[0006] The types of secondary batteries widely used at present include lithium-ion batteries, lithium polymer batteries, nickel-cadmium batteries, nickel-metal hydride batteries, nickel-zinc batteries, etc. The operating voltage of a unit secondary battery cell (i.e., a unit battery cell) is about 2.5 volts to 4.5 volts (V). Therefore, when a higher output voltage is required, a battery pack is formed by connecting a plurality of battery cells in series. In addition, depending on the charge / discharge capacity required for the battery pack, a battery pack can be constructed by connecting a plurality of battery cells in parallel. Therefore, the number of battery cells contained in the battery pack can be set in various ways according to the required output voltage or charge / discharge capacity. Summary of the invention

[0007] Technical issues

[0008] The battery pack may include a plurality of battery cells connected in series and / or in parallel. As a result, even when there is a battery cell in an abnormal state among the plurality of battery cells, it may not be easy to detect the abnormality in the battery cell.

[0009] Therefore, there is a need for a method for detecting a battery cell in an abnormal state among a plurality of battery cells.

[0010] Technical problems of the embodiments disclosed herein are not limited to the above-mentioned technical problems, and other unmentioned technical problems will be clearly understood by a person of ordinary skill in the art through the following description.

[0011] Technical Solution

[0012] An electronic device according to an embodiment disclosed herein includes: a battery module including M battery cells, where M is an integer greater than or equal to 2; a detection circuit configured to obtain status values ​​related to the status of the M corresponding battery cells; and a processor, wherein the processor is configured to: obtain input data through the detection circuit, wherein the input data can be represented as an M×N matrix, the M×N items of the input data represent the status values ​​of the M corresponding battery cells, and N represents the number of time periods used to obtain the status values ​​of the M corresponding battery cells; normalize each of the column vectors of the input data to obtain standardized data of the input data, wherein each of the column vectors includes items obtained within the same unit time; obtain determined reference data based on the standardized data; and determine a first status abnormality and / or a second status abnormality of each of the M battery cells based on the values ​​represented by the items of the corresponding row vectors of the determined reference data, wherein the first status abnormality is determined according to a learning-based model and the second status abnormality is determined according to a scheme other than the learning-based model.

[0013] In an embodiment, the processor may be further configured to: identify a row vector among row vectors of reference data that includes at least one value that is less than or equal to a reference threshold normalized score; obtain an output vector of the identified row vector based on a learning-based model; and determine a first state abnormality of the battery cell corresponding to the identified row vector based on the output vector.

[0014] In an embodiment, the processor can also be configured to: remove the offset from each item of the corresponding row vector of the input data; and normalize each of the column vectors of the input data after removing the offset to obtain standardized data, wherein the offset is set for each of the row vectors of the input data, and the offset is set to the value of the first item among the items of the corresponding row vector of the input data.

[0015] In an embodiment, the processor may be further configured to: smooth the items of the corresponding row vectors of the input data; and normalize each of the column vectors of the smoothed input data to obtain normalized data.

[0016] In an embodiment, the processor can also be configured to: obtain the window average of each of the row vectors of the standardized data as the determined reference data; and determine the first state abnormality and / or the second state abnormality of each of the M battery cells based on the window average of the corresponding row vector of the determined reference data, wherein the window average is the average of the items of the corresponding row vectors of the standardized data contained in different time windows, and the time windows do not overlap with each other.

[0017] In one embodiment, the processor can also be configured to: obtain the change amount of the items of the corresponding row vector of the standardized data as the determined reference data; and determine the first state abnormality and / or the second state abnormality of each of the M battery cells based on the change amount of the items of the corresponding row vector of the determined reference data.

[0018] In an embodiment, the processor can also be configured to: identify the sum of the terms of the corresponding row vectors of the reference data; identify the row vector among the row vectors of the reference data corresponding to the sum that is less than or equal to the threshold sum value; and determine that the battery cell corresponding to the identified row vector has a second state abnormality.

[0019] In one embodiment, the learning-based model may be an autoencoder, and the autoencoder may be trained based on state values ​​of normal battery cells.

[0020] An operating method of an electronic device according to an embodiment disclosed herein includes: obtaining input data through a detection circuit of the electronic device, wherein the input data can be represented as an M×N matrix, the M×N items of the input data represent state values ​​of M corresponding battery cells of the electronic device, and N represents the number of time periods used to obtain the state values ​​of the M corresponding battery cells; normalizing each of the column vectors of the input data to obtain standardized data of the input data, wherein each of the column vectors includes items obtained within the same unit time; obtaining determined reference data based on the standardized data; and determining a first state abnormality and / or a second state abnormality of each of the M battery cells based on the values ​​represented by the items of the corresponding row vectors of the determined reference data, wherein the first state abnormality is determined according to a learning-based model, and the second state abnormality is determined according to a scheme other than the learning-based model.

[0021] In an embodiment, determining a first state abnormality and / or a second state abnormality includes: identifying a row vector among row vectors of reference data that includes at least one value that is less than or equal to a reference threshold normalized score; obtaining an output vector of the identified row vector based on a learning-based model; and determining a first state abnormality of a battery cell corresponding to the identified row vector based on the output vector.

[0022] In an embodiment, obtaining standardized data includes: removing an offset from each item of a corresponding row vector of input data; and normalizing each of the column vectors of the input data after removing the offset to obtain standardized data, wherein an offset is set for each of the row vectors of the input data, and the offset is set to the value of the first item among the items of the corresponding row vector of the input data.

[0023] In an embodiment, obtaining the normalized data includes: smoothing items of corresponding row vectors of the input data; and normalizing each of the column vectors of the smoothed input data to obtain the normalized data.

[0024] In an embodiment, obtaining determined reference data includes: obtaining a window average of each of the row vectors of the standardized data as the determined reference data, and determining the first state abnormality and / or the second state abnormality includes: determining the first state abnormality and / or the second state abnormality of each of the M battery cells based on the window average of the corresponding row vectors of the determined reference data, wherein the window average is the average of the items of the corresponding row vectors of the standardized data contained in different time windows, and the time windows do not overlap with each other.

[0025] In an embodiment, obtaining the determined reference data includes: obtaining the change amount of the items of the corresponding row vector of the standardized data as the determined reference data; and determining the first state abnormality and / or the second state abnormality includes: determining the first state abnormality and / or the second state abnormality of each of the M battery cells based on the change amount of the items of the corresponding row vector of the determined reference data.

[0026] In an embodiment, determining a first state abnormality and / or a second state abnormality includes: identifying the sum of terms of corresponding row vectors of determined reference data; identifying a row vector among the row vectors of determined reference data corresponding to a sum that is less than or equal to a threshold sum value; and determining that a battery cell corresponding to the identified row vector has a second state abnormality.

[0027] Technical Effects

[0028] An electronic device and an operating method thereof according to various embodiments disclosed herein may detect a battery cell in an abnormal state among a plurality of battery cells.

[0029] The effects of the electronic device for detecting battery abnormality and the operating method of the electronic device according to the disclosure of this document are not limited to the above-mentioned effects, and other effects not mentioned will be clearly understood by ordinary technicians in this field according to the disclosure of this document. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 is a block diagram of an electronic device in a network environment according to various embodiments of the present disclosure.

[0031] Figure 2 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure.

[0032] Figure 3 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure.

[0033] Figure 4a The input data is illustrated.

[0034] Figure 4b The smoothed input data is shown.

[0035] Figure 4c This shows the input data after the offset has been removed.

[0036] Figure 4d This illustrates the normalized data of the input data after the offset is removed.

[0037] Figure 4e The normalized data after removing the offset is illustrated. Figure 4f Windowed average data of normalized data are illustrated.

[0038] Figure 4g The variation data of the window average data is shown as an example.

[0039] Figure 5 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure.

[0040] Figure 6 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure.

[0041] Regarding the description of the drawings, like reference numerals may be used to refer to like or related components. DETAILED DESCRIPTION

[0042] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, this description is not intended to limit the present disclosure to a specific embodiment, and it should be understood to include various modifications, equivalents and / or substitutes according to the embodiments of the present disclosure.

[0043] It should be understood that the embodiments of this document and the terms used therein are not intended to limit the technical features set forth herein to specific embodiments, but include various variations, equivalents, or alternatives of the corresponding embodiments. With respect to the description of the drawings, similar reference numerals may be used to refer to similar or related elements. It should be understood that the singular form of a noun corresponding to an item may include one or more things unless the relevant context clearly indicates otherwise.

[0044] As used herein, each of phrases such as “A or B,” “at least one of A and B,” “at least one of A or B,” “A, B or C,” “at least one of A, B and C,” and “at least one of A, B or C” may include any one or all possible combinations of the items listed together in the corresponding one phrase. Terms such as “1st,” “2nd,” “first,” “second,” “A,” “B,” “(a),” or “(b)” may be used to simply distinguish a corresponding component from another component, but do not limit the components in other aspects (such as importance or order) unless otherwise mentioned.

[0045] Here, it should be understood that when an element (e.g., a first element) is referred to as being “connected to another element (e.g., a second element)”, “coupled to another element” or “linked to another element”, or “coupled to another element” or “connected to another element” with or without the term “operably” or “communicatively”, this means that the element can be connected to another element (e.g., the second element) directly (e.g., by wire or wirelessly) or indirectly (e.g., via a third element).

[0046] The method according to various embodiments disclosed herein may include and be provided in a computer program product. The computer program product may be traded as a product between a seller and a buyer. The computer program product may be distributed in the form of a machine-readable storage medium (e.g., a compact disc read-only memory (CD-ROM)), or distributed online (e.g., downloaded or uploaded) through an application store, or distributed directly between two user devices. If distributed online, at least a portion of the computer program product may be temporarily generated or at least temporarily stored in a machine-readable storage medium such as a memory of a manufacturer server, a server of an application store, or a relay server.

[0047] According to the embodiments disclosed herein, each component (e.g., module or program) in the above-mentioned components may include a single entity or multiple entities, and some entities in the multiple entities may be individually arranged in different components. According to various embodiments disclosed herein, one or more of the above-mentioned components may be omitted, or one or more other components may be added. Alternatively or additionally, multiple components (e.g., module or program) may be integrated in a single component. In this case, according to various embodiments, the integrated component may still perform one or more functions of each component in the multiple components in the same or similar manner as performed by a corresponding component in the multiple components before integration. According to embodiments disclosed herein, the operation performed by a module, program or another component may be performed sequentially, in parallel, repeatedly or heuristically, or one or more operations may be performed or omitted in different orders, or one or more other operations may be added.

[0048] Figure 1 is a block diagram of an electronic device 101 according to an embodiment of the present disclosure.

[0049] Reference Figure 1 , the electronic device 101 may include a battery module 110, a sensor circuit 120, a communication circuit 130, a memory 140 and a processor 150. According to an embodiment, Figure 1 At least one component (eg, battery module 110) of the illustrated electronic device 101 may be replaced by another component (eg, a battery cell including a plurality of battery modules 110). Figure 1 At least one component (eg, battery module 110) of the electronic device 101 shown in FIG. 1 may be provided in plurality. According to an embodiment, Figure 1 At least one component of the electronic device 101 shown (eg, the sensor circuit 120 or the communication circuit 130) may be excluded from the electronic device 101. According to an embodiment, the electronic device 101 may further include Figure 1 At least one component other than the components shown (eg, a power supply unit (eg, a motor), a display, an input device, or an output device).

[0050] In an embodiment, the electronic device 101 may be a battery management system (BMS). When the electronic device 101 is implemented with a BMS, the electronic device 101 may provide power of the battery module 110 to an external component (eg, a motor).

[0051] In an embodiment, the electronic device 101 may be a battery swap station (BSS). When the electronic device 101 is implemented with a battery swap station, the electronic device 101 may include a plurality of slots for storing and / or charging a plurality of battery modules 110 .

[0052] In an embodiment, the battery module 110 may supply power to one or more components of the electronic device 101. In an embodiment, the battery module 110 may be attached to or detached from the electronic device 101.

[0053] In an embodiment, the battery module 110 may include one or more battery cells 111, 113, or 115. One or more battery cells 111, 113, or 115 may be included in the battery module 110 in a state of being electrically connected to each other. For example, one or more battery cells 111, 113, and 115 may be connected to each other in series and / or in parallel. According to an embodiment, one or more battery cells 111, 113, or 115 may be included in the battery module 110 in a state of being electrically separated from each other.

[0054] In an embodiment, the sensor circuit 120 may obtain information related to the battery module 110. In an embodiment, the sensor circuit 120 may obtain a value (or information) related to the respective state of one or more battery cells 111, 113, or 115. In an embodiment, the value related to the state may represent one or more values ​​of the voltage, current, resistance, state of charge (SoC), state of health (SoH), or temperature of the battery cell, or a combination thereof. Hereinafter, the value related to the state may be referred to as a "state value".

[0055] In an implementation, the sensor circuit 120 may provide information (eg, status values) of one or more battery cells 111 , 113 , or 115 to the processor 150 .

[0056] In an embodiment, the communication circuit 130 may establish a wired communication channel and / or a wireless communication channel between the electronic device 101 and the external electronic device 102 , and transmit and receive data to and from the external electronic device 102 through the established communication channel.

[0057] In an embodiment, the communication circuit 130 may perform communication based on at least one radio access technology (RAT). In an embodiment, the communication circuit 130 may transmit and receive data to and from the external electronic device 102 by using at least one RAT.

[0058] In implementations, memory 140 may include volatile and / or non-volatile memory.

[0059] In an embodiment, the memory 140 may store data used by at least one component of the electronic device 101 (e.g., the processor 150 or the sensor circuit 120). For example, the data may include software (or instructions related thereto), input data, or output data. In an embodiment, the instructions, when executed by the processor 150, may cause the electronic device 101 to perform the operation defined by the instructions.

[0060] In an embodiment, the processor 150 may execute software to control at least one other component (eg, hardware or software component) of the electronic device 101 connected to the processor 150 , and may process or calculate various data.

[0061] Hereinafter, a method in which the electronic device 101 determines an abnormality of one or more battery cells 111 , 113 , or 115 included in the battery module 110 according to an embodiment of the present disclosure will be described.

[0062] In an embodiment, the electronic device 101 may obtain a state value (e.g., one or more values ​​related to voltage, current, resistance, state of charge, state of health, or temperature) of one or more battery cells 111, 113, or 115. In an embodiment, the electronic device 101 may obtain a state value of one or more battery cells 111, 113, or 115 within a specified time window (e.g., 180 seconds (sec)). In an embodiment, the electronic device 101 may obtain a state value of one or more battery cells 111, 113, or 115 within a specified time period (e.g., 0.1 seconds) of a specified time window. For example, the electronic device 101 may obtain a voltage value of one or more battery cells 111, 113, or 115. In another example, the electronic device 101 may obtain a current value of one or more battery cells 111, 113, or 115.

[0063] In an embodiment, the state values ​​of one or more battery cells 111, 113, or 115 obtained within a specified time period of a specified time window may be designated as “input data”. In an embodiment, the input data may be represented as a matrix (e.g., an M×N matrix) including N values ​​of M battery cells 111, 113, or 115. For example, the input data may include N voltage values ​​of M battery cells 111, 113, or 115. In another example, the input data may include one of N current, resistance, state of charge, state of health, or temperature values ​​of M battery cells 111, 113, or 115. For example, each item of the input data may represent the voltage value of M battery cells 111, 113, or 115. In another example, each item of the input data may represent one of the resistance, state of charge, state of health, or temperature values ​​of M battery cells 111, 113, or 115. Here, M represents the number of battery cells 111, 113, or 115, and N represents the number of time periods for obtaining the state values ​​of battery cells 111, 113, or 115. For example, when the number of battery cells 111 , 113 , or 115 is 100, the designated time window is 180 seconds, and the designated time period is 0.1 seconds, M may be 100 and N may be 1800.

[0064] In an embodiment, the input data D input It can be expressed as formula 1.

[0065] [Formula 1]

[0066]

[0067] In formula 1, the input data D input The corresponding term v 1,1 ,……,v 1,N to v M,1 ,……,v M,NIt can represent the state value of the battery cell 111, 113, or 115. Input data D input The corresponding columns (or column vectors) of may represent the M state values ​​obtained in the same time period (or the sensing time of the sensor circuit 120). input The corresponding rows (or row vectors) of may represent N state values ​​obtained from the battery cells 111, 113, or 115, respectively, within a specified time period. For example, each item v of the first column vector 1,1 to v M,1 It can be the state value of the battery cell 111, 113 or 115 obtained in the first time period. 1,1 to v 1,N It may be a state value obtained from the first battery cell within a specified time period. In an embodiment, the input data D input The corresponding term v 1,1 ,……,v 1,N to v M,1 ,……,v M,N can represent state values ​​of the same type. For example, the corresponding item v 1,1 ,……,v 1,N to v M,1 ,……,v M,N A status value that can represent one type of voltage, current, resistance, state of charge, state of health, or temperature.

[0068] In an embodiment, the electronic device 101 can obtain standardized data based on the input data. In an embodiment, the standardized data of the random matrix data can be obtained based on the standardization of each item of the corresponding column vector of the random matrix data. In an embodiment, standardization can convert the value into a standardized score (or Z score). For example, standardization can be based on the following formula 2.

[0069] [Formula 2]

[0070]

[0071] In formula 2, z i,j Represents the normalized score of the jth entry of the i-th row vector of random matrix data. i,j Represents the value of the jth item of the i-th row vector of random matrix data. m j Represents the average value of the j-th column vector of random matrix data. σ j represents the standard deviation of each term of the j-th column vector of the random matrix data. Here, i represents an integer greater than or equal to 1 and less than or equal to the number of rows of the random matrix data, and j represents an integer greater than or equal to 1 and less than or equal to the number of columns of the random matrix data.

[0072] In an embodiment, the electronic device 101 may determine the first state abnormality and / or the second state abnormality of the battery cell 111, 113, or 115 based on the standardized data. In an embodiment, the electronic device 101 may determine the first state abnormality and / or the second state abnormality of the battery cell 111, 113, or 115 by using the determination reference data based on the standardized data. In an embodiment, the determination reference data may be standardized data. In an embodiment, the determination reference data may be data obtained by performing data processing (e.g., removing offsets, averaging time periods, obtaining variations, and / or cascading) on ​​standardized data. In an embodiment, the first state abnormality may mean a state abnormality in which the instantaneous voltage of the battery cell drops during battery cell charging. In an embodiment, the second state abnormality may mean a state abnormality in which the voltage slope of the battery cell decreases (or is low) during battery cell charging. In an embodiment, the first state abnormality may be determined according to a learning-based model, and the second state abnormality may be determined according to a scheme other than a learning-based model (e.g., a rule-based model).

[0073] Next, an operation of the electronic device 101 determining that the first state of the battery cell 111 , 113 , or 115 is abnormal will be described.

[0074] In an embodiment, the electronic device 101 may identify a row vector including items having values ​​less than or equal to a reference threshold normalized score among each item of a corresponding row vector that determines reference data. According to an embodiment, the electronic device 101 may identify a row vector including items having values ​​greater than or equal to a reference threshold normalized score among each item of a corresponding row vector that determines reference data. According to an embodiment, the electronic device 101 may identify a row vector including items having values ​​exceeding a reference threshold normalized score range among each item of a corresponding row vector that determines reference data. In an embodiment, the reference threshold normalized score may be an experimentally determined score for extracting candidates having a likelihood of being abnormal in the first state. Below, it is illustrated that the electronic device 101 may identify a row vector including items having values ​​less than or equal to a reference threshold normalized score among each item of a corresponding row vector that determines reference data.

[0075] In an embodiment, the electronic device 101 may obtain an output vector of a learning-based model based on a row vector of determined reference data. In an embodiment, the electronic device 101 may obtain an output vector of a row vector of items having values ​​less than or equal to a reference threshold normalization score according to a learning-based model. In an embodiment, the learning-based model may be an autoencoder. In an embodiment, the autoencoder may be an artificial neural network trained to reconstruct determined reference data corresponding to normal behavior. In an embodiment, the autoencoder may be an artificial neural network trained in an unsupervised manner. In an embodiment, the autoencoder may include an encoder and a decoder. In an embodiment, the encoder of the autoencoder may compress the input into a lower dimension, and the decoder of the autoencoder may reconstruct the compressed output of the autoencoder into a previous dimension. Therefore, the input and output of the autoencoder may have the same dimension, and the output of the autoencoder may have a form that includes a reconstruction error (or reconstruction loss) in addition to the input of the autoencoder.

[0076] In an embodiment, the electronic device 101 may determine that the first state of the battery cell is abnormal based on the output vector of the learning-based model. In an embodiment, the electronic device 101 may determine that the first state of the battery cell is abnormal based on the reconstruction error of the autoencoder. In an embodiment, the reconstruction error or the random row vector may be the root mean square (RMS) of a vector obtained by subtracting the output vector of the random row vector from the random row vector.

[0077] In an embodiment, the electronic device 101 may identify a row vector whose reconstruction error exceeds a specified reference reconstruction error among row vectors of the reference data. In an embodiment, the electronic device 101 may determine that the battery cell associated with the row vector exceeding the reference reconstruction error has a first state abnormality. In an embodiment, the reference reconstruction error may be an error determined experimentally in order to identify the first state abnormality.

[0078] Next, an operation of the electronic device 101 determining that the second state of the battery cell 111 , 113 , or 115 is abnormal will be described.

[0079] In an embodiment, the electronic device 101 may determine that the second state of the battery cell 111, 113 or 115 is abnormal based on the sum of the terms of the corresponding row vectors of the reference data. In an embodiment, the electronic device 101 may identify a row vector among the row vectors corresponding to a sum that is less than or equal to a specified reference sum value. According to an embodiment, the electronic device 101 may identify a row vector among the row vectors corresponding to a sum that is greater than or equal to a specified reference sum value. According to an embodiment, the electronic device 101 may identify a row vector among the row vectors corresponding to a sum that exceeds a specified reference sum range. Below, it will be illustrated that the electronic device 101 may identify a row vector among the row vectors corresponding to a sum that is greater than or equal to a specified reference sum value. In an embodiment, the reference sum value may be a value determined experimentally in order to identify the second state abnormality.

[0080] In an implementation, the electronic device 101 may determine that a battery cell associated with a row vector that is less than or equal to a specified reference sum value has a second state abnormality.

[0081] In an embodiment, when a battery cell containing an abnormal state is identified, the electronic device 101 may provide a notification to the user through an output device (or display or communication circuit 130). In an embodiment, the notification provided to the user may include a notification to check the battery cell for abnormality, a notification to replace the battery cell, or a combination thereof.

[0082] Although Figure 1 It is shown that the electronic device 101 obtains the standardized data of the input data and determines the first state abnormality and / or the second state abnormality based on the determined reference data of the standardized data, but this is only an example. According to an embodiment, the electronic device 101 may perform processing to process the input data and determine the first state abnormality and / or the second state abnormality based on the standardized data of the processed input data. In addition, the electronic device 101 may perform processing to process the standardized data of the input data and determine the first state abnormality and / or the second state abnormality based on the processed standardized data (or determined reference data). According to an embodiment, the processing of the input data and / or the processing of the standardized data may be selectively performed.

[0083] Next, an example in which the electronic device 101 processes input data according to an embodiment of the present disclosure will be described.

[0084] In an embodiment, the electronic device 101 may obtain smoothed data (or moving average data) of the input data. In an embodiment, the smoothing of the random matrix data may be performed between the items of the corresponding row vectors of the random matrix data. For example, the smoothing of the random matrix data may be performed based on Formula 3.

[0085] [Formula 3]

[0086]

[0087] In formula 3, vs i,j represents the smoothed value of the jth item of the i-th row vector of the random matrix data. d represents the distance of the item farthest from the jth item among the items used for smoothing. k represents an integer greater than or equal to (jd) and less than or equal to (j+d). αk represents the weight value applied to the kth item. v i,k represents the value of the kth item in the i-th row. The sum of the weight values ​​for smoothing the jth item of the i-th row vector of the random matrix data may be 1. When the weight values ​​for smoothing the jth item of the i-th row vector of the random matrix data are equal to each other, the smoothing may be a moving average.

[0088] In an embodiment, the electronic device 101 may obtain smoothed data (or moving average data) of input data, obtain normalized data of the obtained smoothed data (or moving average data), and determine whether the first state is abnormal and / or the second state is abnormal based on the normalized data.

[0089] In an embodiment, the electronic device 101 can obtain data by removing the offset from the input data. In an embodiment, the offset of the random matrix data can be set for each of the row vectors of the random matrix data. In an embodiment, the offset of the random matrix data can be the value of the first item among the items of the corresponding row vector of the random matrix data. In an embodiment, removing the offset of the random matrix data involves subtracting the offset from each of the row vectors of each item of the row vector of the random matrix data.

[0090] In an embodiment, the electronic device 101 may obtain data by removing an offset from input data, obtain normalized data of the obtained offset-removed data, and determine that the first state is abnormal and / or the second state is abnormal based on the normalized data.

[0091] In an embodiment, the electronic device 101 may sequentially perform offset removal and smoothing on the input data. In an embodiment, the electronic device 101 may obtain normalized data of the data obtained by sequentially performing offset removal and smoothing, and determine that the first state is abnormal and / or the second state is abnormal based on the normalized data. According to the embodiment, offset removal may be performed on the input data first, and then smoothing may be performed, or smoothing may be performed on the input data first, and then offset removal may be performed.

[0092] Below, an example of processing standardized data to obtain determined reference data by the electronic device 101 according to an embodiment of the present disclosure will be described. The standardized data described below may be standardized data of input data, standardized data of smoothed input data, standardized data of input data after removing an offset, or standardized data of smoothed and offset-removed input data.

[0093] In an embodiment, the electronic device 101 may obtain the determination reference data by removing the offset from the normalized data. In an embodiment, the electronic device 101 may determine that the first state is abnormal and / or the second state is abnormal based on the normalized data after the offset is removed (ie, the determination reference data).

[0094] In an embodiment, the electronic device 101 can obtain the period average data of the standardized data as the reference data for determination. In an embodiment, the period average value can involve averaging the items of the corresponding row vectors of the random matrix data at every specified time interval (e.g., 18 seconds). In an embodiment, the time intervals may not overlap with each other. For example, when the specified time interval is 18 seconds, the specified time window is 180 seconds, and the specified time period is 0.1 seconds, the number of columns of the window average data is 10, and the number of columns of the window average data can represent the average of 180 columns of the standardized data.

[0095] In an embodiment, the electronic device 101 may determine that the first state is abnormal and / or the second state is abnormal based on window average data (ie, determine reference data).

[0096] In an embodiment, the electronic device 101 may obtain the variation data of the standardized data as the determination reference data. In an embodiment, the variation may be the difference between adjacent items of each of the row vectors of the random matrix data. For example, the variation of the random item may be a value obtained by subtracting the value of the item before the item from the value of the random item.

[0097] In an embodiment, the electronic device 101 may determine that the first state is abnormal and / or the second state is abnormal based on the variation data (ie, determine the reference data).

[0098] In an embodiment, the electronic device 101 may obtain cascade data of standardized data as determination reference data. In an embodiment, the cascade of random matrix data may cascade other matrix data except the random matrix data in the column direction. In an embodiment, the cascade of random matrix data may involve sequentially cascading random matrix data except the first column, random matrix data of the first column, and random matrix data of the last column. In an embodiment, the electronic device 101 may determine that the first state is abnormal and / or the second state is abnormal based on the cascade data (i.e., determination reference data).

[0099] In an embodiment, the electronic device 101 may obtain the determined reference data by sequentially removing the offset, averaging the window, obtaining the variation, and performing cascading for the standardized data. In an embodiment, the electronic device 101 may determine that the first state is abnormal and / or the second state is abnormal based on the determined reference data obtained by sequentially removing the offset, averaging the window, obtaining the variation, and performing cascading for the standardized data. The order of removing the offset, averaging the window, obtaining the variation, and performing cascading for the standardized data may be set differently according to the embodiment.

[0100] Figure 2 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure. Figure 2 The operations shown can be performed by Figure 1 Executed by electronic device 101.

[0101] Reference Figure 2 , in operation 210, the electronic device 101 may obtain input data. In an embodiment, the input data may include corresponding state values ​​of one or more battery cells 111, 113, or 115 obtained within a specified time period of a specified time window. In an embodiment, the input data may be represented as a matrix (e.g., an M×N matrix) including N values ​​of M battery cells 111, 113, or 115. For example, each item of the input data may represent the voltage value of M battery cells 111, 113, or 115. In another example, each item of the input data may represent one of the resistance, charge state, health state, or temperature value of the M battery cells 111, 113, or 115. Here, M represents the number of battery cells 111, 113, or 115, and N represents the number of time periods for obtaining the state value of the battery cells 111, 113, or 115.

[0102] In operation 220, the electronic device 101 may obtain normalized data. In an embodiment, the normalized data may be normalized data among corresponding items of a column vector of the input data.

[0103] In operation 225, the electronic device 101 may obtain reference data based on the standardized data. In an embodiment, the reference data may be standardized data. In an embodiment, the reference data may be data obtained by performing data processing (e.g., removing an offset, averaging a period, obtaining a variation, and / or performing a cascade) on the standardized data.

[0104] In operation 230, the electronic device 101 may determine that the first state is abnormal. In an embodiment, the electronic device 101 may obtain an output vector based on a learned model based on determining a row vector of reference data. In an embodiment, the electronic device 101 may determine that the first state of the battery cell is abnormal according to the output vector based on the learned model.

[0105] In an embodiment, the electronic device 101 may determine that the first state of the battery cell is abnormal based on the reconstruction error of the autoencoder. In an embodiment, the reconstruction error or the random row vector may be the RMS of a vector obtained by subtracting an output vector of the random row vector from the random row vector.

[0106] In an embodiment, the electronic device 101 may identify a row vector whose reconstruction error exceeds a specified reference reconstruction error among row vectors of the reference data. In an embodiment, the electronic device 101 may determine that the battery cell associated with the row vector exceeding the reference reconstruction error has a first state abnormality.

[0107] In operation 240, the electronic device 101 may determine that the second state is abnormal. In an embodiment, the electronic device 101 may determine that the second state of the battery cell 111, 113, or 115 is abnormal based on determining the sum of corresponding items of the row vector of the reference data. In an embodiment, the electronic device 101 may identify a row vector corresponding to a sum less than or equal to a specified reference sum value among the row vectors. In an embodiment, the electronic device 101 may determine that the battery cell associated with the row vector less than or equal to the specified reference sum value has the second state abnormality.

[0108] Although Figure 2 It is shown that the electronic device 101 obtains the standardized data of the input data and determines the first state abnormality and / or the second state abnormality based on the determined reference data of the standardized data, but this is only an example. According to an embodiment, the electronic device 101 may perform processing to process the input data and determine the first state abnormality and / or the second state abnormality based on the standardized data of the processed input data. In addition, the electronic device 101 may determine the first state abnormality and / or the second state abnormality based on the determined reference data obtained by performing processing to process the standardized data of the input data. According to an embodiment, the processing of the input data and / or the processing of the standardized data may be selectively performed.

[0109] In an embodiment, the method for the electronic device 101 to process input data may include: removing an offset from the input data and smoothing the input data. In an embodiment, the method for the electronic device 101 to process standardized data to obtain determined reference data may include: removing an offset from the standardized data, performing window averaging on the standardized data, and obtaining a variation of the standardized data.

[0110] although Figure 2 230 and 240 are both shown to be performed, but this is only an example. According to an embodiment, operations 230 and 240 may be selectively performed. In some embodiments, neither of operations 230 and 240 may be performed. For example, the electronic device 101 may perform operation 230 to determine that only the first state is abnormal and may not perform operation 240. In another example, the electronic device 101 may perform operation 240 to determine that only the second state is abnormal and may not perform operation 230.

[0111] Figure 3 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure. Figure 4a The input data is illustrated. Figure 4b The smoothed input data is shown. Figure 4c This shows the input data after the offset has been removed. Figure 4d This illustrates the normalized data of the input data after the offset is removed. Figure 4e The normalized data after removing the offset is illustrated. Figure 4f Windowed average data of normalized data are illustrated. Figure 4g The variation data of the window average data is shown as an example. Figure 3 Operations 310, 320, and 330 may include Figure 2 In operation 220. Figure 3 Operations 340, 350, and 360 may include Figure 2 In operation 225. Figure 3 The operations shown can be performed by Figure 1 The electronic device 101 executes. In an embodiment, Figure 4g The change data can be understood as Figure 3 The determined reference data is obtained by the operation.

[0112] Reference Figure 3 , in operation 310 , the electronic device 101 may perform smoothing on the input data.

[0113] In an embodiment, the input data may include corresponding state values ​​of one or more battery cells 111, 113, or 115 obtained within a specified time period of a specified time window. In an embodiment, the input data may be represented as a matrix (e.g., an M×N matrix) including N values ​​of M battery cells 111, 113, or 115. In an embodiment, the state value may be one or more values ​​related to voltage, current, resistance, state of charge, state of health, or temperature.

[0114] Reference Figure 4a, the graph 401 may represent input data 410. The input data 410 may be voltage values ​​of the M battery cells 111, 113, or 115 obtained within a time window of 180 seconds.

[0115] In an embodiment, smoothing of the input data 410 may be performed between the terms of the corresponding row vectors of the input data. For example, smoothing of the input data 410 may be performed based on Equation 3.

[0116] Reference Figure 4b , the graph 402 may represent the smoothed input data 420. Compared to the graph 401, the graph 402 shows that the noise of the input data 420 is reduced by smoothing.

[0117] In operation 320, the electronic device 101 may remove the offset of the smoothed input data 420. In an embodiment, the offset of the smoothed input data 420 may be set for the corresponding row vector of the smoothed input data 420. In an embodiment, the offset of the smoothed input data 420 may be the first item value of the corresponding item of the row vector of the smoothed input data 420. In an embodiment, removing the offset from the smoothed input data 420 may involve subtracting the offset of each of the row vectors of the smoothed input data 420 from the corresponding item of the row vector.

[0118] Reference Figure 4c , the curve 403 may represent the input data 430 after the offset is removed. Compared with the curve 402 , the curve 403 indicates that the initial value of the input data 430 starts from 0.

[0119] In operation 330, the electronic device 101 may perform standardization on the input data 430 after removing the offset. In an embodiment, the standardized data of the input data 430 may be obtained based on the standardization of each item of the corresponding column vector of the input data 430. In an embodiment, the standardization may transform the value into a standardized score (or Z score). For example, the standardization may be based on the following formula 2.

[0120] Reference Figure 4d , the graph 404 may represent the normalized data 440 of the input data 430 .

[0121] In operation 340 , the electronic device 101 may remove an offset of the normalized data 440 .

[0122] In an embodiment, the offset of the normalized data 440 may be set for the corresponding row vector of the normalized data 440. In an embodiment, the offset of the normalized data 440 may be the first item value of the corresponding item of the row vector of the normalized data 440. In an embodiment, removing the offset from the normalized data 440 may involve subtracting the offset of the row vector from the corresponding item of each of the row vectors of the normalized data 440.

[0123] Reference Figure 4e , the graph 405 may represent the normalized data 450 after the offset is removed.

[0124] In operation 350, the electronic device 101 may perform window averaging on the normalized data 450 after the offset is removed. In an embodiment, the window averaging may involve averaging each item of the corresponding row vector of the normalized data 450 at every specified time interval (e.g., 18 seconds). For example, when the specified time interval is 18 seconds, the specified time window is 180 seconds, and the specified time period is 0.1 seconds, then the number of columns of the window average data is 10, and the corresponding column of the window average data may represent the average of 180 columns of the normalized data.

[0125] Reference Figure 4f , the graph 406 may represent window average data 460. Figure 4e Compared with graph 405 , it can be seen that there are 10 windows in graph 406 .

[0126] In operation 360, the electronic device 101 may obtain a variation of the window average data 460. In an embodiment, the variation may be a difference between adjacent items of each of the row vectors of the window average data 460. For example, the variation of a random item may be a value obtained by subtracting the value of the item before the random item from the value of the random item.

[0127] Reference Figure 4g , the curve graph 407 can represent the variation data 470. Figure 4g Compared with the curve graph 407 of FIG. 407 , it can be seen that there are 9 windows in the curve graph 407. This is because there is an item (for example, the last item) in the curve graph 407 for which no change amount is obtained.

[0128] Thereafter, the electronic device 101 may determine that the first state is abnormal and / or the second state is abnormal of the battery cell 111 , 113 , or 115 based on the variation data 470 .

[0129] Figure 5 is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure. Figure 5 The operations can include Figure 2 In operation 230. Figure 5The operations shown can be performed by Figure 1 Executed by electronic device 101.

[0130] You can perform this for each row vector Figure 5 operation.

[0131] Reference Figure 5 In operation 510, the electronic device 101 may determine whether the row vector includes an item that is less than or equal to a reference threshold normalized score. In an embodiment, the electronic device 101 may identify a row vector including items having a value less than or equal to a reference threshold normalized score among items of a corresponding row vector of the determined reference data. According to an embodiment, the row vector may be Figure 4f The row vector of the variation data 470. According to an embodiment, the row vector may be a row vector of data of the cascaded variation data 470. In an embodiment, the row vector may be a row vector of data of the variation data 470 except the first column, the variation data 470 of the first column, and the variation data 470 of the last column sequentially cascaded.

[0132] In operation 520, the electronic device 101 may obtain an output vector of the row vector according to a learning-based model. In an embodiment, the learning-based model may be an autoencoder. In an embodiment, the autoencoder may be an artificial neural network trained to reconstruct determined reference data corresponding to normal behavior. In an embodiment, the output vector may have the same dimension as the row vector and include a reconstruction error (or reconstruction loss).

[0133] In operation 530, the electronic device 101 may determine whether a reconstruction error between the row vector and the output vector exceeds a reference reconstruction error. In an embodiment, the reconstruction error between the row vector and the output vector may be an RMS of a vector obtained by subtracting the row vector from the output vector.

[0134] In an embodiment, when the reconstruction error exceeds the reference reconstruction error, the electronic device 101 may perform operation 540. In an embodiment, when the reconstruction error does not exceed the reference reconstruction error, the electronic device 101 may perform operation 550.

[0135] In operation 540, the electronic device 101 may determine that it has a first state abnormality. In an embodiment, the electronic device 101 may determine that a battery cell corresponding to a row vector has a first state abnormality.

[0136] In operation 550, the electronic device 101 may determine that it does not have the first state abnormality. In an embodiment, the electronic device 101 may determine that the battery cell corresponding to the row vector does not have the first state abnormality.

[0137] Figure 6is a flowchart illustrating the operation of the electronic device according to an embodiment of the present disclosure. Figure 6 The operations can include Figure 2 In operation 240. Figure 6 The operations shown can be performed by Figure 1 Executed by electronic device 101.

[0138] You can perform this for each row vector Figure 6 operation.

[0139] Reference Figure 6 In operation 610, the electronic device 101 may identify the sum of the terms of the row vector. According to an embodiment, the row vector may be Figure 4f The row vector of the variation data 470.

[0140] In operation 620 , the electronic device 101 may determine whether the sum is less than or equal to a reference sum value.

[0141] In an embodiment, when the sum is less than or equal to the reference sum value, the electronic device 101 may perform operation 630. In an embodiment, when the sum is not less than or equal to the reference sum value, the electronic device 101 may perform operation 640.

[0142] In operation 630, the electronic device 101 may determine that there is a second state abnormality. In an embodiment, the electronic device 101 may determine that a battery cell corresponding to a row vector has a second state abnormality.

[0143] In operation 640, the electronic device 101 may determine that it does not have the second state abnormality. In an embodiment, the electronic device 101 may determine that the battery cell corresponding to the row vector does not have the second state abnormality.

Claims

1. An electronic device, comprising: A battery module, the battery module comprising M battery cells, where M is an integer greater than or equal to 2; a detection circuit configured to obtain a status value related to the status of the M corresponding battery cells; as well as processor, Wherein, the processor is configured to: Obtaining input data through the detection circuit, wherein the input data can be represented as an M×N matrix, the M×N items of the input data represent the state values ​​of the M corresponding battery cells, and N represents the number of time periods for obtaining the state values ​​of the M corresponding battery cells; Normalizing each of the column vectors of the input data to obtain normalized data of the input data, wherein each of the column vectors includes items obtained in the same unit time; Obtaining determined reference data based on the standardized data; and determining, based on the values ​​represented by the entries of the corresponding row vectors of the determination reference data, that the first state is abnormal and / or the second state is abnormal for each of the M battery cells; Wherein, the first state anomaly is determined according to a learning-based model, and the second state anomaly is determined according to a scheme other than the learning-based model.

2. The electronic device according to claim 1, wherein: The processor is further configured to: identifying a row vector among the row vectors of the determined reference data that includes at least one value that is less than or equal to a reference threshold normalized score; obtaining an output vector of identified row vectors according to the learning-based model; as well as It is determined that the first state of the battery cell corresponding to the identified row vector is abnormal based on the output vector.

3. The electronic device according to claim 1, wherein: The processor is further configured to: removing an offset from each entry of a corresponding row vector of the input data; and Normalizing each of the column vectors of the input data after the offset is removed to obtain the normalized data, The offset is set for each of the row vectors of the input data, and the offset is set to the value of the first item among the items of the corresponding row vector of the input data.

4. The electronic device according to claim 1, wherein: The processor is further configured to: smoothing the entries of the corresponding row vectors of the input data; and Each of the column vectors of the smoothed input data is normalized to obtain the normalized data.

5. The electronic device according to claim 1, wherein: The processor is further configured to: Obtaining a window average value of each of the row vectors of the standardized data as the determined reference data; and determining the first state abnormality and / or the second state abnormality of each of the M battery cells based on the window average of the corresponding row vector of the determined reference data, The window average is the average of items of corresponding row vectors of standardized data contained in different time windows, and the time windows do not overlap with each other.

6. The electronic device according to claim 1, wherein: The processor is further configured to: Obtaining a change in an item of a corresponding row vector of the standardized data as the determined reference data; and The first state abnormality and / or the second state abnormality of each of the M battery cells is determined based on a change amount of an item of a corresponding row vector of the determination reference data.

7. The electronic device according to claim 1, wherein: The processor is further configured to: identifying a sum of entries of a corresponding row vector of the determined reference data; identifying a row vector among the row vectors of the determined reference data corresponding to a sum that is less than or equal to a threshold sum value; as well as It is determined that the battery cell corresponding to the identified row vector has the second state abnormality.

8. The electronic device according to claim 1, wherein: The learning-based model is an autoencoder, and the autoencoder is trained based on state values ​​of normal battery cells.

9. A method for operating an electronic device, the method comprising the following steps: Obtaining input data through a detection circuit of the electronic device, wherein the input data can be represented as an M×N matrix, the M×N items of the input data represent state values ​​of M corresponding battery cells of the electronic device, and N represents the number of time periods for obtaining the state values ​​of the M corresponding battery cells; Normalizing each of the column vectors of the input data to obtain normalized data of the input data, wherein each of the column vectors includes items obtained in the same unit time; Obtaining determined reference data based on the standardized data; and determining a first state abnormality and / or a second state abnormality for each of the M battery cells based on the values ​​represented by the entries of the corresponding row vectors of the determination reference data, Wherein, the first state anomaly is determined according to a learning-based model, and the second state anomaly is determined according to a scheme other than the learning-based model.

10. The operating method according to claim 9, wherein: The step of determining that the first state is abnormal and / or the second state is abnormal comprises the following steps: identifying a row vector among the row vectors of the determined reference data that includes at least one value that is less than or equal to a reference threshold normalized score; obtaining an output vector of identified row vectors according to the learning-based model; and It is determined that the first state of the battery cell corresponding to the identified row vector is abnormal based on the output vector.

11. The operating method according to claim 9, wherein: The step of obtaining the standardized data comprises the following steps: removing an offset from each entry of a corresponding row vector of the input data; and Normalizing each of the column vectors of the input data after the offset is removed to obtain the normalized data, The offset is set for each of the row vectors of the input data, and the offset is set to the value of the first item among the items of the corresponding row vector of the input data.

12. The operating method according to claim 9, wherein: The step of obtaining the standardized data comprises the following steps: smoothing the entries of the corresponding row vectors of the input data; and Each of the column vectors of the smoothed input data is normalized to obtain the normalized data.

13. The operating method according to claim 9, wherein: The step of obtaining the determined reference data comprises the following steps: obtaining a window average value of each of the row vectors of the standardized data as the determined reference data, and The step of determining the first state abnormality and / or the second state abnormality comprises the following steps: determining the first state abnormality and / or the second state abnormality of each battery cell in the M battery cells based on the window average value of the corresponding row vector of the determination reference data, The window average is the average of items of corresponding row vectors of standardized data contained in different time windows, and the time windows do not overlap with each other.

14. The operating method according to claim 9, wherein: The step of obtaining the determined reference data comprises the following steps: obtaining the variation of the items of the corresponding row vector of the standardized data as the determined reference data; and The determining of the first state abnormality and / or the second state abnormality comprises the following steps: determining the first state abnormality and / or the second state abnormality of each of the M battery cells based on the variation of the items of the corresponding row vector of the determination reference data.

15. The operating method according to claim 9, wherein: The step of determining that the first state is abnormal and / or the second state is abnormal comprises the following steps: identifying a sum of entries of a corresponding row vector of the determined reference data; identifying a row vector among the row vectors of the determination reference data corresponding to a sum that is less than or equal to a threshold sum value; and It is determined that the battery cell corresponding to the identified row vector has the second state abnormality.

Images