Battery diagnostic apparatus and method thereof

Abstract

A battery diagnostic apparatus, according to one embodiment of the present document, comprises: a memory for storing at least one instruction; and at least one processor executing the at least one instruction. The at least one processor may: identify a first input voltage profile including a voltage profile measured by supplying a current having a specified waveform to a battery cell; input the first input voltage profile to a first model for encoding and then restoring same; and diagnose the state of the battery cell on the basis of a feature vector identified on the basis of the first model.

Description

Battery diagnostic device and method thereof

[0001] Cross-citation with related applications

[0002] This application claims the benefit of priority to Korean Patent Application No. 10-2024-0087431, filed July 3, 2024, and Korean Patent Application No. 10-2025-0078638, filed June 16, 2025, the entire contents of which are incorporated herein by reference.

[0003] Technology field

[0004] The embodiments disclosed in this document relate to a battery diagnostic device and method thereof.

[0005] Recently, research and development on secondary batteries has been actively underway. Here, secondary batteries are defined as rechargeable and dischargeable batteries, encompassing both conventional Ni / Cd and Ni / MH batteries, as well as more recent lithium-ion batteries. Recently, their use has expanded to include power sources for electric vehicles, attracting attention as a next-generation energy storage medium.

[0006] With the proliferation of various electronic devices driven by the Fourth Industrial Revolution, battery usage is rapidly increasing. Batteries are emerging as an essential energy source in various fields, and this is fueling the growing importance of technology to diagnose battery conditions to improve battery performance and reliability.

[0007] Recent advancements in artificial intelligence models have led to the development of battery diagnostic devices that utilize battery data. Furthermore, multiple trained models can be used as tools to diagnose the condition of battery cells.

[0008] According to the embodiments disclosed in this document, an object is to provide a battery diagnosis device and method for reducing noise generation rate by using a reduced dimension feature vector identified in a first model that is restored after encoding.

[0009] According to the embodiments disclosed in this document, an object is to provide a battery diagnosis device and method for reducing the overfitting occurrence rate in a second model, which is a regression model, by reducing the noise occurrence rate.

[0010] According to the embodiments disclosed in this document, it is an object to provide a battery diagnosis device and method for improving the diagnosis accuracy of a battery cell state based on a feature vector extracted through a first model in which a pattern learning process is performed.

[0011] According to the embodiments disclosed in this document, an object is to provide a battery diagnosis device and method for reducing training stability degradation due to high-dimensional data processing.

[0012] According to embodiments disclosed in this document, it is intended to provide a battery diagnosis device and method for improving training efficiency even if the number of input data is less than a specified number.

[0013] The technical challenges of this document are not limited to the technical challenges mentioned above, and other technical challenges not mentioned will be clearly understood by those skilled in the art from the descriptions below.

[0014] A battery diagnostic device according to one embodiment of the present document may include a memory storing at least one instruction, and at least one processor executing the at least one instruction.

[0015] According to one embodiment, the at least one processor identifies a first input voltage profile including a voltage profile measured by supplying a current having a waveform specified to a battery cell, inputs the first input voltage profile into a first model that encodes and then restores the first input voltage profile, and diagnoses the state of the battery cell based on a feature vector identified based on the first model.

[0016] According to one embodiment, the at least one processor can identify the feature vector based on a latent vector that is a processing result in a bottleneck layer included in the first model.

[0017] In one embodiment, the at least one processor can diagnose the state of the battery cell based on at least one of at least a portion of the feature vector, at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof.

[0018] According to one embodiment, the at least one processor may diagnose the state of the battery cell based on at least one of at least a portion of the first input voltage profile, at least a portion of the restored voltage profile reconstructed based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof and at least a portion of the feature vector, based on all components of the feature vector being within the predetermined range, or may diagnose the state of the battery cell based on at least one of at least a portion of the first input voltage profile, at least a portion of the restored voltage profile reconstructed based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof, based on at least one of the components of the feature vector being outside the predetermined range.

[0019] In one embodiment, the at least one processor may input at least some of the feature vectors into a second model that outputs a state of the battery cell.

[0020] In one embodiment, the at least one processor is capable of supplying the current to the battery cell within a specified time period.

[0021] According to one embodiment, the state of the battery may include a state of health (SOH).

[0022] According to one embodiment, the first model may include a long short term memory autoencoder (LSTM AE) model.

[0023] A battery diagnosis method according to another embodiment of the present document may include an operation of identifying a first input voltage profile including a voltage profile measured by supplying a current having a specified waveform to a battery cell, an operation of inputting the first input voltage profile into a first model that encodes and then restores the first input voltage profile, and an operation of diagnosing a state of the battery cell based on a feature vector identified based on the first model.

[0024] According to one embodiment, the operation of diagnosing the state of the battery cell based on the feature vector identified based on the first model may include an operation of identifying the feature vector based on a latent vector which is a result of processing in a bottleneck layer included in the first model.

[0025] According to one embodiment, the operation of diagnosing the state of the battery cell based on the feature vector identified based on the first model may include the operation of diagnosing the state of the battery cell based on at least one of at least a portion of the feature vector, at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof.

[0026] According to one embodiment, the operation of identifying the feature vector based on the latent vector, which is a result of processing in the bottleneck layer included in the first model, may include the operation of diagnosing the state of the battery cell based on at least one of at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof and at least a portion of the feature vector, based on all components of the feature vector being included in the predetermined range, or the operation of diagnosing the state of the battery cell based on at least one of at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof, based on at least one of the components of the feature vector being outside the predetermined range.

[0027] According to one embodiment, the operation of diagnosing the state of the battery cell based on the feature vector identified based on the first model may include the operation of inputting at least a portion of the feature vector into a second model that outputs the state of the battery cell.

[0028] In one embodiment, the operation of identifying the first input voltage profile including the voltage profile measured by supplying a current having the specified waveform to the battery cell may include supplying the current to the battery cell within a specified time period.

[0029] According to one embodiment, the state of the battery may include a state of health (SOH).

[0030] According to one embodiment, the first model may include a long short term memory autoencoder (LSTM AE) model.

[0031] According to another embodiment of the present document, a computer-readable recording medium can record a program for performing the battery diagnosis method on a computer.

[0032] A battery diagnostic device according to another embodiment of the present document may include a memory storing at least one instruction, and at least one processor executing the at least one instruction.

[0033] According to one embodiment, the at least one processor identifies a first input voltage profile by supplying a current having a first waveform to a battery cell, or identifies a second input voltage profile based on a current having a second waveform output from the battery cell, inputs the first input voltage profile or the second input voltage profile into a first model that encodes and then restores the first input voltage profile or the second input voltage profile, and diagnoses a state of the battery cell based on a feature vector identified based on the first model.

[0034] According to one embodiment, the at least one processor can identify the first input voltage profile when the state of charge of the battery cell falls within a predetermined state range, and can identify the second input voltage profile when the state of charge of the battery cell does not fall within the predetermined state range.

[0035] In one embodiment, the predetermined state range may include less than a predetermined state value.

[0036] According to one embodiment, the at least one processor can identify the feature vector based on at least some of the components of a latent vector that is a result of processing in a bottleneck layer included in the first model.

[0037] In one embodiment, the at least one processor may diagnose the state of the battery cell based on at least one of at least a portion of the feature vector, at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof, or may diagnose the state of the battery cell based on at least one of at least a portion of the feature vector, at least a portion of the second input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the second input voltage profile, or any combination thereof.

[0038] In one embodiment, the at least one processor may input at least some of the feature vectors into a second model that outputs a state of the battery cell.

[0039] According to one embodiment, the first model may include a long short term memory autoencoder (LSTM AE) model.

[0040] Another embodiment of the present document provides a battery diagnosis method, which may include: identifying a first input voltage profile by supplying a current having a first waveform to a battery cell, or identifying a second input voltage profile based on a current having a second waveform output from the battery cell; inputting the first input voltage profile or the second input voltage profile into a first model that encodes and then restores the first input voltage profile or the second input voltage profile; and diagnosing a state of the battery cell based on a feature vector identified based on the first model.

[0041] According to one embodiment, the operation of identifying the first input voltage profile by supplying a current having the first waveform to the battery cell or identifying the second input voltage profile based on a current having the second waveform output from the battery cell may include the operation of identifying the first input voltage profile when the state of charge of the battery cell is within a predetermined state range, and the operation of identifying the second input voltage profile when the state of charge of the battery cell is not within the predetermined state range.

[0042] In one embodiment, the predetermined state range may include less than a predetermined state value.

[0043] According to one embodiment, the operation of diagnosing the state of the battery cell based on the feature vector identified based on the first model may include an operation of identifying the feature vector based on at least some of the components of a latent vector that is a result of processing in a bottleneck layer included in the first model.

[0044] According to one embodiment, the operation of identifying the feature vector based on at least some of the components of the latent vector, which is a result of processing in the bottleneck layer included in the first model, may include the operation of diagnosing the state of the battery cell based on at least one of at least some of the feature vector, at least some of the first input voltage profile, at least some of the restored voltage profile restored based on the first model, at least some of the differential values ​​of the first input voltage profile, or any combination thereof, or the operation of diagnosing the state of the battery cell based on at least one of at least some of the feature vector, at least some of the second input voltage profile, at least some of the restored voltage profile restored based on the first model, at least some of the differential values ​​of the second input voltage profile, or any combination thereof.

[0045] According to one embodiment, the operation of diagnosing the state of the battery cell based on the feature vector identified based on the first model may include the operation of inputting at least a portion of the feature vector into a second model that outputs the state of the battery cell.

[0046] According to one embodiment, the first model may include a long short term memory autoencoder (LSTM AE) model.

[0047] According to another embodiment of the present document, a computer-readable recording medium can record a program for performing the battery diagnosis method on a computer.

[0048] This technology can reduce the noise generation rate by using a reduced dimension feature vector identified in a first model that is restored after encoding.

[0049] In addition, the present technology can reduce the overfitting rate in the second model, which is a regression model, by reducing the noise rate.

[0050] In addition, the present technology can improve the diagnostic accuracy of the status of a battery cell based on a feature vector extracted through a first model on which a pattern learning process has been performed.

[0051] Additionally, the present technology, according to the embodiments disclosed in this document, can reduce the deterioration of training stability due to high-dimensional data processing.

[0052] In addition, the present technology can improve training efficiency even if the number of input data is less than a specified number.

[0053] In addition, various effects may be provided, either directly or indirectly, through this document.

[0054] FIG. 1 is a block diagram showing a battery pack in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0055] FIG. 2 is a block diagram showing the configuration of a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0056] FIG. 3 illustrates an example of a first model restored after encoding in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0057] FIG. 4 illustrates an example of an operation in which a voltage profile is processed through a first model that is restored after encoding in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0058] FIG. 5 illustrates an example of an operation of diagnosing life information of a battery cell according to the degree of charge of the battery cell in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0059] FIG. 6 illustrates an example of an operation in which the type of data input to a second model, which is a regression model, is changed according to a component of a latent vector in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0060] FIG. 7 illustrates an example of an operation in which the value of data input to a second model, which is a regression model, is changed according to a component of a latent vector in a battery diagnosis device and a battery diagnosis method according to an embodiment of the present document.

[0061] FIG. 8 illustrates an example of an operation in which a method of deriving a state of a battery cell from data output from a second model, which is a regression model, is changed according to a component of a latent vector in a battery diagnosis device and a battery diagnosis method according to an embodiment of the present document.

[0062] FIG. 9 illustrates an example of an operation for diagnosing the status of a battery cell in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0063] FIG. 10 illustrates an example of an operation of diagnosing the state of a battery cell based on a voltage profile of the battery cell obtained in a manner determined according to the degree of charge of the battery cell, in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0064] FIG. 11 is a block diagram showing the hardware configuration of a computing system that performs a battery diagnosis method in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0065] Hereinafter, some embodiments disclosed in this document are described with reference to the accompanying drawings, which illustrate various embodiments of this document. However, this is not intended to limit the present technology to specific embodiments, and it should be understood that various modifications, equivalents, and / or alternatives of the embodiments of this technology are included.

[0066] When assigning reference numerals to components in each drawing, it should be noted that identical components are assigned the same numerals whenever possible, even if they are shown in different drawings. Furthermore, when describing various embodiments disclosed in this document, if a detailed description of a related known configuration or function is deemed to hinder understanding of the embodiments of the present invention, the detailed description will be omitted. The singular form of a noun corresponding to an item may include one or more items, unless the context clearly indicates otherwise.

[0067] In describing the components of the embodiments of this document, terms such as first, second, A, B, (a), (b), etc. may be used. These terms are only intended to distinguish the components from other components, and the nature, order, or sequence of the components may not be limited by the terms. In addition, unless otherwise defined, all terms used herein, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art to which the embodiments disclosed in this document belong. Terms defined in commonly used dictionaries should be interpreted as having a meaning consistent with the meaning they have in the context of the relevant technology, and shall not be interpreted in an idealized or overly formal sense unless explicitly defined in this application.

[0068] In addition, in the present disclosure, expressions such as "more than" or "less than" may be used to determine whether a specific condition is satisfied or fulfilled. However, this is merely a description for expressing an example and does not exclude descriptions such as "more than" or "less than." Conditions described as "more than" may be replaced with "more than," conditions described as "less than," and conditions described as "more than and less than" may be replaced with "more than and less than." In addition, hereinafter, "A" to "B" mean at least one of the elements from A (including A) to B (including B).

[0069] In this document, each of the phrases "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 of the items listed together in that phrase, or all possible combinations thereof.

[0070] In this document, when a component (e.g., a first component) is referred to as being “connected,” “coupled,” or “connected,” with or without the terms “functionally” or “communicatively,” or is referred to as being “coupled” or “connected,” it means that the component can be connected to the other component directly (e.g., wired), wirelessly, or through a third component.

[0071] According to one embodiment, the method according to the various embodiments disclosed in the present document may be provided as included 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., compact disc read-only memory (CD-ROM)), or may be distributed online (e.g., downloaded or uploaded) through an application store or directly between two user devices. In the case of online distribution, at least a portion of the computer program product may be temporarily stored or temporarily generated in a machine-readable storage medium, such as the memory of a manufacturer's server, an application store's server, or an intermediary server.

[0072] According to various embodiments, each component (e.g., a module or a program) of the described components may include one or more entities, and some of the entities may be separated and placed in other components. According to various embodiments, one or more components or operations of the aforementioned components may be omitted, or one or more other components or operations may be added. Alternatively or additionally, a plurality of components (e.g., a module or a program) may be integrated into a single component. In such a case, the integrated component may perform one or more functions of each of the plurality of components identically or similarly to those performed by the corresponding component among the plurality of components prior to integration. According to various embodiments, the operations performed by a module, program, or other component may be executed sequentially, in parallel, iteratively, or heuristically, or one or more of the operations may be executed in a different order, omitted, or one or more other operations may be added.

[0073] Hereinafter, embodiments of the present document will be described in detail with reference to FIGS. 1 to 11.

[0074] FIG. 1 is a block diagram showing a battery pack in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0075] Referring to FIG. 1, a battery pack (1) may include a battery unit (12), a sensor unit (14), a switching unit (16), and a battery management system (BMS) (20). At this time, the battery pack (1) may be equipped with a plurality of battery units (12), sensor units (14), switching units (16), and battery management systems (20).

[0076] According to one embodiment, the battery unit (12) can supply power to a target device (not shown). To this end, the battery unit (12) can be electrically connected to the target device. Here, the target device can include an electrical, electronic, or mechanical device that operates by receiving power from the battery pack (1). For example, the target device can be, but is not limited to, an electric vehicle (EV) or an energy storage system (ESS).

[0077] According to one embodiment, the battery unit (12) may include at least one battery cell (10) that can be charged and discharged. Here, the battery cell (10) may be a basic unit of a battery cell that can charge and discharge electric energy and use it. For example, the battery cell (10) may be a lithium-ion (Li-ion) battery, a lithium-ion polymer (Li-ion polymer) battery, a nickel-cadmium (Ni-Cd) battery, a nickel-metal hydride (Ni-MH) battery, etc., but may not be limited thereto.

[0078] According to one embodiment, a plurality of battery units (12) may be connected in series or parallel. For example, the battery unit (12) may be a battery module, a battery bank, or a collection of battery cells (cell-to-pack structure).

[0079] According to one embodiment, the sensor unit (14) can obtain information related to the battery unit (12). According to one embodiment, the sensor unit (14) can obtain values ​​(or information) related to the state of each of the battery unit (12) or battery cells (10). In one embodiment, the values ​​related to the state may include one or more values ​​for voltage, current, resistance, state of charge (SOC), state of health (SOH), or temperature of the battery cell, or a combination thereof.

[0080] According to one embodiment, the sensor unit (14) can provide information on each of the plurality of battery units (12) to the battery management system (20).

[0081] According to one embodiment, the switching unit (16) may include a device for controlling the current flow for charging or discharging the battery unit (12). For example, the switching unit (16) may include at least one relay and / or magnetic contactor, etc., depending on the specifications of the battery pack (1).

[0082] According to one embodiment, a battery management system (BMS) (20) may monitor voltage, current, temperature, etc. of the battery pack (1) to control or manage the battery pack (1) to prevent overcharge, overdischarge, etc. For example, the battery management system (20) may include a plurality of terminals as an interface for receiving values ​​measured from the various parameters described above, and a circuit connected to these terminals to process the input values. In addition, the battery management system (20) may control the sensor unit (14) and / or the switching unit (16). For example, the battery management system (20) may be connected to a plurality of battery units (12) to monitor the status of each of the plurality of battery units (12) and control ON / OFF of a relay or a contactor, etc.

[0083] According to one embodiment, the operation of the battery management system (20) may be performed by a battery management system (BMS) in the vehicle, as well as by various devices such as a server, cloud, charger, or discharger.

[0084] The upper controller (2) can transmit control signals for multiple battery units (12) to the battery management system (20). Accordingly, the battery management system (20) can be controlled for operation based on signals received from the upper controller (2).

[0085] According to one embodiment, the battery management system (20) may include the battery diagnostic device (201) of FIG. 2. According to another embodiment, the battery management system (20) may be a different system from the battery diagnostic device (201) of FIG. 2. That is, the diagnostic device (201) of FIG. 2 may be included in the battery pack (1) or may be configured as another device external to the battery pack (1). For convenience of explanation, the following description will be made on the assumption that the battery diagnostic device (201) is configured as another device external to the battery pack (1). In addition, the operation of the battery diagnostic device (201) below may be performed by an in-vehicle BMS (battery management system), as well as by various devices such as a server, a cloud, a charger, or a discharger.

[0086] FIG. 2 is a block diagram showing the configuration of a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0087] Referring to FIG. 2, the battery diagnostic device (201) may include a memory (203) and at least one processor (205). The memory (203) may store at least one instruction. The at least one processor (205) may execute at least one instruction.

[0088] According to one embodiment, at least one processor (205) of the battery diagnostic device (201) can identify a voltage profile of a battery cell by supplying a current having a first waveform to the battery cell, or identify a voltage profile of the battery cell based on a second waveform output from the battery cell, and diagnose a state of the battery cell (e.g., life information of the battery cell) based on the identified voltage profile.

[0089] According to one embodiment, at least one processor (205) of the battery diagnostic device (201) can diagnose the state of a battery cell based on values ​​obtained based on an input voltage profile and a first model to improve diagnostic accuracy. In this case, the first model can encode and then restore the input voltage profile.

[0090] In one embodiment, a first model that encodes and then restores may be used to improve diagnostic accuracy for battery cell status (e.g., battery cell life information). The first model may be referred to as a long short-term memory-autoencoder (LSTM-AE), but the embodiments of this document may not be limited thereto.

[0091] According to one embodiment, at least one processor (205) of the battery diagnostic device (201) can input the identified voltage profile as an input voltage profile into the first model.

[0092] According to one embodiment, at least one processor (205) of the battery diagnostic device (201) can input at least one of at least some of the feature vectors identified based on the first model, at least some of the input voltage profiles, at least some of the output voltage profiles restored based on the first model, at least some of the differential values ​​of the input voltage profile, or any combination thereof, into a second model, which is a regression model.

[0093] According to one embodiment, at least one processor (205) of the battery diagnostic device (201) can diagnose the state of a battery cell (e.g., life information of the battery cell) based on a value output from a second model, which is a regression model. Specific details are described below with reference to FIGS. 3 and 4.

[0094] The life information of a battery cell may include the State of Health (SOH) of the battery cell, which represents the ratio of the current maximum charge capacity to the initial design capacity of the battery cell, and may be expressed as a value between 0 and 1, or a value between 0% and 100%. Measurement of the SOH may use a cycle count-based evaluation, an internal resistance change analysis, and / or a capacity degradation rate calculation method, but embodiments of the present document may not be limited thereto.

[0095] FIG. 3 illustrates an example of a first model in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0096] Referring to FIG. 3, the first model may include an autoencoder using an LSTM, but the embodiments of the present document may not be limited thereto. The first model may receive input voltage profiles, which are time-series data, extract feature vectors from the data, and reconstruct the input voltage profiles, which are input data, based on the feature vectors. Here, LSTM is one type of artificial neural network and can be defined as a structure that improves a recurrent neural network (RNN) to reflect long-term features.

[0097] In one embodiment, the first model may include an LSTM encoder, a bottleneck layer, and an LSTM decoder. However, the embodiments of this document are not limited thereto, and a first model that does not use an LSTM may include an encoder, a bottleneck layer, and a decoder.

[0098] In one embodiment, the first model may obtain an encoded feature vector from input data (e.g., an input voltage profile) input through an LSTM encoder through a bottleneck layer. The feature vector may be low-dimensional data compared to the input data.

[0099] According to one embodiment, the first model can obtain restored data (e.g., restored voltage profile) by decoding and restoring a compressed feature vector of input data.

[0100] According to one embodiment, at least one processor (205) of the battery diagnosis device (201) can diagnose the state of a battery cell according to an output value of the second model by inputting at least one of input data, restoration data, feature vector, or any combination thereof into a second model that is a regression model.

[0101] The second model can include various types of regression models, including linear regression models and kernel regression models.

[0102] According to one embodiment, a battery diagnostic device (201) may be configured with a data collection unit, a feature extraction unit, and a life prediction unit. The feature extraction unit may be configured with a first model. The life prediction unit may be configured with a second model. The second model may identify the state of the battery (e.g., battery life information) based on parameters identified by the feature extraction unit.

[0103] FIG. 4 illustrates an example of an operation in which a voltage profile is processed through a first model in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0104] Referring to FIG. 4, a set (401) of input voltage profiles may represent at least one voltage profile measured from each of at least one battery cell included in a battery unit. An input voltage profile (403) included in the set (401) of input voltage profiles may include a voltage profile of a first battery cell among the first to n-th battery cells included in the battery unit. The input voltage profile (403) may include a vector representing a voltage of the first battery cell measured over time. For example, the input voltage profile (403) may include a vector representing a voltage measured at a specified time interval (e.g., at an interval of about 0.1 seconds) for a specified time (e.g., about 60 seconds).

[0105] The set of feature vectors (405) may represent at least one latent vector, which is a result of processing in a bottleneck layer for each of at least one input voltage profile included in the set of input voltage profiles (401). The feature vector (407) may be identified based on at least some of the components of the latent vector of the first model for the input voltage profile (403). The number of components of the feature vector (e.g., the feature vector (407)) may be smaller than the number of components of the input voltage profile (e.g., the input voltage profile (403)).

[0106] The set of restored voltage profiles (409) may represent at least one restored voltage profile identified by decoding each of at least one feature vector included in the set of feature vectors (405). The restored voltage profile (411) may represent a restored voltage profile identified by decoding the latent vector (407).

[0107] According to one embodiment, the voltage profile of the first battery cell may include a first input voltage profile identified by supplying a current having a first waveform to the battery cell within a specified time period, or a second input voltage profile identified based on a current having a second waveform output from the battery cell.

[0108] At this time, which voltage profile among the first input voltage profile and the second input voltage profile is to be identified can be determined based on the charge level of the battery cell.

[0109] For example, at least one processor (205) of the battery diagnostic device (201) can identify the first input voltage profile by supplying a current having a first waveform to the battery cell within a specified time period when the state of charge (e.g., the state of charge (SOC) of the battery cell) of the battery cell is within a predetermined state range (e.g., a state of charge (SOC) of less than about 50%).

[0110] This is because, when the state of charge of a battery cell is lower than a predetermined state value, the stability of the voltage profile obtained by charging the battery cell is higher than the stability of the voltage profile obtained by discharging the battery cell. In addition, when the state of charge of a battery cell is lower than a predetermined state value, the magnitude of the current that can be supplied to the battery cell during charging is greater than the magnitude of the current that the battery cell can output during discharging. Therefore, when the state of charge of a battery cell is lower than a predetermined state value, the diagnostic accuracy of a method for diagnosing the state of the battery (e.g., battery life information such as SOH) based on a first input voltage profile may be higher than the diagnostic accuracy of a method for diagnosing the state of the battery based on a second input voltage profile.

[0111] For example, at least one processor (205) of the battery diagnostic device (201) may identify a second input voltage profile based on a current having a second waveform output from the battery cell when the state of charge (e.g., the state of charge (SOC) of the battery cell) of the battery cell is not within a predetermined state range (e.g., less than about 50% of the state of charge (SOC)). The first waveform and the second waveform may be the same or different.

[0112] This is because, when the degree of charge of a battery cell is greater than a predetermined state value, the stability of the voltage profile obtained by discharging the battery cell is higher than the stability of the voltage profile obtained by charging the battery cell. In addition, when the degree of charge of a battery cell is greater than a predetermined state value, the magnitude of the current that the battery cell can output during discharge is greater than the magnitude of the current that can be supplied to the battery cell during charging. Therefore, when the degree of charge of a battery cell is greater than a predetermined state value, the diagnostic accuracy of a method for diagnosing the state of the battery (e.g., battery life information such as SOH) based on a second input voltage profile may be higher than the diagnostic accuracy of a method for diagnosing the state of the battery based on a first input voltage profile.

[0113] Accordingly, at least one processor (205) of the battery diagnosis device (201) can diagnose the state of the battery cell regardless of the charge level of the battery cell. In addition, since at least one processor (205) of the battery diagnosis device (201) identifies an input voltage profile (e.g., a first input voltage profile, a second input voltage profile) within a specified time, the input voltage profile can be identified within a shorter time compared to the time required for voltage profile identification in a conventional battery diagnosis method, and the battery diagnosis can be performed within a shorter time compared to the time required in a conventional battery diagnosis method.

[0114] According to one embodiment, at least one processor (205) of the battery diagnostic device (201) can identify a state of a battery cell (e.g., a state of health (SOH) of the battery cell) based on at least one of at least a portion of an input voltage profile (403) (e.g., a first input voltage profile, a second input voltage profile), at least a portion of a feature vector (407), at least a portion of a restored voltage profile (411), at least a portion of a differential value of the input voltage profile (403), or any combination thereof.

[0115] The second model, which is a regression model, can output a parameter representing the state of the battery cell (e.g., the state of health (SOH) of the battery cell) as a continuous value based on a regression equation such as mathematical equation 1 by using at least one of at least a portion of the input voltage profile (403), at least a portion of the feature vector (407), at least a portion of the restored voltage profile (411), at least a portion of the differential values ​​of the input voltage profile (403), or any combination thereof.

[0116]

[0117] SOH can indicate the lifespan information of a battery cell. , , are regression weights, which can be determined through learning the second model. , , may include at least one of at least a portion of the input voltage profile (403), at least a portion of the feature vector (407), at least a portion of the restored voltage profile (411), at least a portion of the differential values ​​of the input voltage profile (403), or any combination thereof. The parameters included in the regression equation may be predetermined. In addition, although mathematical equation 1 is in the form of a multilinear equation, the regression equation included in the embodiment of the present document may not be limited thereto.

[0118] FIG. 5 illustrates an example of an operation of diagnosing life information of a battery cell according to the degree of charge of the battery cell in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0119] Hereinafter, it is assumed that at least one processor (205) of the battery diagnosis device (201) of FIG. 5 performs the process of FIG. 5. In addition, in the description of FIG. 5, the operations described as being performed by the battery diagnosis device (201) can be understood as being controlled by at least one processor (205) of the battery diagnosis device (201).

[0120] Referring to FIG. 5, in a first operation (501), at least one processor (205) of a battery diagnosis device (201) according to an embodiment may identify whether the degree of charge of a battery cell is within a predetermined state range (e.g., a SOC range of less than about 50%). If the degree of charge of the battery cell is within the predetermined state range, at least one processor (205) of the battery diagnosis device (201) may perform a second operation (503). If the degree of charge of the battery cell is not within the predetermined state range, at least one processor (205) of the battery diagnosis device (201) may perform a third operation (517).

[0121] In a second operation (503), at least one processor (205) of a battery diagnostic device (201) according to one embodiment can supply a current having a first waveform to a battery cell.

[0122] In the fourth operation (505), at least one processor (205) of the battery diagnostic device (201) according to one embodiment may identify the first input voltage profile. For example, the fourth operation (505) may be performed in a data collection unit included in the battery diagnostic device (201).

[0123] In the fifth operation (507), at least one processor (205) of the battery diagnostic device (201) according to one embodiment may normalize the first input voltage profile.

[0124] For example, at least one processor (205) of the battery diagnostic device (201) can normalize the first input voltage profile by dividing a value obtained by subtracting the minimum value of the first input voltage profile from the first input voltage profile by a value obtained by subtracting the minimum value of the first input voltage profile from the maximum value of the first input voltage profile.

[0125] For example, in a feature extraction unit included in a battery diagnostic device (201), the fifth operation (507) can be performed.

[0126] In a sixth operation (509), at least one processor (205) of a battery diagnostic device (201) according to an embodiment may input a normalized first input voltage profile into a first model. The first model may include a model that is restored after encoding.

[0127] For example, in a feature extraction unit included in a battery diagnostic device (201), the sixth operation (509) can be performed.

[0128] In the seventh operation (511), at least one processor (205) of the battery diagnostic device (201) according to one embodiment can identify a feature vector and a restoration voltage profile based on the first model.

[0129] For example, in a feature extraction unit included in a battery diagnostic device (201), the seventh operation (511) can be performed.

[0130] In the eighth operation (513), at least one processor (205) of the battery diagnostic device (201) according to one embodiment may input at least one of at least a portion of the feature vector, at least a portion of the first input voltage profile, at least a portion of the restored voltage profile, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof, into the second model. The second model may include a regression model.

[0131] For example, the types of parameters input to the second model may be predefined or determined based on predefined criteria. Examples of criteria for determining the types of parameters are described with reference to FIGS. 6 and 7.

[0132] For example, the profile input to the second model among the parameters may be determined according to a predetermined criterion. For example, at least one processor (205) of the battery diagnosis device (201) may input at least one of the following into the second model: at least one component in a predetermined order among the components of the feature vector, at least one component in a predetermined order among the components of the first input voltage profile, at least one component in a predetermined order among the components of the restored voltage profile, at least one value in a predetermined order among the derivative values ​​of the first input voltage profile, or any combination thereof.

[0133] For example, in a life prediction unit included in a battery diagnostic device (201), the eighth operation (513) can be performed.

[0134] In the ninth operation (515), at least one processor (205) of the battery diagnostic device (201) according to one embodiment can estimate the SOH of the battery cell based on the second model.

[0135] For example, in a life prediction unit included in a battery diagnostic device (201), the ninth operation (515) can be performed.

[0136] In a third operation (517), at least one processor (205) of a battery diagnostic device (201) according to one embodiment can output a current having a second waveform from a battery cell.

[0137] In the tenth operation (519), at least one processor (205) of the battery diagnostic device (201) according to one embodiment can identify a second input voltage profile.

[0138] For example, in a data collection unit included in a battery diagnostic device (201), the tenth operation (519) can be performed.

[0139] In the eleventh operation (521), at least one processor (205) of the battery diagnostic device (201) according to one embodiment may normalize the second input voltage profile.

[0140] For example, as in the fifth operation (507), at least one processor (205) of the battery diagnostic device (201) can normalize the second input voltage profile by dividing the value obtained by subtracting the minimum value of the second input voltage profile from the second input voltage profile by the value obtained by subtracting the minimum value of the second input voltage profile from the maximum value of the second input voltage profile.

[0141] For example, in a feature extraction unit included in a battery diagnostic device (201), the 11th operation (521) can be performed.

[0142] In the 12th operation (523), at least one processor (205) of the battery diagnostic device (201) according to one embodiment may input the second input voltage profile into the first model.

[0143] For example, in a feature extraction unit included in a battery diagnostic device (201), the 12th operation (523) can be performed.

[0144] In the 13th operation (525), at least one processor (205) of the battery diagnostic device (201) according to one embodiment can identify a feature vector and a restoration voltage profile based on the first model.

[0145] For example, in a feature extraction unit included in a battery diagnostic device (201), the 13th operation (525) can be performed.

[0146] In operation 14 (527), at least one processor (205) of a battery diagnostic device (201) according to an embodiment may input at least one of at least a portion of the feature vector, at least a portion of the second input voltage profile, at least a portion of the restored voltage profile, at least a portion of the differential values ​​of the second input voltage profile, or any combination thereof, into the second model.

[0147] For example, as in the 8th operation (513), an example of a criterion for determining the type of parameter is described with reference to FIGS. 6 to 7.

[0148] For example, as in the 8th operation (513), the profile input to the second model among the parameters can be determined according to a predetermined criterion.

[0149] For example, in a life prediction unit included in a battery diagnostic device (201), the 14th operation (527) can be performed.

[0150] In the 15th operation (529), at least one processor (205) of the battery diagnostic device (201) according to one embodiment can estimate the SOH of the battery cell based on the second model.

[0151] For example, in a life prediction unit included in a battery diagnostic device (201), the 15th operation (529) can be performed.

[0152] FIG. 6 illustrates an example of an operation in which the type of data input to a second model, which is a regression model, is changed according to a component of a latent vector in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0153] Hereinafter, it is assumed that at least one processor (205) of the battery diagnosis device (201) of FIG. 6 performs the process of FIG. 6. In addition, in the description of FIG. 6, the operations described as being performed by the battery diagnosis device (201) can be understood as being controlled by at least one processor (205) of the battery diagnosis device (201).

[0154] The first operation (601), the second operation (603), and the third operation (605) may be performed in place of the eighth operation (513) and the ninth operation (515) of FIG. 5, or may be performed in place of the fourteenth operation (527) and the fifteenth operation (529).

[0155] Referring to FIG. 6, in a first operation (601), at least one processor (205) of a battery diagnosis device (201) according to an embodiment may identify whether all components of a feature vector are included in a predetermined range (e.g., a range less than a predetermined feature value). If all components of the feature vector are included in the predetermined range, at least one processor (205) of the battery diagnosis device (201) may perform a second operation (603). If any of the components of the feature vector are not included in the predetermined range, at least one processor (205) of the battery diagnosis device (201) may perform a third operation (605).

[0156] In a second operation (603), at least one processor (205) of a battery diagnosis device (201) according to an embodiment can estimate a first SOH of a battery cell by inputting at least some of the feature vectors and at least some of the input voltage profiles (e.g., a first input voltage profile, a second input voltage profile), at least some of the restored voltage profiles (e.g., a restored voltage profile for the first input voltage profile, a restored voltage profile for the second input voltage profile), at least some of the differential values ​​of the input voltage profiles, or any combination thereof, into a second model, if all components of the feature vectors fall within a predetermined range.

[0157] In a third operation (605), at least one processor (205) of a battery diagnostic device (201) according to an embodiment may estimate a second SOH of a battery cell by inputting at least one of at least a portion of an input voltage profile (e.g., a first input voltage profile, a second input voltage profile), at least a portion of a restored voltage profile (e.g., a restored voltage profile for the first input voltage profile, a restored voltage profile for the second input voltage profile), at least a portion of a differential value of an input voltage profile, or any combination thereof into a second model.

[0158] In other words, since noise may occur in the feature vector if at least one component of the feature vector is not included in a predetermined range, at least one processor (205) of the battery diagnosis device (201) can identify life information of a battery cell (e.g., second SOH) by inputting at least one parameter excluding the feature vector into the second model to reduce the influence of noise.

[0159] According to one embodiment, in a battery diagnostic device (201) performing the operation of FIG. 6, the type of parameter input to the second model may change as the component of the feature vector gradually increases or decreases.

[0160] FIG. 7 illustrates an example of an operation in which the value of data input to a second model, which is a regression model, is changed according to a component of a latent vector in a battery diagnosis device and a battery diagnosis method according to an embodiment of the present document.

[0161] Hereinafter, it is assumed that at least one processor (205) of the battery diagnosis device (201) of FIG. 7 performs the process of FIG. 7. In addition, in the description of FIG. 7, the operations described as being performed by the battery diagnosis device (201) can be understood as being controlled by at least one processor (205) of the battery diagnosis device (201).

[0162] The first operation (701), the second operation (703), and the third operation (705) may be performed in place of the eighth operation (513) and the ninth operation (515) of FIG. 5, or may be performed in place of the fourteenth operation (527) and the fifteenth operation (529).

[0163] Referring to FIG. 7, in a first operation (701), at least one processor (205) of a battery diagnosis device (201) according to an embodiment may identify whether all components of a feature vector are included in a predetermined range (e.g., a range less than a predetermined feature value). If all components of the feature vector are included in the predetermined range, at least one processor (205) of the battery diagnosis device (201) according to an embodiment may perform a second operation (703). If at least one component of the feature vector is not included in the predetermined range, at least one processor (205) of the battery diagnosis device (201) according to an embodiment may perform a third operation (705).

[0164] In a second operation (703), at least one processor (205) of a battery diagnostic device (201) according to an embodiment can estimate a first SOH of a battery cell by inputting at least one of at least a portion of an input voltage profile, at least a portion of a restored voltage profile, at least a portion of a differential value of an input voltage profile, or any combination thereof, and at least a portion of a feature vector into a second model.

[0165] In the third operation (705), at least one processor (205) of the battery diagnostic device (201) according to one embodiment may replace components of the feature vector that are outside a predetermined range with boundary values ​​of the predetermined range.

[0166] In other words, at least one processor (205) of the battery diagnosis device (201) may replace components of the feature vector that are greater than an upper limit of a predetermined range with an upper limit value, and replace components of the feature vector that are less than a lower limit of the predetermined range with a lower limit value, in order to reduce the influence of noise, since noise may occur in the feature vector. Thereafter, at least one processor (205) of the battery diagnosis device (201) may perform a second operation (703) based on the feature vector in which some components are replaced.

[0167] According to one embodiment, in a battery diagnostic device (201) performing the operation of FIG. 7, as the components of the feature vector gradually increase or decrease, a parameter indicating the state of the battery cell (e.g., the state of health (SOH) of the battery cell) may converge to a specific value.

[0168] FIG. 8 illustrates an example of an operation in which a method of deriving a state of a battery cell from data output from a second model, which is a regression model, is changed according to a component of a latent vector in a battery diagnosis device and a battery diagnosis method according to an embodiment of the present document.

[0169] Hereinafter, it is assumed that at least one processor (205) of the battery diagnosis device (201) of FIG. 8 performs the process of FIG. 8. In addition, in the description of FIG. 8, the operations described as being performed by the battery diagnosis device (201) can be understood as being controlled by at least one processor (205) of the battery diagnosis device (201).

[0170] The first operation (801), the second operation (803), and the third operation (805) may be performed in place of the eighth operation (513) and the ninth operation (515) of FIG. 5, or may be performed in place of the fourteenth operation (527) and the fifteenth operation (529).

[0171] In a first operation (801), at least one processor (205) of a battery diagnosis device (201) according to an embodiment may identify whether all components of a feature vector are included in a predetermined range (e.g., a range less than a predetermined feature value). If all components of the feature vector are included in the predetermined range, at least one processor (205) of a battery diagnosis device (201) according to an embodiment may perform a second operation (803). If at least one component of the feature vector is not included in the predetermined range, at least one processor (205) of a battery diagnosis device (201) according to an embodiment may perform a third operation (805).

[0172] In a second operation (803), at least one processor (205) of a battery diagnostic device (201) according to an embodiment can estimate the SOH of the battery cell based on the first SOH and the second SOH.

[0173] The first SOH can represent an SOH output from a second model by inputting at least one of the input voltage profile, at least one of the restored voltage profile, at least one of the differential values ​​of the input voltage profile, or any combination thereof, and at least one of the feature vectors into the second model.

[0174] The second SOH can represent an SOH output from the second model by inputting at least one of at least a portion of the input voltage profile, at least a portion of the restored voltage profile, at least a portion of the differential values ​​of the input voltage profile, or any combination thereof into the second model.

[0175] For example, at least one processor (205) of the battery diagnostic device (201) can estimate the SOH of the battery cell based on an average value of the first SOH and the second SOH.

[0176] In a third operation (805), at least one processor (205) of a battery diagnosis device (201) according to an embodiment may estimate the SOH of a battery cell based on the second SOH. In other words, at least one processor (205) of the battery diagnosis device (201) may estimate the SOH of a battery cell based on the second SOH in order to reduce the influence of noise, since noise may occur in the feature vector.

[0177] According to one embodiment, in a battery diagnostic device (201) performing the operation of FIG. 6, as the components of the feature vector gradually increase or decrease, the types of parameters input to the second model and the method of deriving the state of the battery cell may be changed.

[0178] FIG. 9 illustrates an example of an operation for diagnosing the status of a battery cell in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0179] Hereinafter, it is assumed that at least one processor (205) of the battery diagnosis device (201) of FIG. 9 performs the process of FIG. 9. In addition, in the description of FIG. 9, the operations described as being performed by the battery diagnosis device (201) can be understood as being controlled by at least one processor (205) of the battery diagnosis device (201).

[0180] In a first operation (901), at least one processor (205) of a battery diagnostic device (201) according to one embodiment can identify a first input voltage profile measured by supplying a current having a first waveform to a battery cell.

[0181] In a second operation (903), at least one processor (205) of a battery diagnostic device (201) according to an embodiment may input a first input voltage profile into a first model. The first model may include a model that is restored after encoding.

[0182] In a third operation (905), at least one processor (205) of a battery diagnostic device (201) according to an embodiment can diagnose the state of the battery cell based on a feature vector identified based on the first model.

[0183] FIG. 10 illustrates an example of an operation of diagnosing the state of a battery cell based on a voltage profile of the battery cell obtained in a manner determined according to the degree of charge of the battery cell, in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0184] Hereinafter, it is assumed that at least one processor (205) of the battery diagnosis device (201) of FIG. 10 performs the process of FIG. 10. In addition, in the description of FIG. 10, the operations described as being performed by the battery diagnosis device (201) can be understood as being controlled by at least one processor (205) of the battery diagnosis device (201).

[0185] In a first operation (1001), at least one processor (205) of a battery diagnostic device (201) according to an embodiment can identify a first input voltage profile by supplying a current having a first waveform to a battery cell, or can identify a second input voltage profile based on a current having a second waveform output from the battery cell.

[0186] In a second operation (1003), at least one processor (205) of a battery diagnostic device (201) according to one embodiment may input a first input voltage profile or a second input voltage profile into the first model.

[0187] In a third operation (1005), at least one processor (205) of a battery diagnostic device (201) according to an embodiment can diagnose the state of a battery cell based on a feature vector identified based on the first model.

[0188] FIG. 11 is a block diagram showing the hardware configuration of a computing system that performs a battery diagnosis method in a battery diagnosis device and a battery diagnosis method according to one embodiment of the present document.

[0189] Referring to FIG. 11, a computing system (1100) according to an embodiment disclosed in the present document may include an MCU (1110), a memory (1120), an input / output I / F (1130), and a communication I / F (1140).

[0190] The MCU (1110) may be at least one processor that executes various programs stored in the memory (1120) (e.g., a battery cell data collection program, a graph generation program, a data analysis program, a data decomposition algorithm, a normalization program, a battery cell diagnosis program, etc.), processes various information including battery cell characteristic data and latent variables through these programs, and performs the functions of the battery diagnosis device (201) shown in the above-described FIGS. 2 to 10.

[0191] The memory (1120) can store various programs such as a battery cell data collection program, a graph generation program, a data analysis program, a data decomposition algorithm, a normalization program, and a battery cell diagnosis program.

[0192] Such memories (1120) may be provided in multiple numbers as needed. The memories (1120) may be volatile memories or non-volatile memories. As volatile memories (1120), RAM, DRAM, SRAM, etc. may be used. As non-volatile memories (1120), ROM, PROM, EAROM, EPROM, EEPROM, flash memories, etc. may be used. The examples of the memories (1120) listed above are merely examples and are not limited to these examples.

[0193] The input / output I / F (1130) can provide an interface that enables data transmission and reception between an input device (not shown) such as a keyboard, mouse, or touch panel, and an output device (not shown) such as a display and the MCU (1110).

[0194] The communication I / F (1140) is a component capable of transmitting and receiving various data with the server, and may be any device capable of supporting wired or wireless communication. For example, the diagnostic device (201) can transmit and receive various types of information, including battery cell shape models, from a separately provided external server via the communication I / F (1140).

[0195] In this way, a computer program according to an embodiment disclosed in this document may be implemented as a module that is recorded in a memory (1120) and processed by an MCU (1110) to perform each function illustrated in FIG. 2, for example.

[0196] In the above, although all components constituting the embodiments disclosed in this document have been described as being combined or operating in combination as one, the embodiments disclosed in this document are not necessarily limited to such embodiments. That is, within the scope of the purpose of the embodiments disclosed in this document, all of the components may be selectively combined and operated one or more times.

[0197] In addition, terms such as "include," "comprise," or "have" described above, unless specifically stated to the contrary, should be interpreted to imply the inclusion of the corresponding component, and thus should not be interpreted to exclude other components, but rather to include other components. All terms, including technical or scientific terms, have the same meaning as commonly understood by a person of ordinary skill in the art to which the embodiments disclosed in this document belong, unless otherwise defined. Commonly used terms, such as terms defined in a dictionary, should be interpreted to be consistent with the contextual meaning of the relevant technology, and shall not be interpreted in an idealized or overly formal sense, unless explicitly defined in this document.

[0198] The foregoing disclosure outlines features of several embodiments to enable those skilled in the art to better understand the aspects of the present disclosure. Those skilled in the art will readily appreciate that the present disclosure can be readily used as a basis for designing or modifying other structures to achieve the same purposes or advantages of the embodiments introduced herein. Furthermore, those skilled in the art will recognize that such equivalent structures do not depart from the scope of the present disclosure, and that various changes, substitutions, and modifications can be made herein without departing from the scope of the present disclosure.

Claims

1. Memory that stores at least one instruction; and comprising at least one processor executing at least one instruction; At least one processor, Identifying a first input voltage profile comprising a voltage profile measured by supplying a current having a specified waveform to a battery cell, Input the first input voltage profile into the first model that encodes and then restores it, Based on the feature vector identified based on the first model, configured to diagnose the state of the battery cell, Battery diagnostic device.

2. In claim 1, At least one processor, A method configured to identify the feature vector based on a latent vector, which is a processing result in the bottleneck layer included in the first model, Battery diagnostic device.

3. In claim 2, At least one processor, A device configured to diagnose a state of the battery cell based on at least one of at least a portion of the feature vector, at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof. Battery diagnostic device.

4. In claim 2, At least one processor, Diagnosing the state of the battery cell based on at least one of the first input voltage profile, at least one of the restored voltage profile restored based on the first model, at least one of the differential values ​​of the first input voltage profile, or any combination thereof and at least one of the feature vector, or Based on at least one of the components of the feature vector being outside a predetermined range, at least one of the first input voltage profile, at least one of the restored voltage profiles restored based on the first model, at least one of the derivatives of the first input voltage profile, or any combination thereof, is configured to diagnose the state of the battery cell. Battery diagnostic device.

5. In claim 1, At least one processor, configured to input at least some of the feature vectors to a second model that outputs the state of the battery cell; Battery diagnostic device.

6. In claim 1, At least one processor, configured to supply the above current to the battery cell within a specified time, Battery diagnostic device.

7. In claim 1, The condition of the above battery is: Composed to include SOH (state of health), Battery diagnostic device.

8. In claim 1, The above first model is, It is configured to include an LSTM AE (long short term memory autoencoder) model, Battery diagnostic device.

9. An operation of identifying a first input voltage profile including a measured voltage profile by supplying a current having a specified waveform to a battery cell; An operation of inputting the first input voltage profile into a first model that encodes and then restores the first input voltage profile; and An operation for diagnosing the state of the battery cell based on a feature vector identified based on the first model, How to diagnose a battery.

10. In claim 9, Based on the feature vector identified based on the first model, the operation of diagnosing the state of the battery cell is as follows: An operation for identifying the feature vector based on a latent vector, which is a result of processing in a bottleneck layer included in the first model, How to diagnose a battery.

11. In claim 10, Based on the feature vector identified based on the first model, the operation of diagnosing the state of the battery cell is as follows: An operation of diagnosing a state of the battery cell based on at least one of at least a portion of the feature vector, at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof. How to diagnose a battery.

12. In claim 10, The operation of identifying the feature vector based on the latent vector, which is a result of processing in the bottleneck layer included in the first model, An operation of diagnosing the state of the battery cell based on at least one of at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the differential values ​​of the first input voltage profile, or any combination thereof and at least a portion of the feature vector, based on all components of the feature vector being included in the predetermined range; or An operation of diagnosing a state of the battery cell based on at least one of at least a portion of the first input voltage profile, at least a portion of the restored voltage profile restored based on the first model, at least a portion of the derivative values ​​of the first input voltage profile, or any combination thereof, based on at least one component of the feature vector being outside a predetermined range. How to diagnose a battery.

13. In claim 9, Based on the feature vector identified based on the first model, the operation of diagnosing the state of the battery cell is as follows: An operation of inputting at least some of the feature vectors into a second model that outputs the state of the battery cell, How to diagnose a battery.

14. In claim 9, An operation of identifying the first input voltage profile, which includes a voltage profile measured by supplying a current having the specified waveform to the battery cell, Including an operation of supplying the current to the battery cell within a specified time, How to diagnose a battery.

15. In claim 9, The condition of the above battery is: Including SOH (state of health), How to diagnose a battery.

16. In claim 9, The above first model is, Including the LSTM AE (long short term memory autoencoder) model, How to diagnose a battery.

17. A computer-readable recording medium having recorded thereon a program for performing the method of any one of claims 9 to 16 on a computer.

Images