KR102574397B1 - Battery diagnosis method and apparatus - Google Patents

Abstract

A method and device for diagnosing a battery module are disclosed. The battery module diagnosis device measures the impedance of each cell by applying the cell measurement frequency to each cell through the welding part between the cells, and the relationship between the impedance distribution in the cell unit and the welding defect is learned in the cell analysis model. Input the impedance to determine the welding defect of each cell.

Description

Battery module diagnosis method and apparatus {Battery diagnosis method and apparatus}

Embodiments of the present invention relate to a method and apparatus for diagnosing a state of a battery module, and more particularly, to a method and apparatus for diagnosing defective welding between cells in a battery module.

Electrochemical Impedance Spectroscopy (EIS) is a method of analyzing resistance characteristics according to frequency by applying AC electrical signals of various frequencies to a battery and measuring a response thereto. The time to analyze the resistance characteristics is inversely proportional to the magnitude of the frequency, and in particular, it takes a considerable amount of time to analyze the resistance characteristics at a low frequency of less than several Hz.

Another battery test method is the AC-IR (Alternating Current Internal Resistance) method. AC internal resistance measurement generally measures the response of a battery to an AC signal of 1 kHz and is measured without changing the frequency.

The electric capacity of battery systems is gradually increasing, as the use of batteries is expanding from small devices to electric vehicles, etc. Accordingly, individual battery cells are being designed with low internal resistance and high energy capacity. As the design direction of batteries changes and diversifies, the existing AC internal resistance measurement method measured at 1 kHz has difficulty in accurately identifying the characteristics of a large capacity battery, and the impedance spectroscopy method cannot meet the demand for battery measurement due to the long inspection time. .

Defects of a battery module including a plurality of cells are largely divided into defects of the cell itself and defects of electrical connection between the cells. Impedance spectroscopy or alternating current internal resistance measurement methods can only identify defects in cell units, but have limitations in identifying connection defects between cells. The existing measurement frequency of 1 kHz is a frequency range in which ohmic resistance is measured based on low-capacity batteries, but in high-capacity batteries, inductance interference is severe, so it is not suitable for diagnosing high-capacity batteries. Also, single-frequency measurements can only look at electrical connections and are difficult to inspect for defects in the cell itself.

A technical problem to be achieved by embodiments of the present invention is to provide a method and apparatus for diagnosing a battery state including electrical defects between cells in a battery module including a plurality of cells.

An example of a method for diagnosing a battery module according to an embodiment of the present invention for achieving the above technical problem is a method for diagnosing a battery module including a plurality of cells, by measuring a cell measurement frequency through a welding portion between cells. measuring the impedance of each cell by applying it to each cell; and determining the welding defects of each cell by inputting the impedance of each cell to the cell analysis model that has learned the relationship between the impedance distribution of each cell and the welding defect.

An example of a battery module diagnosis apparatus according to an embodiment of the present invention for achieving the above technical problem is a measurement unit that measures the impedance of each cell by applying a cell measurement frequency to each cell through a welding portion between cells. ; and a cell diagnosis unit configured to identify welding defects of each cell by inputting the impedance of each cell to the cell analysis model having learned the relationship between the impedance distribution and the welding defect.

According to an embodiment of the present invention, electrical connection defects (eg, welding defects, etc.) between a plurality of cells in a battery module may be identified. In another embodiment, the state of each cell or the entire battery module (for example, good or bad) may be determined together with an electrical connection defect. As another embodiment, by using only a certain number of frequencies for diagnosing the battery module, the time required for diagnosing can be reduced compared to the conventional impedance spectroscopy.

1 is a diagram showing an example of a battery module diagnostic device according to an embodiment of the present invention;
2 is a flowchart illustrating an example of a method for diagnosing a battery module according to an embodiment of the present invention;
3 is a view showing an example of a battery module according to an embodiment of the present invention;
4 is a diagram showing an example of an equivalent circuit of a battery module according to an embodiment of the present invention;
5 is a diagram showing an example of a cell analysis model according to an embodiment of the present invention;
6 is a diagram showing an example of a module analysis model according to an embodiment of the present invention;
7 is a diagram showing an example of a process of selecting a measurement frequency according to an embodiment of the present invention;
8 is a diagram showing an example of a frequency used for measurement of a battery module according to an embodiment of the present invention, and
9 is a diagram showing an example of the configuration of a battery module diagnostic device according to an embodiment of the present invention.

Hereinafter, a battery module diagnosis method and device according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

1 is a diagram showing an example of a battery module diagnostic device according to an embodiment of the present invention.

Referring to FIG. 1, the battery module diagnosis device 100 (hereinafter, referred to as a 'diagnosis device') receives an impedance measured by applying a frequency to the battery module 130, analyzes the impedance, and analyzes the electrical power of the battery module. A battery state including a connection defect (eg, welding defect 160, etc.) is output.

The battery module 130 includes a plurality of cells. A plurality of cells may be connected in series or parallel. The plurality of cells may be connected in various ways such as welding or busbar. An example of a battery module 130 including a plurality of cells is shown in FIG. 3 .

A frequency may be applied to the battery module 130 to measure the impedance of each cell and/or the entire battery module. Impedance measured by applying a frequency to each cell is named 'cell impedance 140', and impedance measured by applying a frequency to the entire battery module is named 'module impedance 150'. A method of measuring the cell impedance 140 and the module impedance 150 will be reviewed again in FIGS. 3 and 4 .

The diagnostic device 100 may analyze the cell impedance 140 through the cell analysis model 110 or analyze the module impedance 150 through the module analysis model 120 to determine the cell state such as welding defect 160. there is. As another embodiment, the diagnosis apparatus 100 may analyze the cell impedance 140 and the module impedance 150 together through the module analysis model 120 to determine the state of the cell and/or battery module. The cell analysis model 110 and the module analysis model 120 for analyzing cells and/or battery modules will be reviewed again in FIGS. 5 and 6 .

2 is a flowchart illustrating an example of a method for diagnosing a battery module according to an embodiment of the present invention.

Referring to FIGS. 1 and 2 together, the diagnosis apparatus 100 determines the cell impedance 140 of the battery module 130 (S200). For example, the diagnostic device 100 applies the cell measurement frequency to each cell through the welding parts (A, B, C, D, A', B', C', D') between the cells as shown in FIG. Thus, the impedance of each cell can be measured. As an example, when the battery module 130 is composed of N cells, the diagnosis apparatus 100 may measure the cell impedance 140 of all or some of the N cells.

The diagnostic device 100 determines the module impedance 150 of the battery module 130 (S210). For example, the diagnosis apparatus 100 may measure the module impedance 150 by applying at least one module measurement frequency to the battery module 130 . Although this embodiment describes the case of determining both the cell impedance 140 and the module impedance 150, this is only an example and the step of determining the module impedance 150 (S210) may be omitted according to the embodiment. there is. However, in the following, for convenience of explanation, a case in which the module impedance is identified (S210) will be assumed and described.

The diagnosis apparatus 100 determines the welding defects of each cell by inputting the impedance 140 of each cell to the cell analysis model 110 having learned the relationship between the impedance distribution in units of cells and the welding defects (S220). Here, defective welding does not mean only disconnection between cells, but is interpreted as including various types of electrical connection defects that are not normal, such as increased resistance in the connection area between cells due to poor welding or intermittent current flow. It can be.

As another embodiment, the cell analysis model 110 may be a model obtained by additionally learning a relationship between an impedance distribution in a cell unit and a cell state. For example, the cell analysis model 110 receives the impedance of each cell and outputs whether or not each cell has a welding defect and the state of each cell (eg, whether or not it is defective, State of Health (SOH), etc.) can be The cell analysis model 110 will be described in detail again in FIG. 5 .

The diagnostic device 100 inputs the impedance of the battery module 130 to the module analysis model 120 that has learned the relationship between the impedance distribution and the battery state, and determines the state of the battery module (eg, pass/fail, SOH, etc.) Find out (S230). As another embodiment, the diagnosis apparatus 100 may input the cell impedance 140 and the module impedance 150 together into the module analysis model 120 to determine the state of each cell and the battery module together. The module analysis model 120 will be described in detail again in FIG. 6 .

The cell measurement frequency used to measure the cell impedance 140 and the module measurement frequency used to measure the module impedance 150 may be the same as or different from each other. In another embodiment, the cell measurement frequency and/or the module measurement frequency may be a frequency used in electrochemical impedance spectroscopy or an AC internal resistance measurement method. However, electrochemical impedance spectroscopy is advantageous in identifying various electrochemical characteristics of a cell and/or battery module, but has a disadvantage in that it takes a long time to measure impedance. Since the AC internal resistance measurement method uses only one specific frequency (eg, 1 kHz), the time required for impedance measurement can be reduced, but it is difficult to accurately determine the characteristics of the cell and/or battery module.

Accordingly, this embodiment suggests a method of using a plurality of frequencies selected within a certain range as cell measurement frequencies and/or module measurement frequencies so that accurate characteristics of cells and/or battery modules can be grasped in a short time. This will be reviewed again in FIG. 7 .

3 is a diagram showing an example of a battery module according to an embodiment of the present invention, and FIG. 4 is a diagram showing an example of an equivalent circuit of the battery module according to an embodiment of the present invention.

Referring to FIG. 3 , the battery module includes a plurality of cells. In this embodiment, a battery module 300 in which a plurality of cells are connected by welding (A, B, C, D) and a plurality of cells are connected using bus bars (A', B', C', D'). One battery module 310 is shown. The defects of the battery modules 300 and 310 may include electrical connection defects between cells as well as electrochemical defects inside the cells. There is a limit to the user's ability to visually check these electrical connection defects one by one.

In this embodiment, when measuring cell impedance in order to determine an electrical connection between cells of the battery modules 300 and 310 (ie, a welding defect), a measurement terminal is connected to a welding portion. For example, the diagnostic device 100 connects the two electrodes to the A and B (or A'B') welds, connects the B and C (or B' and C') welds, or connects the C and D (or The impedance of the corresponding cell can be measured by connecting to the welding part of C' and D'). As another embodiment, the diagnosis apparatus 100 may connect two electrodes to the two welding points A and C to measure the impedance of the two cells. In addition, in this embodiment, module impedance can be measured by connecting two electrodes to the first cell and the last cell of the battery module.

Referring to FIG. 4 , each cell in the equivalent circuit of the battery modules 300 and 310 may be expressed as a unit parallel circuit (circuits 1 to 6). The diagnostic device 100 may measure cell impedance by connecting two measuring electrodes to each node. When each cell has a parallel circuit structure, the diagnosis apparatus 100 may measure the impedance between two nodes adjacent to each cell to measure the impedance for a unit parallel circuit. At this time, the measured temperature or state of charge (SOC) of the battery modules 300 and 310 may be set in advance to be in a predefined condition.

When electrical connection defects (eg, welding defects, etc.) between cells occur, this affects cell impedance. In addition, cell impedance and/or module impedance are affected by electrochemical characteristics inside each cell. Therefore, by measuring cell impedance and/or module impedance through nodes 1 to 7, it is possible to determine not only the state of the cell itself but also the state of electrical connection such as a welded portion between cells.

5 is a diagram showing an example of a cell analysis model according to an embodiment of the present invention.

Referring to FIG. 5 , the cell analysis model 110 is an artificial intelligence model that predicts and outputs welding defects 510 between cells when cell impedance 500 for each cell is received. The cell analysis model 110 may be implemented with various conventional artificial neural networks such as CNN (Convolutional Neural Network).

The cell analysis model 110 requires a training process for analyzing the cell impedance 500 . A data set for training the cell analysis model 110 may include cell impedance 500 in a normal state and cell impedance 500 in various states of poor welding. The diagnosis device 110 may train the cell analysis model 110 using a supervised learning method using the cell impedance labeled as a normal state or a welding defect state.

As another embodiment, the cell analysis model 110 may be a model that outputs a cell state 520 together with a welding defect 510 . For example, the cell analysis model 110 may be a model that determines whether or not a corresponding cell is in a welding defect state when the cell impedance 500 is input, and also outputs a cell state (defective state, etc.). To this end, the cell analysis model 110 may train the cell analysis model 110 by a supervised learning method using the cell impedance 500 labeled as a welding defect and a cell state.

The cell analysis model 110 is a model that sequentially receives the cell impedance 500 in units of cells and outputs welding defects of each cell, or data obtained by concatenating the cell impedances 500 for a plurality of cells. When input is received, it may be a model that outputs welding defects between each cell. For example, in the example of FIG. 3 , since a weld between two cells affects both adjacent cells, in order to determine the state of the weld B, the impedance of the cell between A-B and the cell between B-C is measured together. By inputting the cell analysis model 110, it is possible to determine whether or not a defect 510 exists.

In another embodiment, in order to more accurately identify electrical connection defects such as poor welding 510, the state of charge (SOC) of the battery module that may affect the cell impedance 500 may be preset to a predefined value. there is. For example, after making the SOC value of at least one learning battery module a predefined value (eg, 90% state of charge, etc.), the cell impedance measured by applying the cell measurement frequency to each cell of the learning battery module It can be made into a training dataset. Diagnosis device 100 makes the SOC value of the battery module the same as the SOC value of the learning battery module when creating the training data set, and then measures the cell impedance 500 of the battery module to identify welding defects 510 and the like. there is. As another embodiment, the diagnostic device 100 measures the cell impedance 500 of the battery module in the same environment as the environmental factors such as temperature or humidity when the training data set is generated to determine welding defects 510 and the like. can

6 is a diagram showing an example of a module analysis model according to an embodiment of the present invention.

Referring to FIG. 6 , the module analysis model 120 is an artificial intelligence model that predicts and outputs a state 620 of a module when a module impedance 610 is input. The module analysis model 120 may be implemented with various conventional artificial neural networks such as CNN (Convolutional Neural Network).

The module analysis model 120 requires a training process to perform analysis of the module impedance 610 . A dataset for training the module analysis model 120 may include module impedance 610 in normal state and various abnormal states (eg, defective state, various SOH states, etc.). The diagnosis apparatus 100 may train the module analysis model 120 by a supervised learning method using the module impedance 610 labeled as state values of various battery modules.

As another embodiment, the module analysis model 120 may be a model that receives the cell impedance 600 together with the module impedance 610 and predicts the state (whether defective, SOH, etc.) of the cell and battery module. To this end, the module analysis model 120 may be trained using the cell impedance 600 and the module impedance 610 labeled with various state values of the cell and battery module.

As another embodiment, the SOC of the battery module that may affect the measurement of the module impedance 610 may be preset to a predefined value as shown in FIG. 5 . For example, after setting the SOC value of at least one learning battery module to a predefined value (eg, 90% charge, etc.), the module impedance 610 measured by applying the module measurement frequency to the learning battery module is trained. can be made into a dataset. When diagnosing the battery thereafter, the diagnostic device 100 measures the module impedance 610 from the battery module after making the SOC value of the battery module the same as the SOC value of the learning battery module when generating the training data set, and then measuring the module state ( 620) can be identified. As another embodiment, the diagnostic device 100 may determine the module state 620 by measuring the module impedance 610 of the battery module in the same environment as the environmental factors such as temperature or humidity when generating the training dataset. there is.

7 is a diagram showing an example of a process of selecting a measurement frequency according to an embodiment of the present invention.

Referring to FIG. 7 , the diagnosis apparatus 100 may use a part of a frequency band as a cell measurement frequency and/or a module measurement frequency instead of the entire frequency range used in the conventional EIS. For example, if the entire frequency range used for battery impedance measurement in the conventional EIS ranges from 0.1 to 1000 Hz, the present embodiment measures impedance only for some frequency bands and does not measure impedance for the remaining frequency bands. The time required for battery diagnosis can be shortened.

However, it is difficult to determine the exact state of the cell and/or battery module only with the impedance measured using a randomly selected frequency in a frequency range used in EIS or the like. Accordingly, this embodiment suggests a characteristic frequency band that can well reflect the characteristics of the battery module, rather than using a frequency at an arbitrary position among the entire frequency range.

7 is a graph 700 showing a distribution of impedance values measured by applying frequencies from low to high frequencies to a battery at regular intervals as a curve on a complex plane. The x-axis is the real axis and the y-axis is the imaginary axis. The impedance distribution graph 700 of the present embodiment can be obtained by connecting distributions of impedance values measured using various conventional battery impedance measurement methods such as EIS. The graph 700 of this embodiment is only an example for better understanding, and the shape of the impedance distribution graph 700 may vary depending on the type of battery module.

In this embodiment, the impedance distribution graph 700 appearing on the complex plane includes at least one semicircular shape and an ohmic resistance shape. In one implementation, the diagnostic device 100 includes a semicircular maximum value area 720 and/or minimum value area 710, and/or maximum value area and/or minimum value area of the ohmic resistance area in the impedance distribution graph 700. At least one impedance value existing in 730 may be extracted as a characteristic impedance component. For example, three impedance values corresponding to the minimum point 730, the semicircular maximum point 720, and the minimum point 710 of the ohmic resistance region, or a plurality of The impedance value can be extracted as a characteristic impedance component.

Since the impedance distribution graph 700 is a graph showing impedance values measured at each frequency on multiple planes, the diagnostic device 100 can extract a frequency band corresponding to a characteristic impedance component from the impedance distribution graph 700 as a characteristic frequency component. can The diagnosis apparatus 100 may determine a characteristic frequency component corresponding to a characteristic impedance component as at least one measurement frequency used in the diagnosis apparatus 100 . For example, the diagnostic device 100 sets a plurality of frequencies corresponding to the maximum value area 720 and the minimum value area 710 and the minimum value area 730 of the semicircular shape as the cell measurement frequency and/or the module measurement frequency. can be determined by

In this embodiment, the areas 710 , 720 , and 730 for extracting the feature impedance component are only examples for understanding, and may be modified in various ways according to the embodiment. For example, the diagnostic apparatus 100 may automatically extract a point where the slope of the impedance distribution graph 700 suddenly changes or a point where the maximum value or minimum value changes, etc. through a graph analysis algorithm to determine a characteristic frequency component.

As another embodiment, the impedance distribution graph 700 may be obtained several times to determine the characteristic impedance component and the characteristic frequency component. For example, impedance distributions may be obtained several times for the same battery module or impedance distributions may be obtained several times for different battery modules of the same type. At this time, the value of the characteristic frequency component in each impedance distribution graph may not be perfectly matched. In this case, the diagnosis apparatus 100 may determine the value of the characteristic frequency component through a statistical method. For example, the diagnostic apparatus 100 obtains a normal distribution of characteristic frequency components corresponding to a plurality of characteristic impedance components obtained from a plurality of impedance distributions. The diagnostic apparatus 100 may determine, as the feature frequency component, a frequency that satisfies (μ-2σ < frequency < μ + 2σ) based on a normal distribution.

The cell measurement frequency and/or module measurement frequency identified through this embodiment is the process of generating a training dataset for learning the cell analysis model 110 and/or the module analysis model 120 and the diagnosis of the battery module of FIG. 2 can be used in common methods.

As another embodiment, a frequency selection process for a cell measurement frequency and a frequency selection process for a module measurement frequency may be performed respectively. For example, an impedance distribution graph of cells of the battery module (ie, cell impedance distribution) may be different from an impedance distribution graph of the battery module (ie, module impedance distribution). The diagnosis apparatus 100 may obtain a distribution graph of cell impedance and select a cell measurement frequency therefrom, and may also obtain a distribution graph of module impedance and select a module measurement frequency therefrom.

8 is a diagram showing an example of a frequency used for measurement of a battery module according to an embodiment of the present invention.

Referring to FIG. 8 , the diagnosis apparatus 100 includes at least one cell measurement frequency 810 used to measure the cell impedance of the battery module 800 . When a plurality of cells 802, 804, and 806 are connected in series, and each cell 802, 804, and 806 has a parallel circuit structure as shown in FIG. may be a frequency of

For example, the diagnosis apparatus 100 applies the first to third cell measurement frequencies 810 to the first cell 802 to determine the impedance distribution 820 of the first cell, and also measures the first cell. The first to third cell measurement frequencies 810, which are the same as the frequencies used in ) is applied to the Nth cell 806 to determine the impedance distribution 824 of the Nth cell. That is, the cell measurement frequencies 810 applied to the cells 802, 804, and 806 are the same.

The diagnostic device 100 measures the module impedance 840 of the battery module 800 using at least one module measurement frequency 830 . As an example, the cell measurement impedance 810 and the module measurement impedance 830 may be identical to or different from each other. For example, when obtaining an impedance distribution graph as shown in FIG. 7, the distribution graph of cell impedance and the distribution graph of module impedance may be different from each other. In this case, the cell measurement frequency 810 (eg, N different frequencies) Some or all of the frequencies constituting M and the module measurement frequency 830 (eg, M (which may be the same as or different from N) different frequencies) may be different from each other.

9 is a diagram showing an example of the configuration of a battery module diagnostic device according to an embodiment of the present invention.

Referring to FIG. 9, the battery module diagnosis apparatus 100 includes a measurement unit 900, a cell diagnosis unit 910, a module diagnosis unit 920, a frequency selection unit 930, a cell analysis model 940, and a module analysis. A model 950 and a learning unit 960 are included. As an example, the battery module diagnosis apparatus 100 may be implemented as a computing device including a memory, a processor, and an input/output device. In this case, each configuration may be implemented as software, loaded into a memory, and then executed by a processor. In addition, the diagnostic apparatus 100 according to the embodiment may omit some components or further include other components.

The measurement unit 900 measures the impedance of each cell by applying a cell measurement frequency to each cell through a welded portion between the cells. An example of a structure of a welded portion between cells measured by the measuring unit 900 is shown in FIG. 3 . As another embodiment, the measurement unit 900 may measure the impedance of the battery module by applying a module measurement frequency to the battery module. The cell measurement frequency and/or the module measurement frequency may include at least one or more different frequencies.

The cell diagnosis unit 910 determines the welding defect of each cell by inputting the impedance of each cell to the cell analysis model 940 having learned the relationship between the impedance distribution and the welding defect. As another embodiment, the cell diagnosis unit 910 may determine not only the welding defect of each cell but also the state of the cell through the cell analysis model 940 . An example of a cell analysis model 940 is shown in FIG. 5 .

The module diagnosis unit 920 determines the state of the battery module by inputting the impedance of each cell and the impedance of the battery module to the module analysis model 950 having learned the relationship between the impedance distribution and the battery state. An example of a module analysis model 920 is shown in FIG. 6 .

The frequency selection unit 930 measures the frequency corresponding to the semicircular shape, minimum point area or ohmic resistance area in the curve of the impedance distribution measured by applying the frequency of the predefined section to the learning battery module, cell measurement frequency and / or module measurement frequency can be selected. An example of this is shown in FIG. 7 .

The learning unit 960 trains and generates the cell analysis model 940 and the module analysis model 950 . If the cell analysis model 940 and the module analysis model 950 have been previously trained, the learning unit 960 may be omitted. As an example, a frequency applied to the learning battery module to generate a training data set for training the cell analysis model 940 and the module analysis model 950 may be a frequency selected by the frequency selection unit 930. As another embodiment, the measuring unit 900 charges and discharges the battery module to have the same SOC as the SOC of the learning battery module used when generating the training dataset, and then applies a cell measurement frequency and/or a module measurement frequency to the cell impedance. and/or module impedance.

The present invention can also be implemented as computer readable program codes on a computer readable recording medium. A computer-readable recording medium includes all types of recording devices in which data that can be read by a computer system is stored. Examples of computer-readable recording media include ROM, RAM, CD-ROM, magnetic tape, floppy disk, and optical data storage devices. In addition, the computer-readable recording medium may be distributed to computer systems connected through a network to store and execute computer-readable codes in a distributed manner.

So far, the present invention has been looked at mainly with its preferred embodiments. Those skilled in the art to which the present invention pertains will be able to understand that the present invention can be implemented in a modified form without departing from the essential characteristics of the present invention. Therefore, the disclosed embodiments should be considered from a descriptive point of view rather than a limiting point of view. The scope of the present invention is shown in the claims rather than the foregoing description, and all differences within the equivalent scope will be construed as being included in the present invention.

Claims (12)

In the diagnosis method of a battery module including a plurality of cells,
Measuring impedance of each cell by applying a cell measurement frequency to each cell through a welded portion between the cells; and
Including; identifying the welding defects of each cell by inputting the impedance of each cell to the cell analysis model that has learned the relationship between the impedance distribution and the welding defect in each cell;
The measuring step is
Including; measuring the impedance of the two cells to determine whether or not the welding region between the two cells is defective,
The determining step includes the step of inputting the impedance of the two cells into the cell analysis model to determine whether or not a welded portion between the two cells is defective,
The cell analysis model is implemented as an artificial neural network trained by a supervised learning method using a training dataset including cell impedances labeled as normal or defective welding,
and making the SOC value of the battery module equal to the SOC value of the learning battery module used to create the training dataset of the cell analysis model before the measuring step. According to claim 1,
A method for diagnosing a battery module, further comprising the steps of inputting the impedance of each cell to a cell analysis model that has learned the relationship between the impedance distribution on a cell basis and the cell state, and identifying the state of each cell. The method of claim 1, wherein the measuring step,
A method for diagnosing a battery module, comprising: inputting a plurality of cell measurement frequencies of different frequencies to a cell to measure an impedance distribution. According to claim 3,
A method of diagnosing a battery module, characterized in that the cell measurement frequencies applied to each cell are the same. The method of claim 1, wherein the measuring step,
Selecting a frequency corresponding to a semicircular shape, minimum point area, or ohmic resistance area in the impedance distribution curve measured by applying the frequency of a predefined section to the learning battery module as a cell measurement frequency and applying it to each cell; A method for diagnosing a battery module, characterized in that for. According to claim 1,
measuring impedance by applying at least one module measurement frequency to the battery module; and
The step of determining the state of the battery module by inputting the impedance of each cell and the impedance of the battery module to the module analysis model that has learned the relationship between the impedance distribution and the battery state; of the battery module, characterized in that it further comprises diagnosis method. According to claim 6,
The method of diagnosing a battery module, characterized in that the cell measurement frequency and the module measurement frequency are different frequencies. a measurement unit for measuring impedance of each cell by applying a cell measurement frequency to each cell through a welded portion between the cells; and
A cell diagnosis unit for determining the welding defect of each cell by inputting the impedance of each cell to the cell analysis model that has learned the relationship between the impedance distribution and the welding defect;
The measurement unit measures the impedance of the two cells in order to determine whether or not the welding portion between the two cells is defective,
The cell diagnosis unit inputs the impedance of the two cells into the cell analysis model to determine whether or not a welding portion between the two cells is defective,
The cell analysis model is implemented as an artificial neural network trained by a supervised learning method using a training dataset including cell impedances labeled as normal or defective welding,
The battery module diagnostic device, characterized in that the SOC value of the battery module is the same as the SOC value of the learning battery module used to create the training dataset of the cell analysis model. According to claim 8,
A frequency selection unit for selecting a frequency corresponding to a semicircular shape, minimum point area, or ohmic resistance area in the impedance distribution curve measured by applying the frequency of a predefined section to the learning battery module as the cell measurement frequency; further comprising Battery module diagnostic device characterized by. According to claim 8,
The battery module diagnostic device further comprising a module diagnosis unit for determining the state of the battery module by inputting the impedance of each cell and the impedance of the battery module to the module analysis model having learned the relationship between the impedance distribution and the battery state. . According to claim 10,
The battery module diagnosis device, characterized in that the cell measurement frequency and the module measurement frequency for measuring the impedance of the battery module are different frequencies. A computer-executable recording medium on which a computer program for performing the method according to claim 1 is recorded.

Images