KR102553873B1 - System and method for generating battery diagnosis data using AI model based on image - Google Patents

Abstract

A system for generating data for battery diagnosis according to the present invention comprises: an artificial intelligence (AI) unit learning training data that consists of a pair of image data representing the characteristic resistance and image data representing the pseudo-open circuit voltage (pOCV) corresponding to the characteristic resistance; a profile generation unit generating profile resistance values which are different types of characteristic resistances for target batteries to which a test current is applied; a pre-processing unit generating first image data representing the profile resistance value; a main processing unit controlling the AI unit to generate second image data, or image data representing pOCV as data corresponding to the first image data by using the trained AI model; and a data processing unit processing the second image data to generate pOCV characteristic data of the target battery. Accordingly, a pOCV profile can be rapidly and accurately estimated and generated.

Description

System and method for generating battery diagnosis data using AI model based on image}

The present invention relates to a system for generating data for diagnosing a battery, and more specifically, to a battery characteristic resistance according to a short-time power supply, image data representing the characteristic resistance, and an artificial intelligence model using the characteristic resistance as an input element. A system and method for precisely estimating and generating data on pOCV that can be used as a standard reference for diagnosing characteristics.

As the demand for portable electronic products such as laptops, video cameras, and mobile phones that use electricity as a driving source increases rapidly, and as mobile robots, electric bicycles, electric carts, and electric vehicles become common, high-performance secondary batteries capable of repeated charging and discharging are possible. Research on batteries is actively progressing.

Commercially available secondary batteries include nickel cadmium batteries, nickel hydrogen batteries, nickel zinc batteries, and lithium secondary batteries. It has the advantage of a very low discharge rate, high energy density and high operating voltage, so it has been studied more intensively than other types of secondary batteries and has been applied more extensively to actual products.

Recently, secondary batteries have been widely used not only in small devices such as portable electronic devices, but also in medium and large devices such as electric vehicles and energy storage systems (ESSs).

Unlike liquid or gaseous fossil fuels, these secondary cell-based batteries generate (charge) electrical energy based on electrochemical reactions and consume (discharge) the generated energy, so the capacity of the battery is measured in a physical way. It is impossible to do.

Therefore, the remaining capacity of the battery (SOC, State of Charge) or SOH (State of Health), which quantitatively indicates the degree of battery capacity deterioration, is estimated by a mapping method using an open-circuit voltage value and a mapping table, current integration method, and internal resistance. It is estimated through various mathematical operations, such as the relative ratio calculation method of SOC according to environmental conditions such as temperature and temperature.

Estimating or diagnosing status information such as SOC or SOH, and feeding back the results is an important factor at all times, such as available capacity and available output, and is particularly important in electric vehicles (EVs) that must provide driving distance. can be said to be important.

In the case of an electric vehicle, it is essential to drive over an appropriate distance, as well as electric and electronic equipment, heaters, and air conditioners, so almost all devices in the car are powered by batteries, so high-power and large-capacity batteries are essential. It becomes so expensive that it accounts for a significant portion of the vehicle price.

The high price of such a vehicle (or medium or large-sized) battery has become an important obstacle to the commercialization of electric vehicles, along with long charging time and efficiency problems.

In order to solve this problem, recently, a service model in which a battery is replaced, shared, or subscribed is attracting attention. In order for this service model to expand, policy tasks must be addressed together, but above all, a value judgment on the remaining life of the current battery must be made first.

After the battery has been used for a certain period of time, its usefulness decreases due to capacity deterioration and output deterioration, and safety also acts as a risk factor. In order to improve efficiency, it is necessary to quickly and accurately diagnose the SOH of the current battery and predict the remaining life accordingly.

However, in the conventional method of estimating SOH using the capacity degradation state compared to the initial state, it is difficult to utilize objective indicators because other degradation pathways may be expressed at any time in future cycle life behavior.

In addition, in the case of the method using pOCV (pseudo open circuit voltage), current must be supplied for a long time to generate actual pOCV of the battery, so it is difficult to implement promptness or immediateness, as well as energy inefficiency issues due to long-term current supply is necessarily involved, so there is a fundamental limit to being a market-friendly method.

The present invention was devised to solve the above-mentioned problems in the background as described above, and using a characteristic resistance calculated by a test current (pulse current) applied for a short time and an image-based artificial intelligence model using this as an input element, a battery, The purpose is to provide a data generation system and method for battery diagnosis that can quickly and precisely estimate and generate pOCV data that can be used as important reference data for diagnosing remaining life, SOH, etc.

Other objects and advantages of the present invention can be understood by the following description, and will be more clearly understood by the embodiments of the present invention. In addition, the objects and advantages of the present invention can be realized by the configuration shown in the claims and the combination of the configuration.

In order to achieve the above object, the battery diagnosis data generation system of the present invention learns training data representing a pair of image data representing characteristic resistance and image data representing pseudo open circuit voltage (pOCV) corresponding to characteristic resistance. artificial intelligence department; a profile generator for generating profile resistance values, which are different types of characteristic resistance, for the target battery to which the test current is applied; a pre-processing unit generating first image data representing the profile resistance value; a main processing unit controlling the artificial intelligence unit to generate second image data, which is image data representing a pseudo open circuit voltage (pOCV), as data corresponding to the first image data by using the learned artificial intelligence model; and a data processor generating pOCV characteristic data of the target battery by processing the second image data.

Here, the characteristic resistance of the present invention may be one or more of a resistance component ratio and a resistance growth rate determined by the following formula.

In the above formula, RCR is the resistance component ratio, R _a1 is The target battery resistance value, R _a2 , at a first time point when the first critical change occurred in the voltage behavior of the target battery according to the application of the test current is the target battery at a second time point after the first time point The resistance value, RGR, is the resistance growth rate, and R _i2 is the resistance value of the new battery having the same specifications as the target battery at the second time point.

Depending on the embodiment, the characteristic resistance of the present invention may be one or more of the resistance component ratio, the resistance growth rate, and the resistance vector size determined by the following formula.

In the above equation, RVS is the magnitude of the resistance vector, and R _i1 is the resistance value of the new battery at the first point in time.

Preferably, the second time point may be n (n is a natural number equal to or greater than 1) seconds after the first point in time.

Specifically, the pre-processing unit of the present invention may be configured to generate the first image data by setting at least one of the resistive component ratio and the resistive growth rate to at least one of color space data of the first image data.

The pre-processing unit of the present invention may be configured to generate the first image data by setting at least one of the resistive component ratio, the resistive growth rate, and the resistive vector size to at least one of color space data of the first image data. there is.

Furthermore, the artificial intelligence unit of the present invention may be configured to generate the second image data so that color space data representing voltage values have time-series values according to directions of pixels constituting the image data.

Preferably, the data processing unit of the present invention generates pOCV characteristic data of the target battery using media data, which is color space data included in the second image data, and unit SOC divided into unit sizes according to the number of media data. can be configured to

According to another aspect of the present invention, a data generation method for diagnosing a battery includes an artificial intelligence learning step of learning training data including image data representing characteristic resistance and image data representing pseudo open circuit voltage (pOCV) corresponding to the characteristic resistance. ; A profile generating step of generating profile resistance values, which are different types of characteristic resistance, for the target battery to which the test current is applied; a preprocessing step of generating first image data representing the profile resistance value; a second image generation step of generating second image data, which is image data representing a pseudo open circuit voltage (pOCV), as data corresponding to the first image data by using the learned artificial intelligence model; and a data processing step of generating pOCV characteristic data of the target battery by processing the second image data.

According to an embodiment of the present invention, a pOCV profile equivalent to the actual pOCV of a corresponding battery can be quickly and accurately estimated and generated with a resistance component calculated by a pulse current applied for a short time of about several seconds.

In addition, in the case of the present invention, organic convergence processing with an artificial intelligence model based on image data can be optimized by applying processing to process a characteristic resistance element and graft it to image data, thereby providing data related to battery characteristics. It is possible to generate a more reliable estimate of .

According to an embodiment of the present invention, the voltage profile of a battery can be modeled as time-series data by serializing result data of an artificial intelligence model through mathematical processing, thereby providing a basis for more diverse subsequent application processing.

Furthermore, in the case of the present invention, since pOCV data of a battery can be estimated and generated only with current power applied for a short time, the problem of long-term current application required to check the actual pOCV and consequent energy inefficiency can be fundamentally solved. .

Diagnostic data such as aging, remaining useful life (RUL, Remaining Useful Life), SOH, etc., which are important criteria for diagnosis, evaluation, replacement, circulation, and recycling of batteries, based on the unique characteristics of the present invention. can be generated more economically and quickly, providing an application solution that is more optimized for building a battery circulation ecosystem.

The following drawings attached to this specification illustrate preferred embodiments of the present invention, and together with the detailed description of the present invention serve to more effectively understand the technical idea of the present invention, the present invention is described in these drawings should not be construed as limited to
1 is a block diagram showing the detailed configuration of a data generating system according to a preferred embodiment of the present invention;
2 is a diagram schematically illustrating the logical structure of the AI model included in the artificial intelligence unit of FIG. 1;
3 is a flowchart illustrating a processing procedure according to a preferred embodiment of the present invention for generating pOCV characteristic data;
4 is a flowchart illustrating a processing procedure according to an embodiment of the present invention for generating a characteristic resistance of a battery;
5 is a flowchart illustrating one embodiment of processing in which pOCV characteristic data is generated;
6 is a diagram showing behavior characteristics of a battery according to application of a pulse current;
7 is a diagram showing the reliability result of processing according to the present invention based on the capacity loss rate.

Hereinafter, preferred embodiments of the present invention will be described in detail with reference to the accompanying drawings. Prior to this, the terms or words used in this specification and claims should not be construed as being limited to the usual or dictionary meaning, and the inventor appropriately uses the concept of the term in order to explain his/her invention in the best way. It should be interpreted as a meaning and concept consistent with the technical idea of the present invention based on the principle that it can be defined.

Therefore, since the embodiments described in this specification and the configurations shown in the drawings are only one of the most preferred embodiments of the present invention and do not represent all of the technical ideas of the present invention, various equivalents that can replace them at the time of the present application It should be understood that there may be waters and variations.

As shown in FIG. 1, the data generation system 100 of the present invention includes a pulse current application unit 110, a sensing unit 120, a profile generation unit 130, a pre-processing unit 140, a main processing unit 150, It may be configured to include a data processing unit 160, an artificial intelligence unit 170, a learning DB unit 175 and an estimation diagnosis unit 180.

Prior to the detailed description of the present invention, each component of the data generation system 100 of the present invention shown in FIG. 1 should be understood as a logically separated component rather than a physically separated component.

That is, since each component corresponds to a logical component for realizing the technical idea of the present invention, even if each component is integrated or separated, if the function performed by the logical configuration of the present invention can be realized, the scope of the present invention It should be interpreted that it is within the scope, and any component performing the same or similar function should be construed as within the scope of the present invention regardless of whether or not the name is identical.

Meanwhile, the data generation method of the present invention may be associated with hardware configuration, but since it uses information or data output from the hardware configuration, it is not directly limited to specific hardware.

Since the data generation method of the present invention is implemented as a set of processing or algorithms related to data processing, processing, operation, etc. as a whole, it is implemented by a combination of logical configurations shown in FIG. Of course, it can be implemented in the form of software to be driven.

The pulse current applying unit 110 of the present invention applies a test current such as a pulse current to a battery (hereinafter referred to as a 'target battery') to which diagnostic data is generated (S310, see FIG. 3). .

Secondary cell-based batteries may be implemented as battery cells, cell assemblies, battery modules, battery packs, etc. by various combinational configurations according to application fields or embodiments. Since data for determining characteristics is generated, in the following description, a battery may be an object commonly referred to as a battery cell, a cell assembly, a battery module, or a battery pack.

In this way, when the test current is applied to the target battery 10, the profile generating unit 130 of the present invention detects and processes the sensing unit 120 to thirty different types of characteristic resistance of the target battery 10. resistance) is generated (S320).

Since the battery is based on an electrochemical reaction, etc., as its use continues, deterioration such as damage to the cathode (cathode) and loss of active lithium occurs, resulting in an increase in internal resistance (IR).

In addition, the internal resistance of the battery may change rapidly due to external factors such as temperature environment, current rate (C-rate), and depth of discharge (DOD), and lithium precipitation due to reaction/diffusion restrictions of active lithium Resistance increase due to development, ion diffusion by foreign substances formed on the surface of separators and electrodes, crack damage to electrodes, resistance increase due to electrolyte depletion, etc., charge transfer due to formation of a solid electrolyte interphase (SEI) at the electrode-electrolyte interface resistance increases, etc.

In this way, since the resistance component of the battery or its change directly or indirectly represents the intrinsic state of the battery due to use, the present invention provides the characteristic resistance and A method of estimating the intrinsic state of the battery, for example, pOCV, using the relative change between them as an index is applied.

Specifically, the characteristic resistance may be at least one of a resistance component ratio and a resistance growth rate determined by the following formula.

In the above formula, RCR is the resistance component ratio, RGR is the resistance growth rate, R _a1 is the target battery resistance value at the first time point (t ₁ ) when the first critical change occurs among the voltage behavior of the target battery according to the application of the test current, R _a2 is the target battery resistance value at the second time point (t ₂ ), which is the time point after the first time point (t ₁ ), and R _i2 is the second pristine battery having the same or equivalent specifications as the target battery. It is the resistance value at the time point (t ₂ ). R _a2 And R _i2 may be set to a resistance value measured from the first time point to the second time point according to the embodiment.

Depending on the embodiment, the characteristic resistance may include a resistance vector size determined by the following formula.

In the above formula, RVS is the magnitude of the resistance vector, and R _i1 is the resistance value of the new battery at the first time point (t ₁ ).

Details related to the above-described equations are additionally described with reference to FIG. 4 illustrating a processing process related to generation of characteristic resistance of the battery and FIG. 6 illustrating behavior characteristics of the battery according to application of a test current (pulse current).

When a test current is applied to the target battery 10 (S400), the behavior characteristic of the voltage increasing over time is shown. As shown in FIG. 6 by the inherent characteristics of the battery, the voltage behavior of the target battery 10 within a few seconds A critical change occurs in (S410).

The time point at which the critical change occurs is the first time point t ₁ , and the resistance value of the target battery 10 at the first time point t ₁ becomes R _a1 (S420). Based on the embodiment of FIG. 6 , when the magnitude of the test current is 2 mA, R _a1 becomes 110 Ω (220 mV/2 mA) according to Ohm's law.

The second time point t ₂ is a reference time point at which changes in voltage behavior are stabilized after the first time point t ₁ , and may be set to several seconds or more based on the first time point t ₁ . In this way, when the reference time elapses at the first time point t ₂ (S430), the resistance value at the second time point t ₂ is measured (S440).

R _a2 is the resistance value of the target battery 10 at the second time point t ₂ , which is 132Ω in the embodiment of FIG. 6, and is the same as or equivalent to the target battery 10. At the second point (t ₂ ), the resistance value R _i2 becomes 65Ω (130mV/2mA), and R _i1 That is, the resistance value of the new battery at the first time point (t ₁ ) becomes 50Ω (100mV/2mA).

The first time point (t ₁ ) resistance value and the second time point (t ₂ ) resistance value of the new battery can be measured in the same environment and conditions as the target battery 10, as well as provided during the manufacturing process of the battery. It can also be secured through existing specification (spec,) information, etc.

Therefore, the process of acquiring the resistance value at the first time point (t ₁ ) and the second time point (t ₂ ) of the new battery (S450) is to measure the resistance value of the target battery 10 as illustrated in FIG. It is not limited to the later form.

Based on the embodiment of FIG. 6, the resistance component ratio (RCR), resistance growth rate (RGR), and resistance vector size (RVS) defined in the present invention are calculated as follows.

As such, the profile generator 130 of the present invention generates one or more of the data sets consisting of [0.2, 2, 60.4] corresponding to RCR, RGR, and RVS as the profile resistance value of the target battery 10 (S320). , S460).

When the profile resistance value of the target battery 10 is generated, the pre-processing unit 140 of the present invention generates image data (hereinafter referred to as 'first image data') representing the profile resistance value (S330, S500). see Figure 5).

Specifically, the pre-processing unit 140 of the present invention sets one or more of the resistance component ratio (RCR), the resistance growth rate (RGR), and the resistance vector size (RVS) to one or more of the color space data of the first image data, 1 Can be configured to generate image data.

The color of pixels constituting image data can be expressed in various color spaces such as RGB, CMYK, CIE, and YCbCr according to criteria such as color mixing and similarity to the human visual system. Conversion between different color spaces can be performed by a non-complex mathematical conversion equation.

For example, when RGB is used as a color space, the preprocessing unit 140 of the present invention sets one or more of a resistance component ratio (RCR), a resistance growth rate (RGR), and a resistance vector size (RVS) to one of (R, G, B). With the above settings, the first image data is generated.

The size of the first image data may be variously set in consideration of an interfacing environment with an artificial intelligence model. In addition, of course, various modifications are possible, such as the profile resistance value being included in all pixels or some pixels of the first image data, or the profile resistance value being composed of a single pixel.

Depending on the embodiment, for example, if the resistance component ratio and resistance growth rate are set for two components of R, G, and B, a default value according to a predetermined standard may be set for the other one of R, G, and B. .

The present invention is configured to estimate and generate pOCV data corresponding to a characteristic resistance value, that is, a profile resistance value, using artificial intelligence (AI).

In order to increase the application efficiency among various methodologies for implementing artificial intelligence and to further improve the precision and reliability of mutual matching between input and output elements, the present invention is an image-based artificial intelligence model created by mutually converting an input image domain and an output image domain. It can be configured to apply.

To this end, the artificial intelligence unit 170 of the present invention learns training data in which image data representing the characteristic resistance and image data representing the pseudo open circuit voltage (pOCV) corresponding to the characteristic resistance form a pair (S300). ) is made up of

The training data may be updated and stored in the learning DB unit 175 of the present invention, and result data calculated by the artificial intelligence unit 170 may also be used as training data cyclically.

The pseudo open voltage constituting the training data is data generated based on the voltage actually observed through the application of current for a long time, etc., and is clustered into one or more aggregates along with the characteristic resistance classified according to the temperature environment, the battery usage environment, and the degree of aging of the battery. can get angry

When the pseudo open circuit voltage is observed, it undergoes sampling processing and/or normalization processing based on a predetermined unit, and image data representing the pseudo open circuit voltage is generated through a process of mapping the result values to each pixel of the image data.

As one configuration of training data, the characteristic resistance forming a pair with the pseudo open circuit voltage and image data representing this characteristic resistance can be generated in a method corresponding to the profile resistance value and the first image data described above is of course

2 is a diagram schematically illustrating the logical structure of the AI model 200 loaded in the artificial intelligence unit 170.

The AI model 200 that can be applied to the present invention includes a generator G (210), a discriminator Y (220), a generator F (230), and a discriminator X (240). ), and domain conversion can be achieved by generating an output domain through the generator G 210 when an input domain is input after learning is completed in the AI model 200.

The AI model 200 of the present invention is a model that generates a required output domain based on an input domain, and a plurality of generators and separators form a circular structure to perform learning processing.

x의 형태로 순환 일관성(cycle consistency)을 유지하도록 설계될 수 있다.Specifically, when real pulse resistance image data (Image_Real pulse resistance) 212, which is image data representing actual characteristic resistance, is input, generator G 210 converts the simulated voltage image data (Image_simulated voltage) 214 into a real voltage image. Data (Image_Real voltage) 232 is learned to be similarly generated, and the delimiter Y 220 is learned to discriminate the fake voltage image data by comparing the fake voltage image data 214 and the real voltage image data 232. When this learning process is simply schematized, x → G(x) → F(G(x))

It can be designed to maintain cycle consistency in the form of x.

As described above, when the first image data representing the profile resistance value is generated by the pre-processing unit 140 of the present invention, the artificial intelligence unit 170 of the present invention uses the artificial intelligence model learned as described above. As data corresponding to the first image data, image data (hereinafter referred to as 'second image data') representing the pseudo open circuit voltage (pOCV) is generated (S340 and S510).

In order to increase the efficiency of data processing, etc., the main processing unit 150 may be configured to generate the second image data by controlling the artificial intelligence unit 170 according to an embodiment.

In this case, the artificial intelligence unit 170 of the present invention may be configured to generate second image data so that color space data representing voltage values have time-series values according to the directionality of pixels constituting the image data.

As illustrated in FIG. 5 , the second image data is preferably generated in a mapped form with color space data having a direction such as from left to right. According to this implementation configuration, it is possible to improve the intuitive recognition ability as well as increase the efficiency of calculation processing for rendering of the pOCV model.

Of course, the above description of configuring the color space data of the image data representing the pseudo open voltage to have directionality can be equally applied to the process of learning the training data described above.

When the second image data is generated in this way, the data processing unit 160 of the present invention performs post-processing processing (sampling processing, scaling processing, graphic rendering processing, etc.) on the second image data so that the target battery 10 Generates pOCV characteristic data for (S350, S530).

Specifically, the data processing unit 160 of the present invention uses the color space data included in the second image data and the unit SOC divided into unit sizes according to the number of color space data included in the second image data to target battery 10 ) It can be configured to generate pOCV characteristic data.

For example, if the number of color space data included in the second image data is 200, the SOC having a range of 0 to 100% is divided into 200 equal parts to determine the individual unit SOC, set it on the coordinate axis (X-axis), and The pOCV characteristic data may be generated by assigning each value of the color space data included in the image data to a Y coordinate value according to the position of each of the coordinate axes (X axis) divided into 200 pieces.

When the pOCV characteristic data of the target battery 10 is generated, the estimation diagnosis unit 180 of the present invention estimates the SOH of the target battery 10 using the generated pOCV characteristic data and outputs information data about it. .

7 is a diagram showing reliability results of processing according to the present invention based on initial capacity loss ratio (SOH Loss). The Y-axis of FIG. 7 is the Predicted Capacity Loss calculated based on the pOCV generated by the present invention, and the X-axis of FIG. 7 is the True Capacity Loss calculated based on the actual pOCV.

As confirmed in FIG. 7, they converge to a linear distribution with a slope of 1, and according to the experimental results, the mean absolute error (MAE) is 0.118 and the coefficient of determination (R ² ) is 96.21. Since it is expressed as %, the capacity loss rate based on the pOCV estimated within a few seconds to several tens of seconds through the configuration and processing of the present invention can have a high level of reliability equivalent to the capacity loss rate based on the actual pOCV measured for a long time. there is.

Although the present invention has been described above with limited examples and drawings, the present invention is not limited thereto and will be described below and the technical spirit of the present invention by those skilled in the art to which the present invention belongs. Of course, various modifications and variations are possible within the scope of the claims.

In the description of the present invention described above, modifiers such as first and second are only terms of instrumental concepts used to relatively distinguish components from each other, so they are used to indicate a specific order, priority, etc. It should be interpreted that it is not a term that

Although the accompanying drawings and the like for illustration of the description of the present invention and its embodiments may be shown in a slightly exaggerated form in order to emphasize or highlight the technical contents according to the present invention, the above-described contents and matters shown in the drawings Taking into account, it should be interpreted that it is obvious that various types of modifications can be applied at the level of those skilled in the art.

The method for generating data for diagnosing a battery according to the present invention described above may be implemented as computer readable codes on a computer readable recording medium. Computer-readable recording media include all types of recording devices (CD-ROM, RAM, ROM, floppy disk, magnetic disk, hard disk, magneto-optical disk, etc.) in which data that can be read by a computer is stored. It also includes a server for transmission.

100: data generation system 110: pulse current application unit
120: sensing unit 130: profile generating unit
140: pre-processing unit 150: main processing unit
160: data processing unit 170: artificial intelligence unit
175: learning DB unit 180: estimation diagnosis unit

Claims (10)

an artificial intelligence unit learning training data representing a pair of image data representing a characteristic resistance and image data representing a pseudo open circuit voltage (pOCV) corresponding to the characteristic resistance;
a profile generator for generating profile resistance values, which are different types of characteristic resistance, for the target battery to which the test current is applied;
a pre-processing unit generating first image data representing the profile resistance value;
a main processing unit controlling the artificial intelligence unit to generate second image data, which is image data representing a pseudo open circuit voltage (pOCV), as data corresponding to the first image data by using the learned artificial intelligence model; and
and a data processing unit processing the second image data to generate pOCV characteristic data of the target battery. The method of claim 1, wherein the characteristic resistance,
Data generation system for battery diagnosis, characterized in that at least one of the resistance component ratio and the resistance growth rate determined by the formula below.

In the above formula, RCR is the resistance component ratio, R _a1 is The target battery resistance value, R _a2 , at a first time point when the first critical change occurred in the voltage behavior of the target battery according to the application of the test current is the target battery at a second time point after the first time point The resistance value, RGR, is the resistance growth rate, and R _i2 is the resistance value of the new battery having the same specifications as the target battery at the second time point. The method of claim 2, wherein the characteristic resistance,
A data generation system for battery diagnosis, characterized in that at least one of the resistance component ratio, the resistance growth rate, and the resistance vector size determined by the following formula.

In the above equation, RVS is the magnitude of the resistance vector, and R _i1 is the resistance value of the new battery at the first point in time. The method of claim 2, wherein the second time point,
Battery diagnostic data generation system, characterized in that n (n is a natural number of 1 or more) seconds after the first point in time. The method of claim 2, wherein the pre-processing unit,
and generating the first image data by setting at least one of the resistance component ratio and the resistance growth rate to at least one of color space data of the first image data. The method of claim 3, wherein the pre-processing unit,
wherein the first image data is generated by setting one or more of the resistance component ratio, the resistance growth rate, and the resistance vector size to one or more of color space data of the first image data. The method of claim 1, wherein the artificial intelligence unit,
The battery diagnostic data generation system characterized in that the second image data is generated so that the color space data representing the voltage value has a time-series value according to the directionality of pixels constituting the image data. The method of claim 1, wherein the data processing unit,
Battery diagnosis data generation system characterized in that for generating the pOCV characteristic data of the target battery using the color space data recorded in the second image data and the unit SOC divided into unit sizes according to the number of the media data. . An artificial intelligence learning step of learning training data including image data representing characteristic resistance and image data representing pseudo open circuit voltage (pOCV) corresponding to the characteristic resistance;
A profile generating step of generating profile resistance values, which are different types of characteristic resistance, for the target battery to which the test current is applied;
a preprocessing step of generating first image data representing the profile resistance value;
a second image generation step of generating second image data, which is image data representing a pseudo open circuit voltage (pOCV), as data corresponding to the first image data using the learned artificial intelligence model; and
and a data processing step of processing the second image data to generate pOCV characteristic data of the target battery. A computer-readable recording medium on which a program for performing the method according to claim 9 is recorded.

Images