Apparatus and method for predicting maximum temperature

Abstract

An apparatus for predicting a maximum temperature may include: a preprocessing device including a first processor and a first memory storing a plurality of programs; and a temperature determination device including a second memory and a temperature prediction learning device. The first processor may generate input data including cell design parameters for constituting candidate battery cells of the candidate battery module and simulation prediction conditions for predicting a simulation of a maximum temperature of the candidate battery cells. The second memory may store input data, and during a learning operation, the temperature prediction learning apparatus may perform machine learning by using training data among the input data stored in the second memory, and determine a temperature calculation formula that predicts a highest temperature of a candidate battery cell through the machine learning.

Description

Technical Field

[0001] Cross - reference to related applications

[0002] This application claims the priority and benefit of Korean Patent Application No. 10 - 2023 - 0079003, filed with the Korean Intellectual Property Office on June 20, 2023, the entire content of which is incorporated herein by reference.

[0003] The present disclosure relates to an apparatus and method for predicting a maximum temperature. Background Art

[0004] The maximum battery temperature is one of the main factors determining battery life. Battery manufacturers can measure the maximum temperatures of multiple candidate batteries that meet the required specifications. Due to cost and time limitations, it is not practical to actually manufacture multiple candidate batteries to measure the maximum temperature.

[0005] To overcome this, performance tests are performed on the battery cells that make up each of the multiple candidate batteries, various conditions are set to interpret the performance test results, and an analytical simulation is performed based on the set conditions to predict the maximum battery temperature. Through the above - mentioned series of processes, predicting the maximum temperature of each candidate battery takes a considerable amount of time and cost. Since the prediction of the maximum temperatures of multiple candidate batteries may not be performed simultaneously, predicting the maximum temperatures of all multiple candidate batteries takes even more time and cost. In this process, there may be a problem that the prediction of the maximum temperature is performed even on unnecessary candidate batteries. In addition, based on the analytical simulation results, it takes a considerable amount of time to review the performance of each candidate battery. Summary of the Invention

[0006] [Technical Problem]

[0007] The present disclosure attempts to provide an apparatus and method for predicting the maximum temperature of a battery.

[0008] [Technical Solution]

[0009] According to an aspect of the present disclosure, an apparatus for predicting a maximum temperature includes: a pre - processing device including a first processor and a first memory storing multiple programs; and a temperature determination device including a second memory and a temperature prediction learning device. The first processor may generate input data including cell design parameters for cell units that make up a candidate battery module and simulation analysis conditions for simulating the maximum temperature of the candidate cell units. The second memory may store the input data, and during a learning operation, the temperature prediction learning device may perform machine learning by using training data among the input data stored in the second memory and determine a temperature calculation formula for predicting the maximum temperature of the candidate cell units through machine learning.

[0010] The temperature determination device may further include: a second processor configured to control a second memory to provide training data to the temperature prediction training device during a learning operation, and control the temperature prediction training device to perform machine learning using the training data.

[0011] After the learning operation, the second processor may perform a normal operation of determining the maximum temperature of the candidate battery cell according to the updated temperature calculation formula.

[0012] The temperature prediction learning device may perform a learning operation to determine a plurality of first to nth parameters that affect the maximum temperature among the training data and a plurality of first to nth parameter coefficients corresponding to the plurality of first to nth parameters respectively. The training data may include the cell design parameters of the candidate battery cell, the simulation analysis conditions, and the maximum temperature.

[0013] The simulation analysis conditions of the training data may include the maximum retention time for the maximum current of the candidate battery cell.

[0014] The temperature calculation formula may be a regression formula of terms composed of the products of the plurality of first to nth parameters and the plurality of first to nth parameter coefficients.

[0015] The first memory may store an RC extraction program for deriving the resistance * capacitance (RC) value of the candidate battery cell. The first processor may derive the RC value of the equivalent circuit corresponding to the candidate battery cell by applying the hybrid pulse power characterization (HPPC) test data for the candidate battery cell to the RC extraction program.

[0016] The first memory may store a maximum temperature analysis program for determining the maximum temperature of the candidate battery cell. The first processor may determine the maximum temperature by applying the cell design parameters, the RC value of the equivalent circuit corresponding to the candidate battery cell, and the simulation analysis conditions to the maximum temperature analysis program.

[0017] The first memory may store a candidate derivation program that determines cell design parameters based on the cell design data of the required specifications and derives a candidate battery module according to the cell design parameters. The first processor may derive a candidate battery module by applying the cell design data to the candidate derivation program.

[0018] According to another aspect of the present disclosure, a method for predicting the maximum temperature includes: deriving the RC value of the equivalent circuit corresponding to the candidate battery cell constituting the candidate battery module; determining the maximum temperature of the candidate battery cell by using the cell design parameters, the RC value, and the simulation analysis conditions for predicting the maximum temperature of the candidate battery cell; and determining a temperature calculation formula for predicting the maximum temperature of the candidate battery cell by performing machine learning using training data including the cell design parameters, the RC value, and the simulation analysis conditions.

[0019] The method may further include: after the learning operation, determining the maximum temperature of the candidate battery cell according to the updated temperature calculation formula.

[0020] Determining the temperature calculation formula may include determining a plurality of first to nth parameters that affect the maximum temperature among the training data and a plurality of first to nth parameter coefficients corresponding to the plurality of first to nth parameters respectively.

[0021] The temperature calculation formula may be a regression formula of terms composed of the products of the plurality of first to nth parameters and the plurality of first to nth parameter coefficients.

[0022] The training data may include the maximum retention time for the maximum current of the candidate battery cell.

[0023] Deriving the RC value may include deriving the RC value by applying the HPPC test data for the candidate battery cell to an RC extraction program.

[0024] The method may further include determining monomer design parameters based on the monomer design data of the required specifications.

[0025] [Beneficial Effects]

[0026] Devices and methods for predicting the maximum temperature based on the monomer design parameters and analysis conditions required for predicting the maximum temperature may be provided. Description of the Drawings

[0027] Figure 1 is a diagram showing a device for predicting the maximum temperature according to an embodiment.

[0028] Figure 2 is a flowchart showing a preprocessing method for temperature prediction learning according to an embodiment.

[0029] Figure 3 is a flowchart showing a temperature prediction learning method according to an embodiment.

[0030] Figure 4 is a line graph showing the maximum temperature predicted according to the number of stacks in the preprocessing device. Detailed Description of the Embodiments

[0031] When describing the embodiments disclosed in this specification, when it is determined that the detailed description of related known technologies may confuse the key points of the embodiments disclosed in this specification, the detailed description thereof will be omitted. In addition, the drawings are only for facilitating the understanding of the embodiments disclosed in this specification, and do not limit the technical concept disclosed in this specification, and should be understood to include all changes, equivalents or alternatives included in the spirit and scope of the present disclosure.

[0032] Terms including ordinal numbers such as first, second, etc. may be used to describe various components, but the components are not limited by the terms. The terms are only for the purpose of distinguishing one component from another.

[0033] It should be understood that when a component is referred to as "connected to" or "coupled to" another component, the component can be connected or coupled to other components, or there may be intermediate components. In contrast, when a component is referred to as "directly connected to" or "directly coupled to" another component, there are no intermediate components.

[0034] It should also be understood that when used in this specification, the terms "comprises" and / or "comprising" specify the presence of the stated features, integers, steps, operations, components, and / or parts, but do not preclude the presence or addition of one or more other features, integers, steps, operations, components, parts, and / or combinations thereof.

[0035] Figure 1 It is a diagram showing a device for predicting the maximum temperature according to an embodiment.

[0036] As Figure 1 shown, the device 1 for predicting the maximum temperature may include a preprocessing device 10 and a temperature calculation device 20. The preprocessing device 10 may preprocess at least a part of the input data to be provided to the temperature calculation device 20. For example, the input data may include analysis conditions (hereinafter referred to as simulation analysis conditions) for an analysis simulation applied to determine the maximum temperature and monomer design parameters. The monomer design parameters may include battery size (length, width, and thickness), the number of monomer stacks, the loading amount of the positive electrode material, the monomer capacity, the width / thickness ratio of the cathode / anode foil, the thickness / length of the cathode / anode tab, the width / thickness of the cathode / anode lead, and the SiO content. The simulation analysis conditions may include the initial temperature, the ambient temperature, the cooling water temperature, the rapid charging state of charge (SOC) range, the initial SOC, the heat transfer coefficient (with respect to the battery module), the internal thickness of the total wire, the maximum current, the rapid charging target time, the maximum current maximum retention time, etc. Rapid charging may mean performing charging using a high current equal to or greater than a certain value.

[0037] The preprocessing device 10 may include a processor 11 and a memory 12. The memory 12 may include a candidate derivation program 121, an RC extraction program 122, and a maximum temperature analysis program 123. The processor 11 may determine candidate battery cells constituting a candidate battery module by applying the monomer design data provided to the preprocessing device 10 to the candidate derivation program 121, derive the resistance-capacitance (RC) values of the candidate battery cells by applying the hybrid pulse power characterization (HPPC) test data for the candidate battery cells to the RC extraction program 122, and determine the maximum temperature of the candidate battery cells by applying the simulation analysis conditions and the RC values to the maximum temperature analysis program 123. The monomer design data is data regarding required specifications and may include the maximum current during charging, the time during which fast charging can be performed (hereinafter referred to as the fast charging target time), the battery pack size, and the like.

[0038] The device 1 for predicting the maximum temperature may obtain the HPPC test data from an externally provided emulator 2. The processor 11 may determine the candidate battery cells and provide the monomer design parameters for the candidate battery cells to the emulator 2. The emulator 2 may obtain the HPPC test data by performing a physics-based simulation based on the chemical information among the monomer design parameters of the candidate battery cells. For example, the emulator 2 may derive the voltage behavior of the candidate battery cells by performing a physics-based simulation based on the chemical information of the positive electrode material and the negative electrode material constituting the candidate battery cells. The HPPC test data may include data regarding the voltage behavior of the corresponding monomer.

[0039] First, the processor 11 may load the candidate derivation program 121 from the memory 12, input the monomer design data into the candidate derivation program 121, and execute the candidate derivation program 121. The candidate derivation program 121 may generate the monomer design parameters of the monomers to configure a battery module based on the monomer design data. For example, the candidate derivation program 121 may determine the size of the monomers to configure the battery module and the number of a plurality of monomers to be stacked (hereinafter, the number of monomer stacks) to configure the battery module. A battery module refers to a unit energy storage device realized by electrically connecting at least two battery cells. The candidate derivation program 121 may determine the size of the monomers according to the monomer design data and determine the number of monomer stacks that can be realized using the determined size of the monomers. The size of the monomer may be defined as the length of the monomer, the width of the monomer, and the thickness of the monomer. A battery pack may be realized by electrically connecting a plurality of battery modules. The monomer may include a positive electrode material load, a negative electrode material load, and foils attached to each of the positive electrode material load and the negative electrode material load. Considering the thicknesses of the positive electrode material load and the negative electrode material load, the candidate derivation program 121 may determine the number of stacks constituting the battery module.

[0040] The candidate export program 121 can export the dimensions of multiple monomers that meet the required specifications and the number of monomer stacks for each of the dimensions of the multiple monomers, and can export candidate battery modules in consideration of the production possibility of each of the multiple battery modules achieved by the exported dimensions of the multiple monomers and the exported number of monomer stacks for each of the dimensions of the multiple monomers. In the candidate export program 121, factors for determining production possibility and criteria can be installed.

[0041] In Figure 1 and with reference to Figure 1 it is described that the memory 12 includes the candidate export program 121 and the processor 11 executes the candidate export program 121, but the present disclosure is not limited thereto. The memory 12 does not store the candidate export program 121, and the processor 11 may receive information about candidate battery modules from the outside.

[0042] The processor 11 can load the RC export program 122 from the memory 12, input HPPC test data of monomers (candidate battery monomers) for candidate battery modules into the RC export program 122, and execute the RC export program 122. The RC export program 122 can implement an equivalent circuit representing the electrical characteristics of the corresponding monomer based on the input HPPC test data of the candidate battery monomers, and estimate the RC value of the candidate battery monomers based on the equivalent circuit. The RC export program 122 can implement an equivalent circuit based on the voltage behavior of the monomer according to the HPPC test data of the candidate battery monomers, and can determine the RC value of the candidate battery monomers by multiplying the resistance value and the capacitance value of the implemented equivalent circuit.

[0043] The processor 11 can load the maximum temperature analysis program 123 from the memory 12, input monomer design parameters, simulation analysis conditions, and the RC value of the candidate battery monomers into the maximum temperature analysis program 123, and execute the maximum temperature analysis program 123. The maximum temperature analysis program 123 can export the data necessary for temperature analysis among the monomer design parameters, and determine the maximum temperature by solving the control equation of the maximum temperature based on the exported data (e.g., the dimensions of the monomer, the temperature of the cooling water, the heat transfer coefficient, etc.) and the RC value. In the maximum temperature analysis program 123, the corresponding control equation can be installed in advance, the monomer design parameters and the RC value can be independent variables, and the maximum temperature can be the dependent variable in the control equation. The maximum temperature analysis program 123 can determine the maximum temperature by performing a calculation operation to solve the corresponding control equation.

[0044] The temperature determination device 20 according to the embodiment can receive input data from the preprocessing device 10, store the input data, perform a training operation by using the stored input data, and perform a normal operation of predicting the maximum temperature based on the input data provided from the preprocessing device 10 after the training operation.

[0045] The temperature determination device 20 may include a processor 201, a memory 202, and a temperature prediction learning device 203. The processor 201 may determine which operation between the normal operation and the learning operation is to be performed, control the operation of the temperature prediction learning device 203 when the determined operation mode is the learning operation, and perform temperature calculation when the determined operation mode is the normal operation.

[0046] The processor 201 may determine the operation mode according to an externally provided command, or may determine the operation mode according to preset specific control conditions and the operation modes corresponding to the specific control conditions. For example, the processor 201 may determine the operation mode as the learning operation every predetermined period or whenever the accumulated new training data reaches a certain reference amount.

[0047] In the learning operation, the processor 201 may instruct the temperature prediction learning device 203 to perform machine learning. The processor 201 may determine the training data among the input data stored in the memory 202, and control the memory 202 and the temperature prediction learning device 203 so that the temperature prediction learning device 203 can read the training data from the memory 202. The processor 201 may determine the data added after the most recent learning operation among the input data stored in the memory 202 as the training data. In the learning operation, the temperature prediction learning device 203 may perform machine learning to output a temperature calculation formula that determines the maximum temperature during single-cell charging by using the read training data. The temperature prediction learning device 203 may determine the temperature calculation formula through the learning operation and update the temperature calculation formula to the processor 201. In the learning operation, the processor 201 may control the maximum temperature to be included in the training data. That is, the temperature prediction learning device 203 may perform machine learning by receiving information related to the single-cell design parameters, information related to the maximum temperature simulation analysis conditions, and the predicted maximum temperature to output a temperature calculation formula for determining the maximum temperature.

[0048] The temperature prediction learning device 203 can perform a learning operation to determine a plurality of first to nth parameters that affect the maximum temperature among the training data and a plurality of first to nth parameter coefficients corresponding to the plurality of first to nth parameters respectively. The training data during the learning operation includes the cell design parameters of the candidate battery cells, the simulation analysis conditions, and the maximum temperature. During the learning operation, the temperature prediction learning device 203 can determine the parameters among the cell design parameters and the simulation analysis conditions that affect the maximum temperature, that is, the parameters with a significant correlation level with the maximum temperature, as the plurality of first to nth parameters. The temperature prediction learning device 203 can determine a plurality of first to nth parameter coefficients indicating the degree of correlation by learning the degree of correlation between the plurality of first to nth parameters and the maximum temperature in the learning operation. The temperature prediction learning device 203 can output the plurality of first to nth parameters and the plurality of first to nth parameter coefficients as output data through the learning operation using the input data.

[0049] In addition, the temperature prediction learning device 203 can output a temperature calculation formula by using a combination of the plurality of first to nth parameters and the plurality of first to nth parameter coefficients. For example, the temperature calculation formula can be implemented as a regression equation of terms composed of the products of the plurality of first to nth parameters and the plurality of first to nth parameter coefficients corresponding to the plurality of first to nth parameters respectively.

[0050] In normal operation, the processor 201 can perform an operation to determine the maximum temperature by applying the input data to the temperature calculation formula. That is, in the learning operation, the temperature prediction learning device 203 can update the temperature calculation formula of the processor 201 through the learning operation, and in normal operation, the processor 201 can determine the maximum temperature by extracting information on the plurality of first to nth parameters constituting the temperature calculation formula from the input data and substituting the information into the temperature calculation formula. When the operation mode is normal operation, the preprocessing device 10 can only perform preprocessing on the plurality of first to nth parameters.

[0051] For example, the temperature calculation formula can include the stacking number, the cooling water temperature, the initial SOC, the fast charging time, the maximum current, and the maximum current retention time as the plurality of first to sixth parameters. In this regard, the plurality of first to sixth parameter coefficients can be positive real numbers or negative real numbers.

[0052] The operation method of the device 1 for predicting the maximum temperature according to the embodiment - that is, the method for predicting the maximum temperature - can include a preprocessing method and a temperature prediction learning method.

[0053] Figure 2 is a flowchart showing a preprocessing method for temperature prediction learning according to an embodiment.

[0054] The processor 11 can determine the sizes of various types of cells that can configure the battery module and the number of cell stacks corresponding to the sizes of the various types of cells based on the single-cell design data (S1).

[0055] The processor 11 can derive candidate battery modules by considering the production possibility of each of the multiple battery modules achieved using the sizes of the multiple types of cells and the number of cell stacks derived in step S1 (S2).

[0056] Instead of steps S1 and S2, the processor 11 can receive information about candidate battery modules from the outside.

[0057] The processor 11 can send information about candidate battery cells to the emulator 2 and receive HPPC test data for the candidate battery cells from the emulator 2 (S3).

[0058] The processor 11 can implement an equivalent circuit representing the electrical characteristics of the corresponding cell based on the HPPC test data for the candidate battery cells obtained in step S3, and can derive the RC value of the candidate battery cell based on the equivalent circuit (S4).

[0059] The processor 11 can determine the maximum temperature by using the single-cell design parameters, simulation analysis conditions, and RC value of the candidate battery cell (S5).

[0060] Figure 3 is a flowchart showing a temperature prediction learning method according to an embodiment.

[0061] The processor 201 can determine an operation mode as one of a normal operation and a learning operation according to an external command or a control condition (S10).

[0062] When the operation mode determined in step S10 is the learning operation, the temperature prediction learning device 203 can read training data from the memory 202 (S11).

[0063] In the learning operation, the temperature prediction learning device 203 can perform machine learning by using the read training data to output a temperature calculation formula for determining the maximum temperature of the cell during charging (S12).

[0064] The temperature prediction learning device 203 can determine the temperature calculation formula through the learning operation and update the temperature calculation formula to the processor 201 (S13).

[0065] When the operation mode determined in step S10 is the normal operation, the processor 201 can determine the maximum temperature by applying the input data to the temperature calculation formula (S14).

[0066] The plurality of first through nth parameters determined by the learning operation of the temperature determination device 20 includes a parameter for the maximum current maximum application time. The maximum current maximum application time refers to the maximum time among the times during which the maximum current is applied to the candidate battery cell during charging. The current applied to the candidate battery cell during charging may vary, and there may be a plurality of sections in which the fluctuating current is the maximum current. Among the plurality of sections, the longest time is the maximum current maximum application time.

[0067] Table 1 below shows the error between the predicted maximum temperature and the measured temperature during normal operation after the temperature determination device 20 performs a learning operation using the plurality of first through nth parameters that do not include the maximum current maximum application time and the error rate.

[0068] (Table 1)

[0069] Number of stacks Measured maximum temperature Predicted maximum temperature Error Error rate St_n1 69.52 65.84 3.68 5.29% St_n2 64.52 62.03 2.49 3.86% St_n3 60.77 59.53 1.24 2.04% St_n4 56.50 55.29 1.21 2.14% St_n5 52.20 52.95 0.76 1.45%

[0070] Table 2 below shows the error between the predicted maximum temperature and the measured temperature during normal operation after the temperature determination device 20 performs a learning operation using the plurality of first through nth parameters that include the maximum current maximum application time and the error rate.

[0071] (Table 2)

[0072] Number of stacks Measured maximum temperature Predicted maximum temperature Error Error rate St_n1 69.52 69.06 0.46 0.66% St_n2 64.52 64.18 0.34 0.52% St_n3 60.77 59.65 1.12 1.85% St_n4 56.50 55.60 0.90 1.59% St_n5 52.20 52.70 0.50 0.96%

[0073] A relationship of St_n1 < St_n2 < St_n3 < St_n4 < St_n5 is established between the values indicating the stack numbers in Table 1 and Table 2. In Table 1, the highest error rate is 5.29%, and in Table 2, the highest error rate is reduced to 1.85%. That is, it can be seen that when learning using the plurality of first through nth parameters including the maximum current maximum application time, the prediction accuracy of the maximum temperature increases.

[0074] Hereinafter, a method for determining training data for temperature prediction learning according to an embodiment will be described.

[0075] Figure 4 is a line graph showing the maximum temperature predicted according to the number of stacks in the preprocessing device.

[0076] In Figure 4 a line graph of the predicted maximum temperature in each of the stack numbers N1, N2, N3, N4, N5, N6, and N7 is shown. The variables N1 to N7 indicating the stack numbers may be N1 < N2 < N3 < N4 < N5 < N6 < N7, and the difference between two adjacent stacks among the variables N1 to N7 may be the same.

[0077] As Figure 4As shown, when the line 41 indicated by the dashed line is a linear fitting line, the predicted maximum temperatures for the stack numbers N2 and N4 are different for the linear fitting line. In the case where the data corresponding to the stack numbers N2 and N4 is excluded from the training data, when the temperature prediction learning device 203 performs machine learning by using the corresponding training data to derive the temperature calculation formula, the prediction differences shown in Table 3 below occur. Table 3 is a table showing the measured maximum temperature and the predicted maximum temperature under the condition of 15 minutes for fast charging. The predicted maximum temperature can be derived from the temperature calculation formula learned and determined by the temperature prediction learning device 203. When the temperature prediction learning device 203 performs machine learning by using all the training data including the stack numbers N2 and N4 to derive the temperature calculation formula, the prediction differences shown in Table 4 below occur.

[0078] As shown in Tables 3 and 4, when all the training data is used for machine learning, the accuracy of the prediction results of the maximum temperatures of the stack numbers N2 and N4 is improved.

[0079] (Table 3)

[0080] Number of stacks Measured temperature Predicted temperature Difference Percentage N2 72.41 73.05 0.65 0.88% N4 69.31 68.63 0.67 0.98%

[0081] (Table 4)

[0082] Number of stacks Measured temperature Predicted temperature Difference Percentage N2 72.41 72.93 0.53 0.72% N4 69.31 68.66 0.65 0.94%

[0083] Table 5 below shows the errors and error rates between the measured maximum temperature and the predicted maximum temperature for the stack numbers N2 and N4 under the conditions of 15 minutes, 18 minutes, 20 minutes, 25 minutes, and 30 minutes of fast charging time when the temperature prediction learning device 203 performs machine learning by using the training data and derives the temperature calculation formula under the condition of excluding the data corresponding to the stack numbers N2 and N4.

[0084] (Table 5)

[0085] Number of stacks / quick charging time Measured maximum temperature Predicted maximum temperature Error Error rate [%] N2 / 15 minutes 72.41 71.93 0.47 0.65 N2 / 18 minutes 67.06 67.24 0.18 0.26 N2 / 20 minutes 63.08 63.50 0.42 0.67 N2 / 25 minutes 58.55 58.75 0.20 0.34 N2 / 30 minutes 54.07 54.46 0.39 0.73 N4 / 15 minutes 69.31 70.23 0.92 1.33 N4 / 18 minutes 64.28 64.06 0.23 0.35 N4 / 20 minutes 60.57 60.51 0.06 0.10 N4 / 25 minutes 56.42 57.06 0.64 1.14 N4 / 30 minutes 52.29 52.13 0.15 0.29

[0086] Table 6 below shows the errors and error rates between the measured maximum temperature and the predicted maximum temperature for the stack numbers N2 and N4 under the conditions of 15 minutes, 18 minutes, 20 minutes, 25 minutes, and 30 minutes of fast charging time when the temperature prediction learning device 203 performs machine learning by using all the training data and derives the temperature calculation formula.

[0087] (Table 6)

[0088] Number of stacks / quick charging time Measured maximum temperature Predicted maximum temperature Error Error rate [%] N2 / 15 minutes 72.41 73.95 1.55 2.13 N2 / 18 minutes 67.06 67.99 0.93 1.38 N2 / 20 minutes 63.08 63.99 0.92 1.45 N2 / 25 minutes 58.55 59.09 0.55 0.93 N2 / 30 minutes 54.07 54.56 0.49 0.91 N4 / 15 minutes 69.31 68.46 0.85 1.22 N4 / 18 minutes 64.28 65.26 0.97 1.52 N4 / 20 minutes 60.57 61.02 0.44 0.73 N4 / 25 minutes 56.42 56.82 0.41 0.72 N4 / 30 minutes 52.29 52.56 0.28 0.53

[0089] The highest error rate in Table 5 is 1.33%, which is less than the highest error rate of 2.13% in Table 6. The results in Table 5 can be regarded as an overfitted temperature calculation formula derived through learning.

[0090] As described above, training data including all stacking numbers that can be applied to the candidate battery modules can improve the prediction accuracy of the highest temperature.

[0091] Although the embodiments of the present disclosure have been described in detail above, the scope of the present disclosure is not limited thereto, and various modifications and improvements made by those of ordinary skill in the art to which the present disclosure pertains also fall within the scope of the present disclosure.

Claims

1. An apparatus for predicting the maximum temperature, the apparatus comprising: A preprocessing device, the preprocessing device including a first processor and a first memory storing a plurality of programs; And A temperature determination device, the temperature determination device including a second memory and a temperature prediction learning device, Wherein, the first processor is configured to generate input data including monomer design parameters and simulation analysis conditions, the monomer design parameters being used to constitute candidate battery monomers of a candidate battery module, and the simulation analysis conditions being used for simulating the prediction of the maximum temperature of the candidate battery monomers; The second memory stores the input data; And During a learning operation, the temperature prediction learning device is configured to perform machine learning by using training data among the input data stored in the second memory, and determine a temperature calculation formula for predicting the maximum temperature of the candidate battery monomers through machine learning.

2. The apparatus according to claim 1, wherein: The temperature determination device further includes A second processor, the second processor being configured to control the second memory to provide the training data to the temperature prediction training device during the learning operation, and control the temperature prediction training device to perform machine learning by using the training data.

3. The apparatus according to claim 2, wherein: The second processor is configured to After the learning operation, perform a normal operation of determining the maximum temperature of the candidate battery monomers according to the updated temperature calculation formula.

4. The apparatus according to claim 2, wherein: The temperature prediction learning device is configured to perform the learning operation to determine a plurality of first to nth parameters that affect the maximum temperature among the training data and a plurality of first to nth parameter coefficients corresponding to the plurality of first to nth parameters respectively, and The training data includes the monomer design parameters, the simulation analysis conditions, and the maximum temperature of the candidate battery monomers.

5. The apparatus according to claim 4, wherein: The simulation analysis conditions include The maximum retention time for the maximum current of the candidate battery monomers.

6. The apparatus according to claim 4, wherein: The temperature calculation formula is a regression formula of terms formed by the product of the plurality of first to nth parameters and the plurality of first to nth parameter coefficients.

7. The apparatus according to claim 1, wherein: The first memory stores A resistance * capacitance (RC) extraction program, the RC extraction program deriving the RC value of the candidate battery monomers, and The first processor is configured to Derive the RC value of the equivalent circuit corresponding to the candidate battery monomers by applying hybrid pulse power characterization (HPPC) test data for the candidate battery monomers to the RC extraction program.

8. The apparatus according to claim 1, wherein: The first memory stores A maximum temperature analysis program, the maximum temperature analysis program determining the maximum temperature of the candidate battery monomers, and The first processor is configured to The maximum temperature is determined by applying the monomer design parameters, the RC values of the equivalent circuits corresponding to the candidate battery monomers, and the simulation analysis conditions to the maximum temperature analysis program.

9. The apparatus according to claim 1, wherein: The first memory stores a candidate derivation program that determines the monomer design parameters based on monomer design data of required specifications, and derives the candidate battery module according to the monomer design parameters, and the first processor is configured to derive the candidate battery module by applying the monomer design data to the candidate derivation program.

10. A method for predicting a maximum temperature, the method comprising: deriving resistance-capacitance (RC) values of equivalent circuits corresponding to candidate battery monomers constituting a candidate battery module; determining the maximum temperature of the candidate battery monomers by using the monomer design parameters for the candidate battery monomers, the RC values, and simulation analysis conditions for a simulation for predicting the maximum temperature; and performing machine learning by using training data including the monomer design parameters, the RC values, and the simulation analysis conditions to determine a temperature calculation formula for predicting the maximum temperature of the candidate battery monomers.

11. The method according to claim 10, further comprising: after performing machine learning, determining the maximum temperature of the candidate battery monomers according to the updated temperature calculation formula.

12. The method according to claim 10, wherein: the determination of the temperature calculation formula includes determining a plurality of first to nth parameters that affect the maximum temperature among the training data and a plurality of first to nth parameter coefficients respectively corresponding to the plurality of first to nth parameters.

13. The method according to claim 12, wherein: the temperature calculation formula is a regression formula of terms constituted by the products of the plurality of first to nth parameters and the plurality of first to nth parameter coefficients.

14. The method according to claim 10, wherein: the training data includes a maximum retention time of a maximum current for the candidate battery monomers.

15. The method according to claim 10, wherein: the derivation of the RC values includes deriving the RC values by applying hybrid pulse power characterization (HPPC) test data for the candidate battery monomers to an RC extraction program.

16. The method according to claim 10, further comprising: determining the monomer design parameters based on monomer design data of required specifications.

Images