Battery state prediction device and battery state prediction method

Abstract

The battery state prediction device disclosed herein includes a communication interface for receiving battery data from a battery, and at least one processor that generates a battery state prediction model including a first prediction model for predicting lithium loss, a second prediction model for predicting active material loss in the positive electrode, and a third prediction model for predicting active material loss in the negative electrode, and inputs the battery data into the battery state prediction model to acquire battery state data.

Description

[Technical Field]

[0001] This application claims priority rights under Korean Patent Application No. 10-2023-0097583 dated July 26, 2023, and Korean Patent Application No. 10-2024-0089939 dated July 08, 2024, and all content disclosed in the documents of said patent applications is incorporated herein by reference. The embodiments disclosed herein relate to a battery state prediction device and a battery state prediction method for predicting the state of a battery. [Background technology]

[0002] In recent years, research and development on rechargeable batteries has been actively pursued. Here, rechargeable batteries refer to batteries that can be charged and discharged, and include both conventional Ni / Cd batteries, Ni / MH batteries, and the more recent lithium-ion batteries. Among rechargeable batteries, lithium-ion batteries have the advantage of having a much higher energy density than conventional Ni / Cd batteries and Ni / MH batteries. In addition, lithium-ion batteries can be manufactured to be small and lightweight, so they are used as power sources for mobile devices. Furthermore, the range of applications for lithium-ion batteries has expanded to include power sources for electric vehicles, and they are attracting attention as a next-generation energy storage medium.

[0003] The ability to accurately predict the State of Health (SOH) of a battery is extremely important from the perspectives of battery safety, performance, lifespan, economy, and environmental protection, and various SOH prediction technologies exist.

[0004] However, conventional SOH prediction techniques only considered the LLI (Loss of Lithium Inventory) of the negative electrode, and therefore had the limitation of not being able to take into account additional factors that occur during cycle experiments and actual battery use, rather than storage degradation. [Overview of the project] [Problems that the invention aims to solve]

[0005] According to one embodiment disclosed in this document, a battery state prediction model is provided that includes a prediction model that utilizes a P2D (Pseudo two-dimensional) model to explain the degradation that occurs in cycle experiments in the process of predicting the state of health (SOH) of a battery, and that takes into account the LAM (Loss of Active Material) in the positive and negative electrodes.

[0006] The technical problems of the embodiments disclosed in this document are not limited to those mentioned above, and other technical problems not mentioned can be clearly understood by those skilled in the art from the following description. [Means for solving the problem]

[0007] A battery state prediction device according to one embodiment includes a communication interface for receiving battery data from a battery, and at least one processor that generates a battery state prediction model including a first prediction model for predicting lithium loss, a second prediction model for predicting active material loss in the positive electrode, and a third prediction model for predicting active material loss in the negative electrode, and inputs the battery data into the battery state prediction model to acquire battery state data.

[0008] The at least one processor can generate the first predictive model based on an acceleration factor that quantifies the stress experienced at the electrode-solid electrolyte interface (SEI).

[0009] The at least one processor can determine the change in the chemical ratio of the active material due to lithium loss based on the first prediction model, and determine the open circuit potential (OCP) of the positive electrode and the negative electrode based on the change in the chemical ratio of the active material. The at least one processor can predict the state of the battery based on the amount of change in the open-circuit potential.

[0010] The at least one processor can generate the second prediction model based on cation mixing and particle cracking induced volume change. The at least one processor can input the battery data into the second prediction model to determine the degree of degradation of the battery.

[0011] The at least one processor can determine the correlation between the lithium loss and the active material loss in the negative electrode, and generate the third prediction model based on the correlation.

[0012] The at least one processor can input the change in battery capacity derived by the first prediction model into the second prediction model or the third prediction model to derive a degradation coefficient, and predict the state of the battery based on the degradation coefficient.

[0013] A method for predicting the state of a battery according to an embodiment includes receiving battery data from the battery, generating a battery state prediction model including a first prediction model for predicting lithium loss, a second prediction model for predicting active material loss in the positive electrode, and a third prediction model for predicting active material loss in the negative electrode, and inputting the battery data into the battery state prediction model to obtain battery state data.

[0014] Generating the first prediction model can include generating the first prediction model based on an acceleration factor that quantifies the stress received by the electrode and the solid electrolyte interface (SEI).

[0015] The battery state prediction method according to one embodiment can further include determining a change in the chemical ratio of the active material due to the lithium loss based on the first prediction model, and determining the open circuit potential (OCP) of the positive electrode and the negative electrode based on the change in the chemical ratio of the active material. The battery state prediction method according to one embodiment can further include predicting the state of the battery based on the amount of change in the open circuit potential.

[0016] Generating the second prediction model can include generating the second prediction model based on cation mixing and particle cracking induced volume change.

[0017] The battery state prediction method according to one embodiment can further include inputting the battery data into the second prediction model to determine the degree of degradation of the battery. Generating the third prediction model can include determining the correlation between the lithium loss and the loss of the active material in the negative electrode, and generating the third prediction model based on the correlation.

[0018] The battery state prediction method according to one embodiment can further include inputting the change in the battery capacity derived by the first prediction model into the second prediction model or the third prediction model to derive a degradation coefficient, and predicting the state of the battery based on the degradation coefficient.

Advantages of the Invention

[0019] According to the battery state prediction device according to one embodiment, through simulation based on the operating conditions of the battery, the degradation tendencies of the positive electrode and the negative electrode occurring under each operating condition can be predicted.

[0020] According to one embodiment of the battery state prediction device, the accuracy of SOH diagnosis and RuL (Remaining Useful Life) prediction of the battery can be improved. According to one embodiment of the battery state prediction device, the battery voltage can be diagnosed based on the battery's MoL (Middle of Life) state. [Brief explanation of the drawing]

[0021] [Figure 1] A block diagram showing the configuration of a battery state prediction device according to one embodiment is shown. [Figure 2] This diagram schematically shows the process by which a battery state prediction device according to one embodiment predicts the state of a battery based on a prediction model. [Figure 3] The data analysis results based on the first prediction model of a battery state prediction device according to one embodiment are shown. [Figure 4] The results for a model that considers the acceleration coefficient and a model that does not consider it in a battery state prediction device according to one embodiment are shown. [Figure 5] This graph shows the correlation based on the second prediction model of a battery state prediction device according to one embodiment. [Figure 6] The data analysis results based on the second prediction model of the battery state prediction device according to one embodiment are shown. [Figure 7] The results for a model that considers volume change and a model that does not consider volume change in a battery state prediction device according to one embodiment are shown. [Figure 8] This graph shows the correlation based on the third prediction model of a battery state prediction device according to one embodiment. [Figure 9] The parameter changes based on the second and third prediction models of the battery state prediction device according to one embodiment are shown. [Figure 10] The following shows the prediction results for storage degradation using a battery state prediction device according to one embodiment. [Figure 11] The following shows the prediction results for cycle degradation using a battery state prediction device according to one embodiment. [Figure 12]A control flowchart for a battery state prediction method according to one embodiment is shown. [Modes for carrying out the invention]

[0022] The various embodiments disclosed in this document will be described in detail below with reference to the attached drawings. In this document, the same reference numerals are used for identical components in the drawings, and redundant descriptions of the same components are omitted.

[0023] With respect to the various embodiments disclosed herein, any specific structural or functional descriptions are provided merely as examples for the purpose of illustrating the embodiments, and the various embodiments disclosed herein may be implemented in various forms and should not be construed as being limited to the embodiments described herein.

[0024] Expressions such as “first,” “second,” “first,” or “second,” used in various embodiments, may modify various components regardless of order and / or importance, and do not limit such components. For example, without departing from the scope of rights of the embodiments disclosed herein, the first component may be named the second component, and similarly, the second component may be renamed the first component.

[0025] The terminology used in this document is used solely to describe specific embodiments and is not intended to limit the scope of other embodiments. Singular expressions may include plural expressions unless the context clearly indicates otherwise.

[0026] All terms used herein, including technical or scientific terms, may have the same meaning as those generally understood by a person of ordinary skill in the art of the embodiments disclosed herein. Terms defined in commonly used dictionaries may be interpreted as having the same or similar meaning as in the context of the relevant art, and not as ideal or overly formal unless explicitly defined herein. In some cases, terms defined herein should not be interpreted in a way that excludes the embodiments disclosed herein.

[0027] Figure 1 shows a block diagram illustrating the configuration of a battery state prediction device according to one embodiment. Referring to Figure 1, a battery state prediction device 1 according to one embodiment includes a control unit 100 which includes at least one processor 110 and memory 120, and a communication interface 200 which can communicate with an external device 5 via the communication interface 200 to predict the state of the battery.

[0028] According to one embodiment, the battery state prediction device 1 can monitor voltage, current, and temperature from a voltage sensor 2, a current sensor 3, and a temperature sensor 4, and control and manage them to prevent overcharging and over-discharging, and can include, for example, a BMS (Battery Management System).

[0029] According to one embodiment, the battery state prediction device 1 can consist of a user terminal and / or a server device capable of communicating with an external device 5. Specifically, if the battery state prediction device 1 is a user terminal, the control unit 100 of the battery state prediction device 1 can be configured with the CPU of the user terminal so that battery data can be merged on-device at the user terminal. In this case, the user terminal may include, but is not limited to, a personal computer, terminal, portable telephone, smartphone, handheld device, wearable device, etc.

[0030] Furthermore, if the battery state prediction device 1 is a server device, the server device can be implemented using various computing devices such as a workstation, cloud, data drive, or data station. The server device can be implemented using one or more server devices that are physically or logically separated based on functions, detailed functional configurations, or data, and can send and receive data and process the transmitted and received data through communication between each server device.

[0031] In one embodiment, the battery state prediction device 1 may refer to all electronic devices including the processor 110 and memory 120, and can be mounted and operated in a vehicle. The components of the battery state prediction device 1 will be described in detail below.

[0032] The communication interface 200 may include a wireless communication interface 210 and a wired communication interface 220 for communicating with the external device 5. The communication interface 200 can send and receive programs and various data for calculating battery cell characteristic values, classifying, and estimating lifespan from a separately provided external server.

[0033] The wireless communication interface 210 may include at least one of a short-range communication module and a long-range communication module. The near-field communication module can communicate with an external device 5 adjacent to the battery state prediction device 1 using a near-field communication method. Here, the near-field communication module can utilize one of the following communication methods: Bluetooth®, BLE (Bluetooth Low Energy), Infrared Data Association (IrDA), Zigbee®, Wi-Fi®, Wi-Fi Direct®, Ultra Wideband (UWB), or Near Field Communication (NFC).

[0034] The long-range communication module includes communication modules that perform various types of long-range communication and may include a mobile communication interface. The mobile communication interface can send and receive radio signals with at least one of the following on a mobile communication network: a base station, an external terminal, or an external device 5. The long-range communication module can also communicate with the external device 5 or other electronic devices 5 via a nearby access point (AP). The access point (AP) can connect the local network (LAN) to which the battery state prediction device 1 is connected to a wide area network (WAN) to which the communication server is connected. As a result, the battery state prediction device 1 can connect to the external device 5 and the communication server via the wide area network (WAN) and communicate with each other.

[0035] The wired communication interface 220 can connect to a wired communication network and communicate with an external device 5 via the wired communication network. For example, the wired communication interface 220 can connect to a wired communication network via Ethernet (registered trademark, IEEE 802.3 technical standard) or via CAN communication, and can send and receive data with the external device 5 via the wired communication network.

[0036] A battery state prediction device 1 according to one embodiment may include an input / output interface (not shown). This interface can provide a connection between an input device (not shown), such as a keyboard, mouse, or touch panel, an output device (not shown), such as a display, and a processor 110, enabling data transmission and reception.

[0037] The memory 120 can store various information necessary for operating the battery state prediction device 1. Specifically, the memory 120 can store the operating system and programs necessary for operating the battery state prediction device 1, or it can store the data necessary for operating the battery state prediction device 1.

[0038] Specifically, the memory 120 can store various programs related to calculating the characteristic values ​​of battery cells, classifying them, and estimating their lifespan. The memory 120 can also store various data such as voltage, current, and characteristic value data for each battery cell.

[0039] Furthermore, the memory 120 can store data related to the first prediction model, the second prediction model, and the third prediction model for predicting the battery state. Memory 120 may include volatile memory 120 such as S-RAM (Static Random Access Memory) and D-RAM (Dynamic Random Access Memory) for temporarily storing data. Memory 120 may also include non-volatile memory 120 such as ROM (Read Only Memory), EPROM (Erasable Programmable Read Only Memory), and EEPROM (Electrically Erasable Programmable Read Only Memory) for long-term data storage.

[0040] The processor 110 outputs control signals and controls the battery state prediction device 1 overall. The processor 110 may include one or more CPUs (central processing units) and GPUs (Graphics Processing Units). In this case, the processor 110 may be implemented as an array of multiple logic gates, or as a combination of a general-purpose microprocessor 110 and a memory 120 that stores a program executable by the microprocessor 110.

[0041] The aforementioned memory 120 and processor 110 can be included in the control unit 100, and the control unit 100 can control the aforementioned components to predict conditions related to battery degradation.

[0042] Specifically, the processor 110 can generate a battery state prediction model including a first prediction model, a second prediction model, and a third prediction model, and can input battery data into the battery state prediction model to obtain battery state data.

[0043] Here, the first prediction model includes a model for predicting lithium loss, and the first prediction model can be defined by an acceleration coefficient that quantifies the stress experienced at the electrode-solid electrolyte interface (SEI).

[0044] Furthermore, the processor 110 can determine the change in the chemical ratio of the active material due to lithium loss based on the first prediction model, and determine the open circuit potential (OCP) of the positive and negative electrodes based on the change in the chemical ratio of the active material.

[0045] In other words, the first predictive model can explain the changes in the open-circuit potential of the positive and negative electrodes, and therefore can accurately analyze the physical and chemical changes of the positive and negative electrodes that occur during the battery degradation process.

[0046] Furthermore, the second prediction model includes a model for predicting the loss of active material at the cathode, and this second prediction model can be defined by cation mixing and volume change due to particle cracking.

[0047] Furthermore, the third prediction model includes a model for predicting the loss of active material at the negative electrode, and the third prediction model can be defined by the correlation between lithium loss and the loss of active material at the negative electrode.

[0048] The processor 110 can input the change in battery capacity derived by the first prediction model into a second or third prediction model to derive a degradation factor. That is, the processor 110 can first calculate the change in capacity predicted by the first prediction model and then apply it to the degradation models of the positive and negative electrodes to determine the degradation factors of the positive and negative electrodes.

[0049] Thus, the battery state prediction device 1 according to one embodiment has the effect of improving the accuracy of predicting the degree of degradation based on a first prediction model that predicts lithium loss and second and third prediction models that predict active material loss in the positive and negative electrodes.

[0050] Figure 2 schematically shows the flow of how a battery state prediction device according to one embodiment predicts the state of a battery based on a prediction model. The configurations 101 to 103 in Figure 2 are implemented in the form of software blocks, stored in memory 120, and can be executed by processor 110.

[0051] Referring to Figure 2, the processor 110 can receive battery data via a sensor unit including a voltage sensor 2, a current sensor 3, and a temperature sensor 4. Subsequently, the prediction model generation unit 101 of the control unit 100 can generate multiple prediction models and determine the degree of battery degradation.

[0052] Specifically, the first prediction model generation unit 101-1 of the control unit 100 can generate a prediction model that can explain the LLI (Loss of Lithium Inventory) that occurs in cycle experiments by utilizing a P2D (Pseudo two-dimensional) model.

[0053] Here, the P2D model is a model that mathematically models the operation of lithium-ion batteries, and can be described as a model that predicts battery performance by combining particle-level and domain-level phenomena. Specifically, the P2D model can predict the overall performance of a battery by integrating particle-level diffusion and reaction with domain-level macroscopic transfer phenomena, and can model realistic battery operation including electrode porosity, electrolyte heterogeneity, and polarization effects.

[0054] Furthermore, the LLI phenomenon refers to the phenomenon in which the amount of usable lithium ions in a battery decreases, and can occur due to the formation of the SEI (Solid Electrolyte Interphase) layer, lithium metal plating, electrolyte decomposition, and side reactions.

[0055] When the LLI phenomenon occurs in a battery, problems such as capacity reduction, performance degradation, and shortened lifespan may occur. Therefore, a battery performance prediction device according to one embodiment can predict battery degradation by modeling the LLI phenomenon that occurs in a battery.

[0056] Specifically, the LLI phenomenon that occurs in a cycle is mainly caused by side reactions occurring at the negative and positive electrodes, and such LLI phenomena can be represented by an electrochemical equation in the form of a Butler-Volmer reaction.

[0057] Furthermore, the stress on the negative electrode and SEI layer can cause mechanical defects in the electrode and SEI layer. These defects can expose the surfaces of the negative electrode and SEI layer to the electrolyte, allowing additional SEI to be deposited in the newly exposed areas. This results in more LLI (Long Life Isolation) occurring compared to storage degradation, and the storage degradation model alone can no longer explain cycle degradation.

[0058] According to one embodiment of the battery state prediction device 1, cycle degradation can be simulated by introducing an acceleration coefficient, based on the fact that the aforementioned stress occurs significantly under low temperature and high C-rate conditions.

[0059] In this case, since it is experimentally difficult to directly calculate the stress on the electrodes and SEI layer, the first predictive model can express the concentration difference from the surface to the interior of the electrode as an empirical equation, as shown in the following equation, instead of stress.

[0060] Furthermore, the first prediction model incorporates mathematical models to explain lithium metal plating that occurs at high C-rates and low temperatures, allowing for the simulation of abnormal degradation such as rapid performance degradation due to rapidly increasing LLI.

[0061] Specifically, equations 1 to 7 below are included in the first prediction model and may represent an LLI model due to cycle degradation at the negative electrode. Here, Equation 1 is the total capacity reduction rate during the battery degradation process (

number

Number

Number

Number

[0062] Specifically, in Equation 2, [[ID=​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​RE This refers to the function that determines the growth current of the SEI layer.

number

[0066] Also, in equation 6, i TM This refers to the current related to the transition metal catalytic effect, f TM This represents the function that determines the transition metal catalytic effect current, U p (t) represents the electrode voltage over time, and T(t) can represent the temperature over time.

[0067] Also, in equation 7, i LP This refers to the current associated with lithium plating, and f RE This represents the function that determines the lithium plating current, U n (t) represents the electrode potential over time, and T(t) can represent the temperature over time.

[0068] [Formula 1]

number

number

number

number

[0069] [Formula 5]

number

number

number

[0070] Furthermore, equations 8-10 below are included in the first prediction model and may represent an LLI model due to cycle degradation at the positive electrode. Specifically, in equation 8,

number

number

[0071] Also, in equation 9,

number

[0072] Also, in equation 10, i CEI This refers to the current associated with CEI formation, and f CEI This represents the function that determines the CEI formation current, U P (t) represents the electrode voltage over time, and T(t) can represent the temperature over time.

[0073] [Formula 8]

number

number

number

[0074] The second prediction model generation unit 101-2 of the control unit 100 can model degradation in the cathode based on the correlation between LAMp (Loss of Active Material of cathode) and SOC (State of Charge) in battery storage and cycle degradation.

[0075] Specifically, LAMp refers to the loss of active material in the positive electrode, which is one of the main causes of battery performance degradation. It can refer to the loss or deactivation of active material in the positive electrode during the charging and discharging processes of the battery.

[0076] The second prediction model may include a Pseudo Cation Mixing model that simulates the phenomenon of cations within an electrode changing positions with other cations in the electrode material, and a Particle Cracking Induced Volume Change model that simulates the cracking phenomenon that occurs as electrode particles repeatedly expand and contract during the charging and discharging process of a battery.

[0077] The virtual cation mixing model included in the second prediction model can explain the degradation of the cathode that occurs frequently in the intermediate SOC region (50-60%) by establishing a degradation model based on the assumption of virtual cation mixing.

[0078] Specifically, in the low SOC region, nickel ions (Ni2+) in the NCM crystal structure readily migrate to lithium ions (Li+), but because there are few lithium ion vacancies, cation mixing may be reduced.

[0079] In contrast, in the high SOC region, nickel ions (Ni2+) are oxidized to nickel ions (Ni4+), making it difficult for them to move to lithium ions (Li+), resulting in fewer antisite defects even with a large number of lithium ion vacancies.

[0080] Based on the above characteristics, the second prediction model of the battery state prediction device 1 according to one embodiment can be composed of cation mixing governing equations as shown in the following equations 11 and 12.

[0081] Specifically, in equation 11,

number

[0082] Also, in equation 12, i CM This refers to the current associated with cation mixing, and f CM This refers to a function that determines the current related to cation mixing, and NI Li (t) represents the lithium concentration over time, and T(t) can represent the temperature over time.

[0083] [Formula 11]

number

number

[0084] Furthermore, the volume change model due to particle cracking included in the second prediction model can explain the additional cathode degradation that occurs during the cycle by establishing a degradation model based on the assumption of particle cracking due to volume change.

[0085] Specifically, a volume change model due to particle cracking can be generated by assuming that when current is applied to a battery, the volume of the positive electrode decreases during charging, causing delithiation relative to the positive electrode, resulting in cracks forming between the initial particles, and the electrolyte permeating into the resulting extra space, leading to degradation of the positive electrode.

[0086] Based on the above characteristics, the second prediction model of the battery state prediction device 1 according to one embodiment can be composed of particle crack governing equations as shown in the following equations 13 and 14.

[0087] Specifically, in equation 13,

number

[0088] Also, in equation 14, i CR This refers to the current associated with particle cracking, and f CR ΔV(t) represents the function that determines the current associated with particle cracking, ΔV(t) represents the change in volume over time, and T(t) can represent the temperature over time.

[0089] [Formula 13]

number

number

[0090] The third prediction model generation unit 101-3 of the control unit 100 can model degradation at the anode based on the correlation between LLI and LAMn (Loss of Active Material of anode) in battery storage and cycle degradation.

[0091] Specifically, LAMn refers to the loss of active material in the negative electrode, which is one of the main causes of battery performance degradation. It can refer to the loss or deactivation of active material in the negative electrode during the charging and discharging processes of the battery.

[0092] In one embodiment of the battery state prediction device 1, it is assumed that pore clogging by LLI blocks the surface sites of the negative electrode where lithium can enter, leading to LAMn, and the third prediction model can be configured as shown in the following equation 15.

[0093] Specifically, in equation 15,

number

number

number

[0094] [Formula 15]

number

[0095] Subsequently, the parameter adjustment unit 102 of the control unit 100 can simulate the parameter changes of the physical model using LLI and the parameter changes of the physical model using LAM.

[0096] Specifically, in the case of parameter changes in the physical model due to LLI, LLI is applied to the following equations 16-20, and the voltage usage region (V max , V min By defining ), it is possible to mathematically calculate the changes in the stoichiometry range of the positive and negative electrodes (negative electrode w0~w100, positive electrode z0~z100) according to LLI.

[0097] This allows us to reflect these results in a physical model and explain the changes in the Open Circuit Potential (OCP) of the positive and negative electrodes due to LLI.

[0098] Specifically, in equation 16,

number

number

number

number

number

number

[0099] Also, in equation 17,

number

number

number

number

number

number

[0100] Also, in equation 18,

number

number

number

[0101] Also, in equation 19,

number

number

number

[0102] Also, in equation 20,

number

[0103] [Formula 16]

number

number

number

[0104] [Formula 19]

number

number

[0105] Furthermore, in the case of parameter changes in the physical model due to LAM, the occurrence of LAM causes shrinkage of the OCP profiles at the positive and negative electrodes, which is the main cause of the change in the overall shape of the OCV. Therefore, it is possible to simulate parameter changes due to LAM by introducing a decay factor, and a detailed explanation of this will be given later in Figure 9.

[0106] Subsequently, the prediction result derivation unit 103 of the control unit 100 can predict the battery state, taking into account the degree of battery degradation, from at least one of the first prediction model, the second prediction model, and the third prediction model, and transmit it to the external device 5.

[0107] Figure 3 shows the data analysis results based on the first prediction model of a battery state prediction device according to one embodiment. Referring to Figure 3, it can be confirmed that LLI occurs significantly at low temperatures during cycle degradation, as shown by the hypothetical trend line (a). The battery state prediction device 1 according to one embodiment can predict the battery state by taking into account the LLI that occurs significantly at low temperatures. In Figure 3, the horizontal axis may represent the battery temperature, and the vertical axis may represent the amount of decrease in battery capacity due to LLI.

[0108] Specifically, LLI (Loss of Lithium Inventory), which occurs in large quantities at low temperatures, can be caused by factors such as increased SEI formation, deposition of lithium metal, and degradation of electrolyte performance.

[0109] Among the causes of LLI increase, SEI formation is related to the formation of a solid layer (SEI) on the electrode surface. At low temperatures, SEI formation on the electrode surface may increase. Since the SEI layer is formed when lithium ions react with the electrolyte to create a solid layer on the electrode surface, lithium ions are consumed in this process, which can reduce the amount of usable lithium ions.

[0110] SEI formation primarily occurs during the initial cycles of a battery, but at low temperatures, this process can proceed more actively. As a result, the SEI layer becomes thicker, trapping more lithium ions and potentially affecting battery degradation.

[0111] One of the causes of LLI increase is related to the deposition of lithium metal. At low temperatures, lithium ions are not uniformly inserted into the negative electrode, and there is a higher possibility that they will precipitate in the form of lithium metal. Since such lithium metal cannot be reused, it can increase LLI.

[0112] One of the causes of LLI increase is related to the degradation of electrolyte performance. At low temperatures, the ionic conductivity of the electrolyte decreases, which can slow down the movement rate of lithium ions and increase the internal resistance of the battery. As a result, the degradation of electrolyte performance can lead to non-uniform lithium ion exchange with the electrodes, which can increase LLI.

[0113] One embodiment of the battery state prediction device incorporates mathematical models to explain lithium metal plating that occurs at high C-rates and low temperatures, as shown in equations 1 to 10, and can simulate abnormal degradation such as a rapid performance decrease due to a rapidly increasing LLI.

[0114] Figure 4 shows the results for a model that considers the acceleration coefficient and a model that does not consider it in a battery state prediction device according to one embodiment. In Figure 4, the horizontal axis may represent the predicted value of LLI, and the vertical axis may represent the simulated value of LLI.

[0115] Referring to Figure 4, it can be confirmed that the predicted LLI value predicted by the processor 110 of the battery state prediction device 1 according to one embodiment shows a linear relationship with the simulated value.

[0116] Specifically, Figure 4(a) represents the first prediction model without considering the acceleration coefficient, and Figure 4(b) may represent the first prediction model when considering the acceleration coefficient, as shown in equations 1 to 10.

[0117] In Figure 4(a), the first learning model did not consider acceleration coefficients that quantify the stress on the negative electrode and SEI, so the graph can be shown relatively far from the trend line of the linear relationship at all temperatures.

[0118] In contrast, in Figure 4(b), the first learning model mathematically considers an acceleration coefficient that quantifies the stress on the negative electrode and SEI, so the graph can be represented relatively close to a linear trend line at all temperatures.

[0119] In other words, a linear relationship between the predicted and simulated values ​​of LLI means that the prediction model accurately reflects the changes in LLI, and that the first prediction model accurately reproduces the operation of the actual battery system.

[0120] Furthermore, when experiments were conducted under the same conditions, but with only the application of the acceleration factor changed, the root mean square error (RMSE) in Figure 4(a) was 1.97 Ah, and the root mean square error in Figure 4(b) was 1.38 Ah. This confirms that the prediction performance of Figure 4(b), where the acceleration factor is applied, is superior, even when measured by the root mean square error, which measures the difference between the model's predicted value and the simulated value.

[0121] As a result, the battery state prediction device 1 according to one embodiment, unlike conventional technology, can introduce an acceleration coefficient to quantify the stress on the electrodes and SEI, thereby improving its ability to simulate the degradation of an actual battery system.

[0122] Figure 5 shows a graph of the correlation based on the second prediction model of a battery state prediction device according to one embodiment. Figure 5(a) represents the correlation between LAMp and SOC related to storage degradation in the second prediction model, and Figure 5(b) may represent the correlation between LAMp and SOC related to cycle degradation in the second prediction model.

[0123] Specifically, in the low SOC region shown in Figure 5, nickel ions (Ni2+) in the NCM crystal structure readily migrate to lithium ions (Li+), but because there are fewer lithium ion vacancies, cation mixing may be reduced.

[0124] In contrast, in the high SOC region, nickel ions (Ni2+) are oxidized to nickel ions (Ni4+), making it difficult for them to move to lithium ions (Li+), resulting in fewer antisite defects even with a large number of lithium ion vacancies.

[0125] Thus, although it is difficult to see in the trend lines (a-1) and (b-1), there is a trade-off between the movement of lithium vacancies and nickel ions depending on the charge state, so in the intermediate SOC region, as in regions (a-2) and (b-2), a large amount of active material loss (LAMp) occurs in the positive electrode (volcanic activity). n o) A tendency like this may occur.

[0126] As a result, the battery state prediction device 1 according to one embodiment can simulate the positive electrode active material loss (LAMp) that occurs significantly in the SOC intermediate region, as shown in Figures 6 and 7.

[0127] Figure 6 shows the data analysis results based on the second prediction model of the battery state prediction device according to one embodiment. Referring to Figure 6, it can be seen that when the SOC lift is large, as shown by the hypothetical trend line (a), a large LAMp occurs. The battery state prediction device 1 according to one embodiment can predict the battery state by taking into account the LAMp that occurs when the SOC lift is large. In Figure 3, the horizontal axis may represent the magnitude of the SOC lift, and the vertical axis may represent the magnitude of the LAMp.

[0128] Specifically, when the state of charge (SOC) lift is large during cycle degradation, that is, when the SOC is maintained at a high state for a longer period of time or is frequently charged at a high SOC, active material loss in the cathode can occur due to a combination of factors such as electrolyte decomposition, corrosion of the cathode material, thermal instability, mechanical stress, and SEI layer formation on the electrode surface.

[0129] In response to this, a battery state prediction device 1 according to one embodiment can accurately predict LAMp based on a pseudo-cation mixing model and a particle cracking-induced volume change model, as shown in equations 11-14.

[0130] Figure 7 shows the results for a model that considers volume change and a model that does not consider volume change in a battery state prediction device according to one embodiment. Referring to Figure 7, it can be confirmed that the predicted value of LAMp predicted by the processor 110 of the battery state prediction device 1 according to one embodiment shows a linear relationship with the simulated value.

[0131] Specifically, Figure 7(a) represents the prediction result graph when the processor 110 applies only the virtual cation mixing model from the second prediction model, and Figure 7(b) may represent the prediction result graph when both the virtual cation mixing model and the volume change model due to particle damage are applied.

[0132] In Figure 7(a), the second learning model did not consider volume changes due to particle damage, so the graph can be represented relatively far from the trend line of the linear relationship at all temperatures.

[0133] In contrast, in Figure 7(b), the second learning model takes into account the volume changes due to the negative electrode and particle damage using equations 13 and 14, so the graph can be represented as being relatively close to the linear relationship trend line at all temperatures.

[0134] In other words, a linear relationship between the simulated and predicted values ​​of LAMp means that the prediction model accurately reflects the changes in LAMp, and that the second prediction model accurately reproduces the operation of the actual battery system.

[0135] Furthermore, when experiments were conducted under the same conditions, only changing whether or not volume changes due to particle damage were considered, the mean square error in Figure 7(a) was 1.63 Ah, and the mean square error in Figure 7(b) was 1.28 Ah. This confirms that the prediction performance of Figure 7(b), which incorporates volume changes due to particle damage, is superior, even when measured by the mean square error, which measures the difference between the model's predicted value and the simulated value.

[0136] As a result, the battery state prediction device 1 according to one embodiment can predict LAMp by considering both a virtual cation mixing model and a volume change model due to particle damage, as shown in the following equation 21, which has the effect of improving its ability to simulate the degradation of an actual battery system.

[0137] Specifically, in equation 21,

number

number

number

[0138] [Formula 21]

number

[0139] Figure 8 shows a graph of correlations based on a third prediction model of a battery state prediction device according to one embodiment. Referring to Figure 8, the processor 110 can analyze the correlation between LLI and LAMn, which is the active material loss at the negative electrode, and based on this, it can confirm a tendency for a linear relationship as shown in Figure 8.

[0140] Specifically, the correlation arises because when an SEI layer is formed on the surface of the negative electrode, lithium ions are consumed, and as the SEI layer thickens, more lithium ions are irreversibly consumed, potentially increasing LLI. In this case, as the SEI layer thickens, the active surface area of ​​the negative electrode decreases, which can lead to LAMn generation, thus suggesting a correlation between the two.

[0141] Furthermore, during repeated charge-discharge cycles, the negative electrode material expands and contracts, generating mechanical stress that can lead to decomposition and cracking of the negative electrode material, increasing LAMn. In addition, increased loss of negative electrode material reduces its ability to accommodate lithium ions, which can exacerbate LLI.

[0142] Furthermore, in high SOC conditions or low-temperature environments, lithium ions are less likely to be uniformly inserted into the negative electrode and more likely to precipitate in the form of lithium metal. This increases lithium ion loss (LLI), and the deposition of lithium metal can damage the surface of the negative electrode, potentially degrading the LAMn.

[0143] Based on these correlations, the processor 110 can predict the state of the battery using a model like the one described in equation 15, assuming that pore clogging by LLI blocks surface sites on the negative electrode where lithium can enter, leading to LAMn.

[0144] Figure 9 shows the parameter changes based on the second and third prediction models of the battery state prediction device according to one embodiment. Referring to Figure 9, it is possible to simulate the parameter changes of the physical model due to LAM in the second and third prediction models. Specifically, LAM causes shrinkage of the OCP profiles of the positive and negative electrodes, and LAM is the main cause of the change in the overall shape of the OCV. From this, the processor 110 can introduce a decay factor to simulate the parameter changes due to LAM.

[0145] In this case, Figure 9(a) can be assumed to be a rough representation of the OCV (vertical axis) against volume (horizontal axis) when no LAM occurs, Figure 9(b) is a rough representation of the graph that changes when 20% of LAMp occurs, and Figure 9(c) is a rough representation of the graph that changes when 30% of LAMn occurs.

[0146] A degradation coefficient can be introduced to simulate the changes in the general shape of such graphs, and to simulate the parameter changes due to LAM by introducing the degradation coefficient, the battery capacity can be calculated from the design information as shown in Equation 22 below.

[0147] Specifically, in equation 22, Q represents the battery capacity.

number

number

number

[0148] [Formula 22]

number

[0149] In this case, the shrinkage of the open-circuit voltage (OCV) when active substance loss (LAM) occurs can be interpreted as maintaining the usable range of stoichiometry (z), but changing stoichiometry over time.

[0150] Based on the volume relation equation above, LAM can be reflected by reducing the maximum concentration (C_max) value or the volume fraction (epss) value according to the resulting LAM.

[0151] For example, i) when the C_max value decreases, the relationship z = C_surf(t) / C_max indicates that the z value increases compared to the initial state (BoL) for the same surface concentration. Also, ii) when the volume fraction decreases, the electrochemical specific surface area decreases, so the current density flowing through the electrode particles increases even when the same current as in the initial state is applied. This results in a larger change in stoichiometry.

[0152] In the case of the positive electrode, assuming that the loss of positive electrode active material (LAMp) due to cation mixing is the main cause of degradation, and assuming that the change in crystal structure due to cation mixing changes the theoretical capacity of the electrode particles, the maximum concentration (C_max) value among the parameters of the physical model can be adjusted.

[0153] Furthermore, in the case of the negative electrode, the volume fraction value can be adjusted based on the assumption that the electrochemical specific surface area decreases compared to that of a single particle, assuming that the negative electrode active material loss (LAMn) arises from pore clogging due to lithium inventory loss (LLI).

[0154] With the above assumptions, and by first calculating and applying the capacity change shown in Equation 22, the degradation coefficients for the positive and negative electrodes can be defined as shown in Equations 23-26 below.

[0155] Specifically, in equation 23,

number

number

number

[0156] Also, in Equation 24, [Number] means the degradation coefficient of the positive electrode, [Number] means the value obtained by dividing the capacity in the middle of the life by the initial capacity, [Number] means the value obtained by subtracting the initial stoichiometric value from the final stoichiometric value of the positive electrode in the initial state, [Number] may mean the value obtained by subtracting the initial stoichiometric value from the final stoichiometric value of the positive electrode in the middle of the life.

[0157] Also, in Equation 25, [Number] means the volume ratio of the active material of the negative electrode in the current state, [Number] means the volume ratio of the active material of the negative electrode in the initial state, [Number] may mean the degradation coefficient of the negative electrode. <00009​​​ [Number] means the maximum concentration of the positive electrode in the current state, [Number] means the maximum concentration of the positive electrode in the initial state, [Number] may mean the degradation coefficient of the positive electrode.

[0159] [Formula 23] [Number] [Formula 24] [Number]

[0160] [Formula 25] [Number] [Formula 26] <所 [Number]

[0161] Figure 10 shows the prediction results in storage degradation by the battery state prediction device according to one embodiment, and Figure 11 shows the prediction results in cycle degradation by the battery state prediction device according to one embodiment.

[0162] In Figures 10 and 11(a), the horizontal axis represents time and the vertical axis represents the State of Health (SOH), which is the health status of the battery. In Figures 10 and 11(b), the horizontal axis represents time and the vertical axis represents the lithium ion loss (LLI). In Figures 10 and 11(c), the horizontal axis represents time and the vertical axis represents the active material loss at the positive electrode (LAMp). In Figures 10 and 11(d), the horizontal axis represents time and the vertical axis represents the active material loss at the negative electrode (LAMn).

[0163] Furthermore, in Figures 10 and 11(a) to (d), the points shown as shapes may represent experimental values ​​obtained through experiments, while the graphs shown as lines may represent simulated values ​​obtained through simulations.

[0164] Furthermore, in Figures 10 and 11 (a), (b), (c), and (d), the experimental values ​​(a-1), which are points shown as shapes, may represent results derived under the same conditions as the simulation values ​​(a-2), which are graphs shown as lines. Similarly, the experimental value (b-1) may represent results derived under the same conditions as the simulation values ​​(b-2).

[0165] In other words, Figure 10 could mean that (a-1) and (a-2) are results derived under the same conditions, (b-1) and (b-2) are results derived under the same conditions, (c-1) and (c-2) are results derived under the same conditions, (d-1) and (d-2) are results derived under the same conditions, and (e-1) and (e-2) are results derived under the same conditions.

[0166] Similarly, in Figure 11, (a-1) and (a-2) are results derived under the same conditions, and (b-1) and (b-2), (c-1) and (c-2), (d-1) and (d-2), (e-1) and (e-2), (f-1) and (f-2), and (g-1) and (g-2) can each be considered results derived under the same conditions.

[0167] Thus, according to the battery state prediction device of one embodiment, in Figures 10 and 11, the trend lines of the results enclosed by the same alphabet are obtained with similar outlines, so it can be analyzed that the theory and assumptions used in the prediction model are also valid in the actual environment.

[0168] Specifically, Figure 10 includes the results of predicting SOH, LLI, LAMp, and LAMn in storage degradation, and Figure 11 includes the results of predicting SOH, LLI, LAMp, and LAMn in cyclic degradation.

[0169] First, the battery state prediction device 1 according to one embodiment can improve the prediction accuracy of LLI by introducing an acceleration coefficient that is not considered in conventional technology, as shown in Figure 10(b). Furthermore, the battery state prediction device 1 according to one embodiment can improve the prediction accuracy of LAMp by introducing a cation mixing model and a volume change model due to particle damage, which are not considered in conventional technology, as shown in Figure 10(c).

[0170] Furthermore, the battery state prediction device 1 according to one embodiment can improve the prediction accuracy of LAMn by considering the correlation between LLI and LAMn, which is not considered in conventional techniques, as shown in Figure 10(d). As a result, the accuracy of SOH estimation can be improved as shown in Figure 10(a).

[0171] Next, the battery state prediction device 1 according to one embodiment, as shown in Figures 11(a) to (d), differs from conventional technology that only considers storage degradation models, in that it can model battery degradation by considering cycle degradation.

[0172] As a result, the battery state prediction device 1 according to one embodiment has the effect of improving the accuracy and precision of SOH estimation for predicting the state of the battery, as shown in Figures 10 and 11.

[0173] Figure 12 shows a control flowchart of a battery state prediction method according to one embodiment. Referring to Figure 12, at least one processor 110 can receive battery data from the battery (1200). The processor 110 can then generate a first predictive model related to LLI that predicts lithium loss (1210).

[0174] Furthermore, the processor 110 can generate a second predictive model related to LAMp that predicts active material loss at the positive electrode (1220), and a third predictive model related to LAMn that predicts active material loss at the negative electrode (1230).

[0175] Here, we will explain that the processor 110 generates the first to third prediction models, but the first to third prediction models may also refer to models that are stored in memory 120 and have been generated in advance.

[0176] Subsequently, the processor 110 can input the battery data into one of the prediction models to obtain battery state data (1240). In this case, the processor 110 may input the battery data into one of the prediction models and input its output value into another prediction model to obtain battery state data.

[0177] Specifically, the processor 110 can input the change in battery capacity derived by the first prediction model into a second or third prediction model to derive a degradation coefficient, and then predict the state of the battery based on the degradation coefficient.

[0178] In other words, the processor 110 can first calculate the capacity change due to LLI from the first prediction model, input it into the second and third prediction models, and derive the degradation coefficient based on equations 23 to 26.

[0179] Subsequently, the processor 110 can determine whether or not it has received a command to transmit battery status data (1250). If the processor 110 determines that it has received a command to transmit battery status data (YES in 1250), it can transmit the battery status data to the external device 5 (1260).

[0180] Thus, the battery state prediction device 1 according to one embodiment has the effect of predicting battery degradation to ensure the stability of battery use, providing an integrated model to reflect in battery warranties, and securing data for analyzing the causes of battery failure.

[0181] On the other hand, the disclosed embodiments may be implemented in the form of a recording medium that stores computer-executable instruction words. The instruction words may be stored in the form of program code, and a processor may generate a program module at runtime to perform the operations of the disclosed embodiments. The recording medium may be a computer-readable recording medium.

[0182] Computer-readable recording media include all types of recording media that store computer-readable instruction words. Examples include ROM (read-only memory), RAM (random access memory), magnetic tape, magnetic disks, flash memory, and optical data storage devices.

[0183] Furthermore, computer-readable recording media may be provided in the form of non-transitory storage media. Here, “non-transitory storage media” simply means a tangible device that does not contain signals (e.g., electromagnetic waves), and this term does not distinguish between cases where data is stored semi-permanently and cases where it is stored temporarily. As an example, “non-transitory storage media” may include buffers in which data is temporarily stored.

[0184] According to one embodiment, the methods according to the various embodiments disclosed herein may be provided in a computer program product. The computer program product may be traded as a commodity between a seller and a buyer. The computer program product may be distributed in the form of an instrument-readable recording medium (e.g., compact disc read-only memory (CD-ROM)) or through an application store (e.g., Play Store). TM The computer program product (e.g., download or upload) may be distributed online via a network or directly between two user devices (e.g., smartphones). In the case of online distribution, at least a portion of the computer program product (e.g., a downloadable app) may be at least temporarily stored or temporarily generated on an instrument-readable recording medium such as the memory of a manufacturer's server, an application store server, or an intermediary server.

[0185] Although all components constituting the embodiments disclosed in this document have been described as operating either as a single unit or in combination, the embodiments disclosed in this document are not necessarily limited to such embodiments. That is, within the scope of the purpose of the embodiments disclosed in this document, all components may operate in combination of one or more units.

[0186] Furthermore, terms such as “includes,” “constitutes,” or “possesses,” as described above, mean that they may contain the component in question, and not exclude other components, unless otherwise specified. All terms, including technical or scientific terms, have the same meaning as those generally understood by a person of ordinary skill in the art to which the embodiments disclosed herein belong, unless otherwise specified. Commonly used terms, such as those defined in dictionaries, should be interpreted to be consistent with their meaning in the context of the relevant technology, and not to be interpreted in an ideal or overly formal sense unless explicitly defined herein.

[0187] The above description is merely illustrative of the technical concept disclosed herein, and any person with ordinary skill in the art to which the embodiments disclosed herein belong can make various modifications and variations without departing from the essential characteristics of the embodiments disclosed herein. Therefore, the embodiments disclosed herein are for illustrative purposes only, not to limit the technical concept of the embodiments disclosed herein, and the scope of the technical concept disclosed herein is not limited by such embodiments. The scope of protection of the technical concept disclosed herein must be interpreted according to the claims described below, and all technical concepts within an equivalent scope should be interpreted as being included in the scope of rights of this document.

Claims

1. A communication interface for receiving battery data from the battery, A battery state prediction model is generated which includes a first prediction model for predicting lithium loss, a second prediction model for predicting active material loss in the positive electrode, and a third prediction model for predicting active material loss in the negative electrode, and at least one processor is used to input battery data into the battery state prediction model to acquire battery state data. A battery state prediction device, including a battery state prediction device.

2. The aforementioned at least one processor is The battery state prediction device according to claim 1, which generates the first prediction model based on an acceleration coefficient that quantifies the stress experienced at the electrode and solid electrolyte interface.

3. The aforementioned at least one processor is The battery state prediction device according to claim 2, wherein the change in the chemical ratio of the active substance due to lithium loss is determined based on the first prediction model, and the open-circuit potentials of the positive electrode and the negative electrode are determined based on the change in the chemical ratio of the active substance.

4. The aforementioned at least one processor is The battery state prediction device according to claim 3, which predicts the state of the battery based on the amount of change in the open-circuit potential.

5. The aforementioned at least one processor is A battery state prediction device according to any one of claims 1 to 4, which generates the second prediction model based on cation mixing and volume changes due to particle damage.

6. The aforementioned at least one processor is The battery state prediction device according to claim 5, wherein the battery data is input to the second prediction model to determine the degree of battery degradation.

7. The aforementioned at least one processor is A battery state prediction device according to any one of claims 1 to 4, which determines the correlation between the lithium loss and the active substance loss in the negative electrode, and generates the third prediction model based on the correlation.

8. The aforementioned at least one processor is A battery state prediction device according to any one of claims 1 to 4, comprising inputting the change in battery capacity derived by the first prediction model into the second prediction model or the third prediction model to derive a degradation coefficient, and predicting the state of the battery based on the degradation coefficient.

9. Receiving battery data from the battery, To generate a battery state prediction model that includes a first prediction model for lithium loss, a second prediction model for active material loss in the positive electrode, and a third prediction model for active material loss in the negative electrode, The aforementioned battery data is input into a battery state prediction model to obtain battery state data, A method for predicting battery state, including the following.

10. Generating the first prediction model is The battery state prediction method according to claim 9, comprising generating the first prediction model based on an acceleration coefficient that quantifies the stress experienced at the electrode and solid electrolyte interface.

11. The battery state prediction method according to claim 10, further comprising determining the change in the chemical ratio of the active substance due to lithium loss based on the first prediction model, and determining the open-circuit potentials of the positive electrode and the negative electrode based on the change in the chemical ratio of the active substance.

12. The battery state prediction method according to claim 11, further comprising predicting the state of the battery based on the amount of change in the open-circuit potential.

13. Generating the second prediction model means A battery state prediction method according to any one of claims 9 to 12, comprising generating the second prediction model based on cation mixing and volume changes due to particle damage.

14. The battery state prediction method according to claim 13, further comprising inputting the battery data into the second prediction model to determine the degree of battery degradation.

15. Generating the third prediction model is A battery state prediction method according to any one of claims 9 to 12, comprising determining the correlation between the lithium loss and the active material loss in the negative electrode, and generating the third prediction model based on the correlation.

16. A battery state prediction method according to any one of claims 9 to 12, further comprising inputting the change in battery capacity derived by the first prediction model into the second prediction model or the third prediction model to derive a degradation coefficient, and predicting the state of the battery based on the degradation coefficient.

Images