Battery monitoring device and method

Abstract

The present disclosure relates to a battery monitoring device and method for monitoring the chemical state of an electrolyte of a battery by using terahertz waves, and the technical problem to be solved is to provide a battery monitoring device and method for analyzing various chemical states of an electrolyte, such as lithium ion concentration, solvent composition, electrolyte deterioration, ionic conductivity, and lithium dendrite formation, by using terahertz waves. To this end, the present disclosure comprises: a processor; and a memory for storing instructions executable by the processor, wherein the processor extracts spectral features for respective frequency regions of a preset terahertz wave received from a terahertz sensor, inputs the extracted spectral features into a pre-trained machine learning model, and monitors the state of an electrolyte according to frequency bands through the machine learning model.

Description

Battery monitoring device and method

[0001] The present disclosure relates to a battery monitoring device and method.

[0002]

[0003] The performance and lifespan of lithium-ion batteries depend significantly on the state of the electrolyte, which acts as a transport medium for lithium ions. Most xEV battery electrolytes currently in use consist of a mixture of lithium salts and organic solvents.

[0004] Battery management systems play a crucial role in monitoring and controlling battery voltage, current, temperature, and other parameters to optimize safety, performance, and lifespan. However, battery management systems have limitations in accurately assessing the battery's internal state due to their restricted ability to directly monitor the chemical state of the electrolyte. Consequently, there is a growing need for technology capable of precisely and real-time monitoring of the battery's internal condition.

[0005] The information described above disclosed in the background technology of this invention is intended only to enhance understanding of the background of the present invention and may therefore include information that does not constitute prior art.

[0006]

[0007] The purpose of the present invention is to provide a battery monitoring device and method for monitoring the chemical state of a battery electrolyte using terahertz waves.

[0008] However, the technical problems that the present invention aims to solve are not limited to those described above, and other unmentioned problems can be clearly understood by those skilled in the art from the description of the invention below.

[0009]

[0010] A battery monitoring device according to one embodiment of the present invention for solving the above technical problem comprises: a processor; and a memory for storing instructions executed by the processor, wherein the processor includes a processor that extracts spectrum features according to a frequency range of preset terahertz waves received by a terahertz sensor and inputs them into preset machine learning to monitor the state of an electrolyte according to the frequency band through the machine learning model.

[0011] The processor of the present invention is characterized by estimating the lithium ion concentration of the electrolyte based on terahertz waves in a first frequency range.

[0012] The processor of the present invention is characterized by calculating the solvent composition ratio of the electrolyte based on terahertz waves in the second frequency range.

[0013] The processor of the present invention is characterized by detecting the degradation of the electrolyte based on terahertz waves in the third frequency range and estimating the ionic conductivity of the electrolyte by analyzing the frequency dependence of multiple dielectric constants.

[0014] The characteristics of the spectrum of the present invention include at least one of the peak of the spectrum, the peak area of ​​the spectrum, and the major fluctuation component of the spectrum.

[0015] The processor of the present invention is characterized by estimating the electrolyte state through the machine learning model, correcting the estimated electrolyte state using an LSTM network, and finally outputting the electrolyte state by assigning confidence to the corrected electrolyte state through fuzzy logic and rule-based inference.

[0016] The processor of the present invention is characterized by using the terahertz sensor to extract the electrolyte state at multiple points and 3D mapping the multiple electrolyte states to extract the spatial distribution of the electrolyte state within the battery.

[0017] A battery monitoring method according to one aspect of the present invention is characterized by comprising: a step in which a processor extracts spectrum features according to a frequency range of preset terahertz waves received by a terahertz sensor; and a step in which the processor inputs the spectrum features into preset machine learning and monitors the state of an electrolyte according to the frequency band through the machine learning model.

[0018]

[0019] According to the present invention, various chemical states of an electrolyte, such as lithium ion concentration, solvent composition, electrolyte degradation, ion conductivity, and lithium dendrite formation, can be analyzed using terahertz waves.

[0020] According to the present invention, the internal state of the battery can be analyzed without opening the battery, allowing for continuous and real-time monitoring even during battery operation.

[0021] According to the present invention, various data regarding the electrolyte can be acquired at high speed in milliseconds, thereby enabling the capture of rapid changes in the state of the electrolyte that occur during rapid charging.

[0022] According to the present invention, it is possible to comprehensively evaluate the condition of an electrolyte by simultaneously analyzing various parameters, such as the concentration, composition, and ionic conductivity of the electrolyte, with a single measurement.

[0023] According to the present invention, local changes and distribution of the electrolyte can be identified with micrometer-level spatial resolution, thereby enabling the detection and mapping of non-uniformity within the electrolyte.

[0024] However, the effects obtainable through the present invention are not limited to those described above, and other unmentioned technical effects will be clearly understood by those skilled in the art from the description of the invention below.

[0025]

[0026] The following drawings attached to this specification illustrate preferred embodiments of the present invention and serve to further enhance understanding of the technical concept of the present invention together with the detailed description of the invention provided below; therefore, the present invention should not be interpreted as being limited only to the matters described in such drawings.

[0027] FIG. 1 is a block diagram of a battery monitoring device according to one embodiment of the present invention.

[0028] FIG. 2 is a flowchart of a battery monitoring method according to one embodiment of the present invention.

[0029] FIG. 3 is a diagram showing raw data according to one embodiment of the present invention.

[0030] FIG. 4 is a diagram showing converted data (frequency domain) according to one embodiment of the present invention.

[0031] FIG. 5 is an example of normalized data according to one embodiment of the present invention.

[0032] FIG. 6 is a diagram showing the characteristics of a spectrum according to one embodiment of the present invention.

[0033] Figure 7 shows the state of the electrolyte estimated through a machine learning model according to one embodiment of the present invention.

[0034] FIG. 8 is a diagram showing the state of the electrolyte corrected through time series analysis according to one embodiment of the present invention.

[0035] FIG. 9 is a diagram showing the state of a reliable electrolyte according to one embodiment of the present invention.

[0036]

[0037] Preferred embodiments of the present invention will be described in detail below with reference to the attached drawings. Prior to this, terms and words used in this specification and claims should not be interpreted as being limited to their ordinary or dictionary meanings. Instead, based on the principle that the inventor may appropriately define the concepts of terms to best describe his invention, they should be interpreted in a meaning and concept consistent with the technical spirit of the present invention. Therefore, it should be understood that the embodiments described in this specification and the configurations illustrated in the drawings are merely some of the most preferred embodiments of the present invention and do not represent all of the technical spirit of the present invention; thus, various equivalents and modifications that can replace them may exist at the time of filing this application. Furthermore, as used in this specification, "comprise" or "include" and / or "comprising" or "including" specify the presence of the mentioned features, numbers, steps, actions, parts, elements, and / or groups thereof, and do not exclude the presence or addition of one or more other features, numbers, actions, parts, elements, and / or groups. In addition, when describing embodiments of the present invention, "can" and "can" may include "one or more embodiments of the present invention."

[0038] Additionally, to aid in understanding the invention, the attached drawings are not drawn to actual scale, and the dimensions of some components may be exaggerated. Furthermore, the same reference numerals may be assigned to identical components in different embodiments.

[0039] The statement that two subjects of comparison are 'identical' means that they are 'substantially identical.' Therefore, substantial identity may include deviations considered low in the industry, for example, deviations within 5%. Additionally, the statement that a parameter is uniform in a given area may mean that it is uniform from an average perspective.

[0040] Although terms such as "first," "second," etc., are used to describe various components, it goes without saying that these components are not limited by these terms. These terms are used merely to distinguish one component from another, and unless specifically stated otherwise, the first component may also be the second component.

[0041] Throughout the specification, unless specifically stated otherwise, each component may be singular or plural.

[0042] The fact that any configuration is placed on the "upper (or lower)" of a component or on the "upper (or lower)" of a component may mean not only that any configuration is placed in contact with the upper (or lower) surface of said component, but also that another configuration may be interposed between said component and any configuration placed on (or below) said component.

[0043] Furthermore, where it is stated that one component is "connected," "coupled," or "connected" to another component, it should be understood that while said components may be directly connected or connected to each other, another component may be "interposed" between each component, or that each component may be "connected," "coupled," or "connected" through another component. Additionally, when it is stated that a part is electrically coupled with another part, this includes not only cases where they are directly connected but also cases where they are connected with another component in between.

[0044] Throughout the specification, "A and / or B" means A, B, or A and B unless specifically stated otherwise. That is, "and / or" includes any combination or any combination of the enumerated items. "C to D" means C or more and D or less, unless specifically stated otherwise.

[0045] FIG. 1 is a block diagram of a battery monitoring device according to one embodiment of the present invention.

[0046] Referring to FIG. 1, a battery monitoring device according to one embodiment of the present invention may include a terahertz sensor (100), a communication unit (200), a user interface unit (400), a memory (300), and a processor (500).

[0047] The terahertz sensor (100) emits terahertz waves in the 0.1 to 10 THz band toward the battery (10) and can detect terahertz waves reflected from the battery (10) or transmitted through the battery (10).

[0048] Terahertz waves in the 0.1–10 THz band directly interact with the rotational and vibrational modes of molecules in the electrolyte, and are particularly sensitive to the collective vibrational modes of hydrogen bonding networks. In this embodiment, monitoring the state of the electrolyte using terahertz waves in the 0.5–4 THz band is described as an example.

[0049] Terahertz waves can penetrate non-conductive materials, allowing the electrolyte state of the battery (10) to be analyzed without opening the battery (10).

[0050] The battery (10) may be a battery (10) of an electric vehicle, but is not specifically limited.

[0051] The battery (10) may be a battery cell included in a battery pack. The battery cells may be connected in series, in parallel, or in a combination of series and parallel. The battery cell may be a rechargeable secondary battery. For example, the battery cell may include a nickel-cadmium battery, a lead-acid battery, a nickel-metal hydride battery (NiMH), a lithium-ion battery, a lithium polymer battery, etc. The number of battery cells may be determined according to the required output voltage. Additionally, the battery (10) may be a prismatic battery, a cylindrical battery, or a pouch-type battery.

[0052] The terahertz sensor (100) may be an antenna module to which a terahertz spectroscopic system is applied. The terahertz sensor (100) may include a terahertz source that emits terahertz waves, a terahertz detector that detects terahertz waves, a beam focusing system that focuses terahertz waves, an optical window that transmits terahertz waves, and a galvanometer scanner that scans the surface of the battery (10) at high speed.

[0053] The terahertz wave source can generate terahertz waves in the 0.5 to 4 THz band using a photoconductive antenna array. The terahertz wave source generates terahertz waves using a small femtosecond laser and can control the terahertz waves within the output frequency range of 0.5 to 4 THz. The terahertz wave source can operate in pulse mode, thereby minimizing heat generation.

[0054] The terahertz detector can detect terahertz waves that pass through or are reflected from the battery (10). The terahertz detector can detect minute signal changes with high sensitivity and broadband characteristics by using a superconducting hot electron bolometer.

[0055] The beam focusing system can focus terahertz waves. The beam focusing system is equipped with an off-axis parabolic mirror to focus the terahertz waves so that they are suitable for penetrating the aluminum housing of the battery (10).

[0056] The optical window may be made of a high-density polyethylene material with relatively excellent transmittance to transmit terahertz waves and may be installed in an aluminum housing of the battery (10).

[0057] The terahertz sensor (100) can be implemented in various forms or structures depending on the type, shape, or structure of the battery (10).

[0058] Generally, the battery (10) can be a rectangular battery, a pouch type, or a cylindrical type.

[0059] For example, if the battery (10) is a prismatic battery, the terahertz sensor (100) may be manufactured in a modular form and attached to the outside of the battery (10), i.e., an aluminum case. To attach the terahertz sensor (100) to the prismatic battery (10), a thermally conductive adhesive with excellent thermal management and signal transmission efficiency may be used. Additionally, as the terahertz sensor (100) is manufactured in a modular form, maintenance and upgrades of the terahertz sensor (100) can be easily performed. The terahertz sensor (100) attached to the prismatic battery (10) may have a lithium ion concentration measurement accuracy of ±0.01 mol / L, a sensitivity to detect electrolyte degradation capable of detecting a change of 0.1% of the total volume, a spatial resolution of 500 μm, and a measurement speed of 1 point / s.

[0060] When the battery (10) is cylindrical, the terahertz sensor (100) can be integrated near the electrolyte inlet of the cylindrical battery. The terahertz sensor (100) attached to the cylindrical battery has a measurement frequency range of 0.1 to 2 THz, a spectral resolution of 5 GHz, a lithium ion concentration measurement accuracy of 0.05 mol / L, an electrolyte degradation detection sensitivity capable of detecting a change of 0.1% of the total volume, an ion conductivity estimation error of less than 5%, and a power consumption of less than 100 mW.

[0061] The terahertz sensor (100) can be combined with a battery (10) made of a different material. To do this, the battery (10) may have a portion of its aluminum housing replaced with a terahertz viewing window. In this case, the terahertz sensor (100) may have an optical window, for example, a high-density polyethylene window, bonded to the terahertz viewing window of the aluminum housing, thereby minimizing signal loss.

[0062] The terahertz sensor (100) can be implemented as a multi-sensor array. That is, multiple terahertz sensors (100) may be provided and attached to the surface of the battery (10). The number or location of attachments to the surface of the battery (10) can be set in various ways depending on the type or size of the battery (10) and is not specifically limited. As multiple terahertz sensors (100) are provided and the electrolyte state at the corresponding attachment point is analyzed based on each terahertz sensor (100), the spatial distribution of the electrolyte state within the battery (10) can be extracted through 3D mapping of these electrolyte states. This will be described later.

[0063] The user interface unit (400) can provide a user interface. The user interface unit (400) can receive a control command to monitor the electrolyte status of the battery (10). The user interface unit (400) can receive and output various data and processing results (electrolyte status) for analyzing the electrolyte status from the processor (500). Meanwhile, the processor (500) can analyze the electrolyte status by analyzing the terahertz wave data received from the terahertz sensor (100) without receiving a control command from the user interface unit (400).

[0064] The user interface unit (400) may include, for example, a keyboard, a mouse, a touchpad, a touchscreen, an electronic pen, a touch button, etc. The user interface unit (400) may include a printer, a display, etc. to output data. Here, the display may be implemented as, for example, a TFT-LCD (thin film transistor-liquid crystal display) panel, an LED (light emitting diode) panel, an OLED (organic LED) panel, an AMOLED (active matrix OLED) panel, or a flexible panel.

[0065] The communication unit (200) can provide a communication interface. The communication unit (200) can transmit the electrolyte status to a Battery Management System (BMS), an energy storage system, or a server managing them via a communication network. The communication unit (200) can periodically transmit the electrolyte status at intervals of 100ms.

[0066] For the communication network, SPI (Serial Peripheral Interface Bus) is used as the main communication network, and UART (Universal asynchronous receiver / transmitter) or I2C (Inter-Integrated Circuit) may be used as the auxiliary communication network. In addition, the communication network may employ 3GPP (3rd Generation Partnership Project), LTE (Long Term Evolution), 5G (Generation), WIMAX (World Interoperability for Microwave Access), wired and wireless internet, LAN (Local Area Network), Wireless LAN (Wireless Local Area Network), WAN (Wide Area Network), PAN (Personal Area Network), Bluetooth, Wifi (Wireless Fidelity), etc., but is not specifically limited.

[0067] Meanwhile, the battery management system can comprehensively evaluate the state of the battery (10) based on the state of the battery (10) received from the communication unit (200). In addition, the battery management system can establish a charging and discharging strategy for the battery (10) and predict the lifespan of the battery (10) based on the collected electrolyte state.

[0068] The memory (300) can store various data used by the processor (500). The data may include instructions for performing operations or steps according to an embodiment of the present invention. That is, the memory (300) can store instructions for monitoring the electrolyte status of the battery (10).

[0069] The memory (300) may include at least one storage medium among a flash memory type, a hard disk type, a multimedia card micro type, a card type memory, RAM (Random Access Memory), SRAM (Static Random Access Memory), ROM (Read-Only Memory), PROM (Programmable Read-Only Memory), EPROM (Erasable Programmable Read-Only Memory), and EEPROM (Electrically Erasable Programmable Read-Only Memory).

[0070] The processor (500) may be connected to the memory (300), and the memory (300) may store instructions for performing operations, steps, etc., according to an embodiment of the present invention. Here, the memory (300) may include a magnetic storage medium or a flash storage medium in addition to a volatile storage device that requires power to maintain stored information, but the scope of the present invention is not limited thereto.

[0071] Additionally, the processor (500) may be configured to perform each function separately at the hardware, software, or logic level. In this case, dedicated hardware may be used to perform each function. To this end, the processor (500) may be implemented as or include at least one of an ASIC (Application Specific Integrated Circuit), DSP (Digital Signal Processor), PLD (Programmable Logic Devices), FPGAs (Field Programmable Gate Arrays), CPU (Central Processing Unit), microcontrollers, and / or microprocessors.

[0072] The processor (500) may be implemented as a Central Processing Unit (CPU) or a System on Chip (SoC), and may control multiple hardware or software components connected to the processor (500) by running an operating system or application, and may perform various data processing and operations. The processor (500) may be configured to execute at least one instruction stored in memory (300) and store the execution result data in memory (300).

[0073] The processor (500) may be included within the battery management system or the Electronic Control Unit (ECU) of an electric vehicle. According to another embodiment, the processor (500) may be included within the controller of an energy storage system. According to yet another example, the processor (500) may be implemented in a server that is connected to the battery management system or the energy storage system via communication.

[0074] The processor (500) can control the terahertz sensor (100) to emit terahertz waves and detect terahertz waves that are reflected and transmitted from the battery (10).

[0075] The processor (500) can monitor the electrolyte state of the battery (10) by analyzing the data of the terahertz waves detected by the terahertz sensor (100) through a pre-trained machine learning model.

[0076] The processor (500) can extract a spectrum in the frequency domain from terahertz waves detected by the terahertz sensor (100). That is, the processor (500) can detect a terahertz waveform in the time domain from the data (raw data) of the terahertz waves detected by the terahertz sensor (100) through an electro-optical sampling method. The processor (500) can convert the detected terahertz waveform in the time domain into a spectrum in the frequency domain by performing a Fast Fourier Transform.

[0077] The processor (500) can analyze the state of an electrolyte by extracting features of the spectrum in the frequency domain and inputting the extracted features of the spectrum into a machine learning model that has been trained. The processor (500) can analyze the chemical state of various types of electrolytes according to the frequency band of the terahertz waves.

[0078] A machine learning model can take spectral features as input and output the state of an electrolyte. The machine learning model learns spectral patterns using deep neural networks and can perform quantitative analysis using partial least squares regression. The electrolyte state output by the machine learning model is a predicted value and may have relatively low accuracy.

[0079] Accordingly, the processor (500) can correct the electrolyte state output from the machine learning model using a Long Short-Term Memory (LSTM) network. Then, the processor (500) can finally obtain the electrolyte state by giving confidence to the corrected electrolyte state through fuzzy logic and rule-based inference.

[0080] When a plurality of terahertz sensors (100) are provided, the processor (500) can obtain the electrolyte state from the data of each of these terahertz sensors (100). In this case, the processor (500) can extract the spatial distribution of the electrolyte state within the battery (10) by 3D mapping the electrolyte state obtained based on the terahertz waves of each terahertz sensor (100).

[0081] Electrolyte conditions may include lithium ion concentration, solvent composition ratio, electrolyte degradation, ion conductivity, lithium dendrites, and changes in electrolyte conditions.

[0082] The processor (500) can measure the lithium ion concentration of the electrolyte in the terahertz wave band of the first frequency range. The first frequency range may be 0.5 to 2 THz. That is, the processor (500) can detect changes in terahertz absorption due to changes in the hydrogen structure of the main lithium ions in the 0.5 to 2 THz range. The processor (500) can measure the lithium ion concentration in the electrolyte with an accuracy of ppb (parts per billion) through changes in terahertz absorption.

[0083] Specifically, the first frequency region has characteristics that respond very sensitively to the interaction between lithium ions and solvent molecules in the electrolyte. The solvation shell formed around the lithium ions exhibits unique collective vibration modes near 0.8 THz, 1.2 THz, and 1.7 THz, and the intensity of these peaks is directly correlated with the concentration of lithium ions. Accordingly, the processor (500) can non-invasively measure the lithium ion concentration in the electrolyte by precisely modeling the correlation between the peak intensity and the lithium ion concentration. That is, the processor (500) can detect characteristic peaks in a noise-robust manner by applying a continuous wavelet transform (CWT). The processor (500) can identify peaks at multiple scales using a 'Mexican hat' wavelet and quantify the area of ​​each peak through trapezoidal numerical integration. The processor (500) converts this peak information, namely the area of ​​the peak, into concentration-dependent multivariate data and can estimate the lithium ion concentration with an accuracy of ±0.01 mol / L through a partial least squares regression (PLS-R) model. Accordingly, the solvation state of the lithium ions can be monitored simultaneously, thereby obtaining additional information regarding the ion mobility and the conductivity characteristics of the electrolyte. Measuring the lithium ion concentration in the electrolyte through changes in terahertz absorption ensures consistent measurement results in various operating environments through a temperature correction algorithm, and enables the detection of non-uniformity in lithium ion distribution within the battery cell through real-time concentration change tracking, and allows for the early detection of localized ion depletion phenomena that may occur during rapid charging.

[0084] The processor (500) can calculate the solvent composition ratio of the electrolyte in the terahertz band of the second frequency range. The second frequency range may be the 1 to 3 THz range. That is, the processor (500) can distinguish the rotational and vibrational modes of organic solvent molecules such as ethylene carbonate (EC) and dimethyl carbonate (DMC) in the 1 to 3 THz range, and calculate the solvent composition ratio of the organic solvent based on the distinguished rotational and vibrational modes of the organic solvent molecules. In this case, the processor (500) can quantify the solvent composition ratio with an error of within 0.1%.

[0085] Specifically, the main organic solvents used in xEV battery electrolytes, such as ethylene carbonate (EC), dimethyl carbonate (DMC), and propylene carbonate (PC), have unique rotational and vibrational modes according to their respective molecular structures, which appear as characteristic patterns in the terahertz spectrum in the 1–3 THz range. EC may show major absorption peaks at 1.4 THz and 2.8 THz, DMC may show major absorption peaks at 1.1 THz and 2.2 THz, and PC may show major absorption peaks at 1.6 THz and 2.5 THz. The relative intensity and position of these multiple peaks change systematically depending on the composition ratio of the solvent. Accordingly, the processor (500) can utilize advanced multivariate statistical techniques to effectively analyze high-dimensional information of the terahertz spectrum data.

[0086] First, the processor (500) can reduce the dimensionality of the spectrum data and extract major fluctuating components through Principal Component Analysis (PCA). Next, the processor (500) can model the correlation between the solvent composition ratio and the spectrum pattern by applying Partial Least Squares Discriminant Analysis (PLS-DA). In the modeling process, the processor (500) can utilize a database built with standard samples of various composition ratios as training data and can ensure the robustness of the model through cross-validation. This analysis model can quantify the ratio of each solvent component from complex terahertz spectra with an error of within 0.1% and can provide stable results even under actual operating conditions through correction algorithms for external environmental changes such as temperature and pressure. In addition, the analysis model can detect phenomena such as selective evaporation or decomposition of the solvent occurring during battery life at an early stage by monitoring the temporal change trend of the solvent composition ratio.

[0087] The processor (500) can detect the degradation of the electrolyte in the terahertz wave band of the third frequency range. The third frequency range may be in the 2–4 THz range. That is, the processor (500) can detect the decomposition of the electrolyte in the early stages by monitoring the characteristic absorption peaks of degradation products such as LiF and POF3 in the 2–4 THz range. This is a level at which changes of less than 0.01% of the total electrolyte volume can be detected. In addition, the processor (500) can estimate accurate ion conductivity by directly observing the movement of ions through the measurement of the complex permittivity of the terahertz waves. The processor (500) can detect the formation of lithium dendrites early by utilizing the high sensitivity of the terahertz waves. That is, the processor (500) can detect the fine precipitation of lithium metal in the early stages. This is a level at which detection is possible even when the length of the dendrite is less than 1 μm.

[0088] Specifically, during battery operation, the electrolyte degrades through various chemical and electrochemical reactions, and decomposition products such as LiF, POF3, and HF generated during this process exhibit characteristic absorption peaks in the 2–4 THz range. LiF shows absorption due to a unique vibrational mode around 2.5 THz, POF3 shows absorption due to a unique vibrational mode around 3.2 THz, and HF shows absorption due to a unique vibrational mode around 3.8 THz. The appearance of these peaks and changes in intensity can be used as direct indicators of electrolyte degradation.

[0089] Accordingly, the processor can derive a complex permittivity spectrum by extracting both amplitude and phase information from the time-domain waveform acquired through terahertz time-domain spectroscopy (THz-TDS). The real part of the complex permittivity reflects the dielectric properties of the electrolyte, while the imaginary part reflects the absorption properties. In particular, the frequency dependence of the imaginary part has a direct relationship with ionic conductivity through the Drood-Smith model. A deep learning-based anomaly detection algorithm can be applied for the detection of degradation products.

[0090] The anomaly detection algorithm uses an autoencoder neural network to learn the spectral patterns of steady-state electrolytes and can detect anomalous spectra based on reconstruction errors. These detected anomaly patterns can be interpreted as the types and concentrations of specific degradation products through comparison with a reference database.

[0091] To estimate ionic conductivity, an extended Drude model can be applied to complex permittivity data. The Drude model incorporates parameters such as ion mobility, concentration, and effective mass, and optimal model parameters can be fitted to experimental data through nonlinear optimization techniques. Since the derived ionic conductivity value is directly related to the battery's internal resistance, it can serve as an important indicator for predicting power performance degradation.

[0092] Accordingly, electrolyte degradation can be detected in real time from the early stages, and trace degradation products at the ppm level can also be identified. In addition, local anomalies such as lithium dendrite formation can be detected early through spatial distribution mapping of ion conductivity, and based on this, battery safety can be significantly improved.

[0093] In this way, the processor (500) can improve the accuracy of the electrolyte state by extracting a direct chemical state at the molecular level.

[0094] The processor (500) uses femtosecond laser-based terahertz time domain spectroscopy, so high-speed data acquisition in milliseconds is possible, and through this, rapid state changes of the electrolyte occurring during rapid charging can be captured.

[0095] The processor (500) can analyze the state of the electrolyte in a single measurement through the amplitude and phase of the terahertz waves.

[0096] Since the processor (500) uses terahertz waves, it can detect local changes and distribution of the electrolyte with micrometer-level spatial resolution.

[0097] Hereinafter, a battery monitoring method according to one embodiment of the present invention will be described with reference to FIGS. 2 to 9.

[0098] FIG. 2 is a flowchart of a battery monitoring method according to an embodiment of the present invention, FIG. 3 is a diagram showing raw data according to an embodiment of the present invention, FIG. 4 is a diagram showing transformed data (frequency domain) according to an embodiment of the present invention, FIG. 5 is an example diagram of normalized data according to an embodiment of the present invention, FIG. 6 is a diagram showing the characteristics of a spectrum according to an embodiment of the present invention, FIG. 7 is a state of an electrolyte estimated through a machine learning model according to an embodiment of the present invention, FIG. 8 is a diagram showing the state of an electrolyte corrected through time series analysis according to an embodiment of the present invention, and FIG. 9 is a diagram showing the state of an electrolyte with assigned reliability according to an embodiment of the present invention.

[0099] Referring to FIG. 2, first, the terahertz sensor (100) emits terahertz waves in the 0.1 to 4 THz band toward the battery (10) and can detect terahertz waves reflected from the battery (10) or transmitted through the battery (10).

[0100] The processor (500) can acquire data (raw data) of terahertz waves detected by the terahertz sensor (100) (S100). This data may consist of electric field strength values ​​in the time domain, as shown in FIG. 3.

[0101] The processor (500) can preprocess data acquired from the terahertz sensor (100) (S200). The preprocessed data is as illustrated in FIG. 3. That is, the processor (500) can remove high-frequency noise using wavelet transform. In this case, the processor (500) can perform a four-step decomposition using the 'db4' wavelet and remove noise coefficients by applying a soft threshold method. The processor (500) can estimate a baseline using the asymmetric least squares method and remove noise through this baseline. The processor (500) can calculate the time delay between the data and the reference signal using the cross-correlation method and align the data by shifting it by the calculated time delay.

[0102] The processor (500) can convert the preprocessed data into a frequency domain as illustrated in FIG. 4 (S300). That is, the processor (500) can increase the number of data points by applying zero padding to the preprocessed time domain data. The processor (500) can reduce spectrum leakage by applying a Hamming window function and obtain a frequency spectrum in the form of complex numbers by performing a Fast Fourier Transform. The processor (500) can obtain an amplitude spectrum by calculating the absolute value of the frequency spectrum in the form of complex numbers.

[0103] The processor (500) can normalize the spectrum as illustrated in FIG. 5 (S400). That is, the processor (500) measures the terahertz transmission spectrum of an empty battery (10) without an electrolyte. This may be a reference spectrum. The processor (500) can calculate the transmittance by dividing the sample spectrum by the reference spectrum. Additionally, the processor (500) can convert the transmittance into absorbance by taking the negative logarithm of the transmittance.

[0104] The processor (500) can extract features of the spectrum as illustrated in FIG. 6 (S500). The features of the spectrum may include peaks of the spectrum, peak areas of the spectrum, and the first five principal components among the major fluctuation components of the spectrum. That is, the processor (500) can detect peaks of the spectrum using a successive wavelet transform. In this case, the processor (500) can identify peaks of the spectrum at multiple scales using a 'Maxican hat' wavelet. The processor (500) can calculate the peripheral area of ​​the detected peaks of the spectrum. The processor (500) can extract major fluctuation components by performing Principal Component Analysis (PCA) on the entire spectrum. In this case, the processor (500) can select the first five principal components among the major fluctuation components and use them as feature vectors.

[0105] The processor (500) can apply the characteristics of the spectrum to a machine learning model (S600). To do this, the machine learning model must be trained in advance. The processor (500) trains the machine learning model using a training data set, and in this case, can evaluate the generalization performance of the machine learning model by performing 5-fold cross-validation. Once the machine learning model training is completed through this process, the processor (500) can apply the aforementioned characteristics of the spectrum to the machine learning model to estimate the electrolyte state as illustrated in FIG. 7.

[0106] Additionally, the processor (500) may perform time series analysis and result interpretation steps to ensure the accuracy and reliability of the estimated electrolyte state as described above. Here, the processor (500) may standardize the characteristics of the spectrum (mean 0, standard deviation 1) through data scaling and use them as input values ​​for training a machine learning model. A Random Forest classifier and a regression model may be selected as machine learning models. The classifier classifies the electrolyte state, for example, normal, early deterioration, or deterioration in progress, and the regression model can estimate the lithium ion concentration.

[0107] The processor (500) can perform time series analysis on the electrolyte state obtained through a machine learning model (S700). That is, the processor (500) can organize the continuous electrolyte state obtained through a machine learning model into time series data. The processor (500) can apply an LSTM network to the organized time series data. Then, the processor (500) can predict the state change over a preset time, for example, 10 hours, as shown in FIG. 8.

[0108] The processor (500) can analyze the corrected electrolyte state (S800). That is, the processor (500) can obtain the state of the electrolyte by assigning confidence to the electrolyte state through fuzzy logic and rule-based inference. To this end, the processor (500) can convert the electrolyte states into a fuzzy set by applying fuzzy logic. The processor (500) can evaluate the electrolyte state by applying rule-based inference, that is, a predefined IF-THEN rule, to the fuzzy set. The processor (500) can calculate the confidence interval of the electrolyte state using Bayesian inference. Based on the electrolyte state and confidence, the processor (500) can finally determine the electrolyte state as illustrated in FIG. 9.

[0109] The processor (500) can store and transmit the electrolyte status to the battery management system (S900). The processor (500) can add and encrypt metadata regarding the measurement time, electrolyte status, identification information of the terahertz sensor (100), battery information, and reliability, and then control the communication unit (200) to transmit it to the battery management system via a communication network, e.g., SPI. The battery management system can use the received data to comprehensively evaluate the electrolyte status, establish an optimal charge / discharge strategy based on this, and predict the lifespan of the battery (10).

[0110] As such, the battery monitoring device and method according to one embodiment of the present invention can analyze various chemical states of the electrolyte, such as lithium ion concentration, solvent composition, electrolyte degradation, ion conductivity, and lithium dendrite formation, by using terahertz waves.

[0111] A battery monitoring device and method according to one embodiment of the present invention can analyze the internal state of a battery without opening the battery, thereby enabling continuous and real-time monitoring even during battery operation.

[0112] A battery monitoring device and method according to one embodiment of the present invention can acquire various data regarding the electrolyte at high speed in milliseconds, thereby enabling the capture of rapid changes in the state of the electrolyte that occur during rapid charging.

[0113] A battery monitoring device and method according to one embodiment of the present invention enables a comprehensive evaluation of the condition of an electrolyte by simultaneously analyzing various parameters, such as the concentration, composition, and ionic conductivity of the electrolyte, through a single measurement.

[0114] A battery monitoring device and method according to one embodiment of the present invention can identify local changes and distributions of an electrolyte with micrometer-level spatial resolution, thereby enabling the detection and mapping of non-uniformity within the electrolyte.

[0115] As used herein, the terms “part” and “module” may include units implemented in hardware, software, or firmware, and may be used interchangeably with terms such as logic, logic block, component, or circuit. “Part” and “module” may be a component formed integrally, or a minimum unit of said component or a part thereof that performs one or more functions. For example, according to one embodiment, “part” and “module” may be implemented in the form of an Application-Specific Integrated Circuit (ASIC).

[0116] Although the present invention has been described above with reference to limited embodiments and drawings, the present invention is not limited thereto, and it is obvious that various modifications and variations are possible within the scope of the technical spirit of the present invention and the equivalent scope of the claims described below by those skilled in the art to which the present invention belongs.

Claims

processor; and It includes memory that stores instructions executed by the above processor, and A battery monitoring device comprising a processor that extracts spectral features according to frequency ranges of preset terahertz waves received by a terahertz sensor and inputs them into preset machine learning to monitor the state of the electrolyte according to the frequency band through the machine learning model. In paragraph 1, The above processor is, A battery monitoring device that estimates the lithium ion concentration of an electrolyte based on terahertz waves in the first frequency range. In paragraph 1, The above processor is, Battery monitoring device that calculates the solvent composition ratio of an electrolyte based on terahertz waves in the second frequency range . In paragraph 1, The above processor is, A battery monitoring device that detects the degradation of an electrolyte based on terahertz waves in the third frequency range and estimates the ionic conductivity of the electrolyte by analyzing the frequency dependence of multiple permittivityes. In paragraph 1, The characteristics of the above spectrum are A battery monitoring device comprising at least one of the peak of the spectrum, the peak area of ​​the spectrum, and the major fluctuation component of the spectrum. In paragraph 1, The above processor A battery monitoring device that estimates the electrolyte state through the above machine learning model, corrects the estimated electrolyte state using an LSTM (Long Short-Term Memory) network, assigns reliability to the corrected electrolyte state through fuzzy logic and rule-based inference, and finally outputs the electrolyte state. In paragraph 1, The above processor A battery monitoring device that extracts electrolyte states at multiple points using the above terahertz sensor and extracts the spatial distribution of electrolyte states within the battery by 3D mapping the multiple electrolyte states. A step in which a processor extracts spectral features for each frequency domain of preset terahertz waves received by a terahertz sensor; and A battery monitoring method comprising the step of the processor inputting the characteristics of the spectrum into a pre-learned machine learning model to monitor the state of the electrolyte according to the frequency band through the machine learning model. In paragraph 8, In the step of monitoring the state of the above electrolyte, The above processor is a battery monitoring method that estimates the lithium ion concentration of an electrolyte based on terahertz waves in a first frequency range. In paragraph 8, In the step of monitoring the state of the above electrolyte, The above processor is a battery monitoring method that calculates the solvent composition ratio of an electrolyte based on terahertz waves in a second frequency range. In paragraph 8, In the step of monitoring the state of the above electrolyte, The above processor is a battery monitoring device that detects degradation of the electrolyte based on terahertz waves in the third frequency range and estimates the ionic conductivity of the electrolyte by analyzing the frequency dependence of multiple dielectric constants. In paragraph 8, The characteristics of the above spectrum are A battery monitoring method comprising at least one of the peak of the spectrum, the peak area of ​​the spectrum, and the major fluctuation component of the spectrum. In paragraph 8, In the step of monitoring the state of the above electrolyte, A battery monitoring method in which the processor estimates the electrolyte state through the machine learning model, corrects the estimated electrolyte state using an LSTM network, and finally outputs the electrolyte state by assigning reliability to the corrected electrolyte state through fuzzy logic and rule-based inference. In paragraph 8, In the step of monitoring the state of the above electrolyte, A battery monitoring method in which the processor extracts the electrolyte state at multiple points using the terahertz sensor and extracts the spatial distribution of the electrolyte state within the battery by 3D mapping the multiple electrolyte states.

Images