A lithium ion battery separator spray coating detection system and detection method

Abstract

The application discloses a lithium ion battery diaphragm spraying coating detection system and method, and relates to the technical field of lithium ion batteries. The system comprises a first signal acquisition unit, a second signal acquisition unit, and a processor. The first signal acquisition unit is used for acquiring a first signal from the same target position of a lithium ion battery diaphragm spraying coating, and the first signal is a complex impedance signal representing the overall electrical characteristics of the target position. The second signal acquisition unit is used for synchronously acquiring a second signal from the same target position, and the second signal is a spectrum signal representing the surface state of the target position. The processor is connected with the first acquisition unit and the second acquisition unit. The processor is configured to perform the following steps: determining the surface state characteristic parameter of the target position according to the spectrum signal; correcting the complex impedance signal according to the surface state characteristic parameter to obtain a corrected impedance signal; and determining the internal conductive network distribution characteristics of the lithium ion battery diaphragm spraying coating at the target position based on the corrected impedance signal. The application can effectively detect the electrochemical uniformity of the coating.

Description

Technical Field

[0001] This application relates to the field of lithium-ion battery technology, and in particular to a lithium-ion battery separator spray coating detection system and detection method. Background Technology

[0002] As a core energy storage component in modern electronic devices and electric vehicles, the performance, safety, and cycle life of lithium-ion batteries largely depend on the quality of their internal key materials and the precision of their manufacturing processes. The separator, one of the four core materials of a lithium-ion battery, primarily functions to form a physical barrier between the positive and negative electrodes, preventing internal short circuits while allowing lithium ions to pass freely. To further enhance the overall performance of the battery, such as improving battery safety, electrolyte wettability, and electrode adhesion, a functional coating is typically sprayed onto the surface of a traditional polyolefin (such as PE or PP) separator substrate. This coating is usually a micron-scale thin layer formed after drying from a slurry composed of a polymer binder (such as polyvinylidene fluoride, PVDF) and inorganic particles (such as insulating ceramic particles like alumina and boehmite) or conductive additives (such as carbon nanotubes and graphene).

[0003] In the production process of lithium-ion battery separator coatings, ensuring the consistency and stability of coating quality is crucial. Currently, the industry mainly employs a range of inspection technologies to monitor the macroscopic physical properties of the coating. For example, CCD camera-based optical vision inspection systems are used to identify surface defects in the coating, such as scratches, foreign objects, pinholes, and particle agglomerations; laser thickness gauges or mechanical contact probes are used to measure the coating thickness and assess its uniformity; or X-ray or beta-ray detection technologies are used to calculate the areal density of the coating by measuring the attenuation of rays, thereby indirectly assessing the coating's uniformity.

[0004] However, for functional coatings containing conductive additives, their electrochemical performance depends not only on the uniformity of thickness or the absence of surface defects, but more importantly on the uniformity of the conductive network formed by the conductive additives and insulating polymer binders within the three-dimensional space of the coating. However, during actual spraying and drying processes, factors such as slurry rheology, solvent evaporation rate, and uneven temperature field distribution can easily lead to the agglomeration of conductive particles at the microscale, causing severe heterogeneity in the conductive network distribution within the coating. This results in problems such as excessive local internal resistance, uneven current density distribution, and concentrated heat generation during rapid charging and discharging, ultimately severely impacting the battery's cycle life and safety. The aforementioned testing methods cannot accurately assess the electrochemical uniformity of the coating.

[0005] Therefore, in the field of lithium-ion battery separator coating, there is an urgent need for a way to solve the technical problem that the existing technology cannot accurately assess the electrochemical uniformity of the coating. Summary of the Invention

[0006] This application provides a lithium-ion battery separator coating detection system and method for effectively detecting the electrochemical uniformity of the coating.

[0007] To achieve the above objectives, the embodiments of this application adopt the following technical solutions: In a first aspect, a lithium-ion battery separator coating inspection system is provided, the system comprising: The first signal acquisition unit is used to acquire a first signal from the same target position of the coating sprayed on the lithium-ion battery separator. The first signal is a complex impedance signal that characterizes the overall electrical properties of the target position. The second signal acquisition unit is used to synchronously acquire a second signal from the same target location. The second signal is a spectral signal that characterizes the surface state of the target location. The processor is connected to the first and second acquisition units. The processor is configured to execute: Based on the spectral signal, determine the surface state characteristic parameters of the target location; Based on the surface condition characteristic parameters, the complex impedance signal is corrected to reduce or eliminate the measurement influence caused by the surface condition, thereby obtaining the corrected impedance signal; Based on the corrected impedance signal, the distribution characteristics of the internal conductive network of the lithium-ion battery separator sprayed at the target location are determined.

[0008] In one possible implementation of the first aspect, the first signal acquisition unit includes a multi-frequency impedance analyzer and a microelectrode probe array, and the processor is configured to drive the microelectrode probe array through the multi-frequency impedance analyzer to acquire complex impedance signals at multiple frequencies.

[0009] In another possible implementation of the first aspect, the second acquisition unit includes a near-infrared spectrometer and an optical probe coaxially arranged with the microelectrode probe array, and the processor is configured to drive the optical probe through the near-infrared spectrometer to acquire the spectral signal in the near-infrared band at the target location.

[0010] In another possible implementation of the first aspect, the processor determines surface state characteristic parameters of the target location based on the spectral signal, including: The processor inputs the spectral signal into a pre-trained chemometrics model; The processor uses a chemometric model to reduce the dimensionality of the high-dimensional spectral signal and outputs one or more low-dimensional feature parameters that characterize the surface state, which are then used as surface state feature parameters.

[0011] In another possible implementation of the first aspect, the chemometric model is a principal component analysis model; the processor is configured to specifically execute: The spectral signal is projected onto the principal component space of the principal component analysis model to obtain its score on a preset number of principal components. One or more scores are used as surface state feature parameters.

[0012] In another possible implementation of the first aspect, the processor corrects the complex impedance signal based on surface state characteristic parameters, including: Substitute the surface state characteristic parameters into a preset correction model to determine a complex impedance offset caused by the surface state. The corrected impedance signal is obtained by subtracting the complex impedance offset from the complex impedance signal.

[0013] In another possible implementation of the first aspect, the processor determines the internal conductive network distribution characteristics of the coating at the target location based on the corrected impedance signal, including: Extract the impedance characteristic value at the preset frequency from the corrected impedance signal; Based on the spatial variation of impedance characteristic values ​​obtained at multiple spatially adjacent target locations, an index for characterizing the distribution characteristics of the internal conductive network is calculated.

[0014] In another possible implementation of the first aspect, the impedance characteristic value is the imaginary part of the corrected impedance signal in the high-frequency region.

[0015] Secondly, this application provides a method for detecting the coating of a lithium-ion battery separator, the method comprising: A first signal is acquired from the same target location of the coating sprayed onto the lithium-ion battery separator. The first signal is a complex impedance signal that characterizes the overall electrical properties of the target location. A second signal is simultaneously acquired from the same target location. The second signal is a spectral signal that characterizes the surface state at the target location. Based on the spectral signal, determine the surface state characteristic parameters of the target location; Based on the surface condition characteristic parameters, the complex impedance signal is corrected to reduce or eliminate the measurement influence caused by the surface condition, thereby obtaining a corrected impedance signal. Based on the corrected impedance signal, the distribution characteristics of the internal conductive network of the lithium-ion battery separator coating at the target location are determined.

[0016] In one possible implementation of the second aspect, determining the surface state characteristic parameters of the target location based on the spectral signal includes: The spectral signal is input into a pre-trained chemometric model to reduce the dimensionality of the spectral signal and obtain one or more principal component scores. One or more principal component scores are used as surface state feature parameters.

[0017] By simultaneously acquiring complex impedance and spectral signals from the same target location and using the spectral signals to intelligently correct the complex impedance signals to eliminate interference from surface conditions, the above technical solution can accurately obtain electrical information reflecting the true distribution characteristics of the conductive network inside the lithium-ion battery separator coating. This effectively solves the technical problem that existing detection technologies can only assess the macroscopic physical properties of the coating but cannot accurately assess electrochemical uniformity. It achieves precise detection of the microscopic distribution of conductive additives and the connectivity of the conductive network inside the coating, providing a reliable technical means for timely detection and prevention of non-uniformity in the distribution of the conductive network inside the coating caused by the agglomeration of conductive particles. This effectively avoids defects such as excessive local internal resistance, uneven current density distribution, and concentrated heat generation during rapid charging and discharging in the finished battery, significantly improving the quality control level of the lithium-ion battery separator coating and the cycle life and safety performance of the battery products.

[0018] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0019] Figure 1 A structural block diagram of a lithium-ion battery separator coating detection system provided in this application embodiment; Figure 2 This is a schematic diagram of the distribution of surface state characteristic parameters provided in an embodiment of this application; Figure 3 This is a schematic flowchart of a lithium-ion battery separator coating detection method provided in an embodiment of this application. Detailed Implementation

[0020] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0021] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0022] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0023] Figure 1 This diagram schematically illustrates a structural block diagram of a lithium-ion battery separator coating inspection system according to an embodiment of this application. Figure 1 As shown in the figure, this application provides a lithium-ion battery separator coating inspection system, which may include: The first signal acquisition unit is used to acquire a first signal from the same target position of the coating sprayed on the lithium-ion battery separator. The first signal is a complex impedance signal that characterizes the overall electrical properties of the target position. The second signal acquisition unit is used to synchronously acquire a second signal from the same target location. The second signal is a spectral signal that characterizes the surface state of the target location. The processor is connected to the first and second acquisition units. The processor is configured to execute: Based on the spectral signal, determine the surface state characteristic parameters of the target location; Based on the surface condition characteristic parameters, the complex impedance signal is corrected to reduce or eliminate the measurement influence caused by the surface condition, thereby obtaining the corrected impedance signal; Based on the corrected impedance signal, the distribution characteristics of the internal conductive network of the lithium-ion battery separator sprayed at the target location are determined.

[0024] The first signal acquisition unit drives a microelectrode probe array via a multi-frequency impedance analyzer to acquire complex impedance signals at the target location of the coating on the lithium-ion battery separator. Specifically, the microelectrode probe array comprises multiple metal probes arranged in a specific geometric pattern to contact the coating surface, forming an electrical measurement circuit. The multi-frequency impedance analyzer applies AC voltage signals of different frequencies to the probe array, typically covering a frequency range of 1 Hz to 1 MHz, to obtain the electrical response characteristics of the coating over a wide frequency range.

[0025] Complex impedance signals consist of two components: a real component and an imaginary component. The real component reflects the resistive properties of the coating, while the imaginary component reflects its capacitive properties. Together, they describe the overall electrical characteristics of the target location. In actual measurements, the contact pressure between the probe and the coating surface is typically maintained within a few millinewtons to ensure good electrical contact while avoiding mechanical damage to the coating. By scanning measurements at multiple frequency points, a complete impedance spectrum can be obtained. The low-frequency band primarily reflects the ion conduction characteristics of the coating, while the high-frequency band primarily reflects its electronic conduction characteristics and geometric capacitance effect. This multi-frequency measurement method can comprehensively characterize the complex electrical behavior of the conductive network within the coating, providing a rich information foundation for subsequent data analysis and effectively capturing subtle changes in the coating's electrochemical performance.

[0026] The second signal acquisition unit uses a near-infrared spectrometer to drive an optical probe coaxially positioned with the microelectrode probe array, synchronously acquiring spectral signals at the target location. The coaxial design of the optical and electrical probes ensures complete overlap between the optical and electrical measurement points, eliminating the influence of spatial deviations on the measurement results. The near-infrared spectrometer emits near-infrared light in the wavelength range of 800-2500 nanometers, a band that exhibits spectral response characteristics to organic polymer binders, conductive additives, and inorganic particles. The optical probe transmits the near-infrared light to the coating surface via an optical fiber bundle, with the irradiated area typically controlled within a few square millimeters, consistent with the electrical measurement area.

[0027] The light signal reflected from the coating surface is collected by an optical probe and transmitted back to a near-infrared spectrometer for analysis, yielding reflectance spectral data. The spectral signal reflects surface condition information such as the chemical composition, particle distribution, surface roughness, and material crystallinity of the coating surface. By analyzing the intensity and position of absorption peaks at specific wavelengths, the molecular chain arrangement of the polymer binder, the dispersion of conductive additives, and the aggregation state of inorganic particles can be quantitatively assessed. The synchronous acquisition mechanism ensures complete temporal and spatial correspondence between the spectral signal and the complex impedance signal, providing an accurate reference for subsequent signal correction processing and effectively eliminating systematic errors during the measurement process.

[0028] After receiving the spectral signal, the processor determines the surface state characteristic parameters of the target location using a pre-trained chemometric model. The chemometric model employs principal component analysis (PCA), an algorithm capable of effectively processing high-dimensional spectral data and extracting key information. In practice, the raw spectral data, containing hundreds of wavelengths, is first preprocessed, including baseline correction, smoothing filtering, and standardization, to eliminate the effects of instrument drift and environmental noise. The preprocessed spectral data is then input into the chemometric model, which projects the high-dimensional spectral space to a low-dimensional principal component space through mathematical transformations. The training process of the chemometric model uses a large amount of standard sample spectral data with known surface states, and statistical analysis is used to determine the principal component directions that best explain the variance of the spectral data. In practical applications, the first 3-5 principal components are typically selected as surface state characteristic parameters; these principal components can explain more than 90% of the information in the raw spectral data. Each principal component score reflects a specific characteristic of the coating surface; for example, the first principal component may primarily reflect polymer content, the second principal component may reflect the dispersion of conductive additives, and the third principal component may reflect surface roughness, etc. This dimensionality reduction process transforms complex spectral information into a few easily processed numerical parameters, greatly simplifying the subsequent data processing flow while preserving the core information related to the surface state in the spectral signal.

[0029] The processor uses the obtained surface state characteristic parameters to correct the complex impedance signal, eliminating the interference of surface state on electrical measurements. The correction process is based on a preset correction model, which establishes a quantitative relationship between surface state characteristic parameters and complex impedance measurement offset. Establishing the correction model requires statistical analysis of a large amount of experimental data, using standard samples with known internal conductive network distribution characteristics, and performing complex impedance measurements under different surface state conditions to establish a mapping relationship between surface state parameters and measurement deviation. The correction model typically employs mathematical methods such as multiple linear regression or neural networks, which can accurately predict the complex impedance offset caused by surface factors such as surface roughness, chemical composition changes, and particle aggregation. Specifically, in the correction process, the surface state characteristic parameters at the current measurement position are substituted into the correction model to calculate the corresponding complex impedance offset, which includes both real and imaginary offset components. Then, the corresponding offset is subtracted from the original complex impedance signal to obtain the corrected impedance signal. The correction formula can be expressed as: ; Where Z is the corrected impedance, Z m To measure impedance, For offset function, These are surface state characteristic parameters. This intelligent correction mechanism effectively eliminates the interference of surface state changes on the detection of the internal conductive network, significantly improving the accuracy and reliability of the measurement results.

[0030] Based on the corrected impedance signal, the processor determines the distribution characteristics of the internal conductive network of the coating at the target location through frequency domain analysis and spatial statistical methods. First, impedance characteristic values ​​at specific frequencies are extracted from the corrected complex impedance signal. Typically, the imaginary part of the impedance in the high-frequency region (e.g., 100kHz-1MHz) is chosen as the key characteristic parameter because the impedance response in this frequency band is mainly determined by the conductive network structure within the coating and is less affected by other factors such as ion conduction. The magnitude of the imaginary part of the impedance directly reflects the connectivity of the local conductive network; a smaller value indicates a more complete conductive network, while a larger value indicates defects or breaks in the conductive network. Next, the above measurement and processing process is repeated at multiple spatially adjacent target locations to obtain a series of spatially distributed impedance characteristic values. These data are used to calculate conductive network distribution uniformity indices, such as standard deviation, coefficient of variation, and spatial autocorrelation coefficient. The standard deviation reflects the dispersion of the impedance characteristic values; a smaller value indicates a more uniform conductive network distribution. The spatial autocorrelation coefficient reflects the correlation between impedance characteristic values ​​at adjacent locations and can identify the spatial patterns and clustering characteristics of the conductive network distribution.

[0031] By establishing two-dimensional or three-dimensional spatial distribution maps of impedance characteristic values, the distribution state of the conductive network inside the coating can be visually displayed, accurately identifying defect areas such as conductive particle agglomeration and network breakage. This analysis method based on corrected impedance signals can deeply reveal the true state of the coating's internal microstructure, providing a scientific basis for coating quality assessment and process optimization.

[0032] Figure 2 This illustration shows a schematic diagram of the distribution of surface state characteristic parameters provided in an embodiment of this application, such as... Figure 2 As shown, the spatial distribution of principal component scores is illustrated, with different colors representing different surface state types. Complex high-dimensional spectral information is reduced to three key feature parameters (PC1, PC2, PC3) through principal component analysis, representing surface state characteristics such as polymer content, conductive additive dispersion, and surface roughness, respectively. The different colored regions intuitively reflect the spatial variability of the coating surface's chemical composition and physical state: blue areas indicate locations with high polymer content, green areas indicate areas with good conductive additive dispersion, and yellow areas identify locations with abnormal surface roughness. This visualization method allows for the rapid identification of non-uniform distribution patterns of coating surface states, providing accurate surface state parameter inputs for subsequent complex impedance signal correction, and also offering important feedback information for optimizing the coating preparation process.

[0033] This embodiment achieves accurate detection and comprehensive evaluation of the electrochemical uniformity of lithium-ion battery separator coatings through the complete detection process described above. The dual-signal synchronous acquisition mechanism ensures the spatiotemporal consistency of electrical and optical measurements, eliminating the limitations and measurement errors of traditional single-detection methods. The application of a chemometric model transforms complex spectral information into quantifiable surface state parameters, providing a scientific basis for subsequent signal correction. The intelligent correction algorithm effectively eliminates the interference of surface state on electrical measurements, significantly improving the accuracy of internal conductive network detection. Based on the spatial statistical analysis method of the corrected impedance signal, the uniformity of the internal conductive network distribution of the coating can be quantitatively assessed, allowing for timely detection of microscopic defects and quality issues. This entire technical solution not only solves the technical challenge of accurately evaluating the electrochemical uniformity of coatings using existing detection technologies but also provides advanced technical means for quality control in lithium-ion battery separator manufacturing processes, effectively improving the performance stability and safety reliability of battery products and promoting the quality improvement of lithium-ion batteries.

[0034] In one embodiment of this invention, the first signal acquisition unit includes a multi-frequency impedance analyzer and a microelectrode probe array. The processor is configured to drive the microelectrode probe array through the multi-frequency impedance analyzer to acquire complex impedance signals at multiple frequencies.

[0035] As the core device of the first signal acquisition unit, the multi-frequency impedance analyzer employs direct digital frequency synthesis technology to generate high-precision AC excitation signals. Specifically, the multi-frequency impedance analyzer integrates a high-speed digital signal processor and a 16-bit high-precision digital-to-analog converter, enabling it to generate sinusoidal excitation signals with stable amplitude and accurate phase. The amplitude of the excitation signal is typically set to a small range of 10-100mV to ensure the measurement process is within the linear response region, avoiding electrochemical or thermal damage to the coating material. The impedance analyzer is also equipped with a high-input-impedance differential amplifier and a phase-sensitive detection circuit, enabling accurate measurement of the amplitude and phase difference of current and voltage signals. Through a built-in digital filtering algorithm, the influence of environmental electromagnetic interference and thermal noise is effectively suppressed, ensuring measurement stability in industrial production environments.

[0036] The microelectrode probe array is manufactured using precision micromachining technology and comprises multiple passivated metal probes made of chemically inert platinum or gold alloys. This ensures good electrical contact with the coating material while avoiding interference from electrochemical reactions. The probe array is arranged in a 2×2 or 3×3 rectangular configuration, with the probe spacing precisely controlled within the range of 0.5-2 mm to match the microstructural features of the coating material. Each probe is equipped with an independent elastic support mechanism that adapts to the micro-undulations of the coating surface, ensuring uniform and stable contact pressure. In practice, the probe array makes perpendicular contact with the coating surface using a precision positioning system, with the contact pressure controlled within the range of 1-5 millinewtons, ensuring good electrical contact while avoiding mechanical damage to the coating. The geometric configuration of the probe array allows for a four-terminal measurement method, with two probes applied to apply the excitation current and the other two probes used to measure the voltage response, effectively eliminating the influence of lead resistance and contact resistance on the measurement results. Each probe is connected to a low-noise signal conditioning circuit, including an impedance matching network and common-mode noise suppression circuitry, ensuring reliable transmission of weak signals. The precise design of the microelectrode probe array enables high spatial resolution measurement of the electrical properties of local areas of the coating, providing key hardware support for the fine detection of conductive network distribution.

[0037] The processor drives a multi-frequency impedance analyzer through a dedicated instrument control interface, enabling precise setting of measurement parameters and automated control of the measurement process. The processor first configures the impedance analyzer's operating parameters according to a preset measurement scheme, including key parameters such as frequency scan range, number of frequency points, excitation amplitude, and measurement accuracy. A typical frequency scan scheme uses a logarithmically evenly spaced distribution, setting 50-100 frequency points within the 1Hz-1MHz range to ensure sufficient data density across different frequency bands. The processor sends control commands to the impedance analyzer via a digital communication protocol to initiate automatic frequency scan measurement. During the measurement process, the processor monitors the measurement status in real time, including indicators such as signal quality, measurement stability, and data validity. When an anomaly is detected, the measurement is automatically paused and error handling is performed. The processor is also responsible for real-time data acquisition and buffering, performing preliminary processing on the raw measurement data obtained from the impedance analyzer, including data format conversion, unit conversion, and outlier removal. This intelligent instrument control method achieves a high degree of automation and standardization in the complex impedance measurement process, significantly improving measurement efficiency and data quality consistency.

[0038] The process of acquiring complex impedance signals at multiple frequencies employs a wideband scanning strategy to fully utilize the different electrical response characteristics of the coating material at different frequencies. Specifically, the frequency scan starts at a low frequency of 1 Hz and gradually increases to a high frequency of 1 MHz using a logarithmic scale. The measurement time at each frequency point is automatically adjusted according to the frequency; a longer integration time is used at low frequencies to improve the signal-to-noise ratio, while a shorter integration time is used at high frequencies to improve measurement speed. At each frequency point, the impedance analyzer applies a sinusoidal excitation signal of the corresponding frequency to the microelectrode probe array, simultaneously measuring the amplitude and phase difference of the current and voltage. The complex impedance value is calculated using Ohm's law. The complex impedance data includes a real part (resistive component) and an imaginary part (reactant component), represented as follows: ,in For frequency-dependent resistors, For frequency-dependent reactance, The angular frequency is used. The complex impedance response at different frequencies reflects the multi-layered electrical characteristics of the coating material: the low-frequency region mainly reflects ion conduction and electrochemical interface processes, the mid-frequency region reflects the dielectric properties of the polymer binder, and the high-frequency region mainly reflects the geometric capacitance effect and electronic conduction characteristics of the conductive network. By obtaining the complete impedance spectrum, the electrical behavior of the coating material can be comprehensively characterized, providing a rich information foundation for subsequent data analysis and conductive network evaluation. The multi-frequency measurement strategy ensures the comprehensive capture of the complex electrical characteristics of the coating, significantly improving the accuracy and reliability of conductive network distribution detection.

[0039] This implementation achieves high-precision, wide-bandwidth acquisition of complex impedance signals from lithium-ion battery separator coatings through a precise combination of a multi-frequency impedance analyzer and a microelectrode probe array, providing a reliable measurement basis for accurately evaluating the electrochemical properties of the coating. The multi-frequency scanning strategy fully utilizes the differentiated characteristics of the material's electrical response at different frequencies, comprehensively revealing the complete electrical behavior spectrum of the coating material from ion conduction to electronic conduction through measurements across the entire frequency range from low to high frequencies. The precise design of the microelectrode probe array enables localized measurements with micron-level spatial resolution, accurately locating the distribution of the conductive network within the coating. Intelligent processor control ensures the automation and standardization of the measurement process, significantly improving measurement efficiency and data quality consistency. The entire technical solution effectively overcomes the limitations of traditional single-frequency measurements in comprehensively characterizing the complex electrical properties of coatings, providing an advanced detection method for the quality control of lithium-ion battery separator coatings and significantly improving the accuracy and reliability of coating electrochemical uniformity assessment.

[0040] In one embodiment of this invention, the second acquisition unit includes a near-infrared spectrometer and an optical probe coaxially arranged with the microelectrode probe array. The processor is configured to drive the optical probe through the near-infrared spectrometer to acquire the spectral signal in the near-infrared band at the target location.

[0041] The near-infrared spectrometer employs Fourier transform near-infrared technology, operating in the 800-2500 nm near-infrared spectral region. It can accurately identify the characteristic absorption peaks of polymer binders, conductive additives, and inorganic particles in coating materials. The spectrometer's core components include a broadband halogen tungsten lamp source, a Michelson interferometer, an InGaAs detector array, and a high-speed data acquisition system. The halogen tungsten lamp source provides stable, continuous spectral output. The Michelson interferometer uses a precise moving mirror system to generate optical path difference changes, achieving interference modulation of the incident light. The positional accuracy of the moving mirrors reaches the nanometer level, ensuring high precision in spectral measurements. The InGaAs detector array exhibits excellent near-infrared response characteristics, with a response time of less than 1 microsecond and a dynamic range exceeding 60 dB, enabling simultaneous detection of light signals across a wide spectral range. The spectrometer is also equipped with a temperature stabilization control system, maintaining the temperature of key optical components within ±0.1℃ to eliminate the influence of temperature variations on spectral measurements. The built-in spectral correction algorithm automatically compensates for instrument response function and wavelength drift, ensuring the accuracy of measurement results. This high-precision spectral measurement capability provides a reliable hardware foundation for accurately acquiring information about the surface condition of the coating, and can sensitively detect subtle changes in the composition and microstructure of the coating material.

[0042] The optical probe employs a specially designed coaxial fiber bundle structure, achieving precise coaxial alignment with the microelectrode probe array and ensuring perfect overlap between optical and electrical measurement points. The core components of the optical probe include an illumination fiber bundle, a collection fiber bundle, and a precision lens assembly. The overall structure is cylindrical, with an outer diameter controlled within the range of 8-12 mm to accommodate integrated installation with the microelectrode probe array. The illumination fiber bundle consists of multiple 50-micron diameter multimode fibers arranged in a ring around the probe's central axis, responsible for uniformly transmitting excitation light from the near-infrared spectrometer to the coating surface. The collection fiber bundle, located at the probe's center, consists of large-core fibers with a diameter of 200 microns, used to collect the light signal reflected from the coating surface and transmit it back to the spectrometer. The precision lens assembly includes a focusing lens and a collimating lens. The focusing lens focuses the illumination beam onto the coating surface, forming a uniform illumination spot approximately 2 mm in diameter, precisely coinciding with the measurement area of ​​the microelectrode probe array. The optical probe and the microelectrode probe array are coaxially fixed using precision mechanical clamps, ensuring spatial consistency between the two measurement methods. This coaxial design effectively eliminates the spatial deviation problem in traditional separate measurements, provides an accurate spatial reference for subsequent signal correction, and significantly improves the reliability and accuracy of measurement results.

[0043] The processor enables precise driving and parameter configuration of the near-infrared spectrometer via a dedicated spectrometer control interface, employing a high-speed serial communication protocol to ensure the real-time performance and reliability of control commands and data transmission. The processor first configures the spectrometer's key operating parameters according to a preset measurement scheme, including integration time, number of scans, spectral resolution, and baseline correction mode. The processor also handles the spectrometer's automatic calibration control, automatically performing dark current correction and white plate reference correction before each measurement to eliminate the effects of detector noise and light source intensity variations. During calibration, the processor controls the spectrometer to first measure the dark current background under shading conditions, and then uses a standard white plate reflective reference for light intensity correction to establish an accurate reflectance standard. This intelligent instrument control method achieves a high degree of automation and standardization in the spectral measurement process, ensuring the consistency and reproducibility of measurement results and providing a high-quality spectral data foundation for subsequent data analysis.

[0044] The near-infrared spectral signal acquisition process employs reflectance spectroscopy measurement mode, fully utilizing the specific response characteristics of near-infrared light to organic polymers and inorganic materials. Specifically, during acquisition, the near-infrared spectrometer emits broadband near-infrared light onto the coating surface via an illumination fiber bundle, with the irradiation power density controlled within a safe range to avoid photothermal damage to the coating material. The reflected light from the coating surface contains the material's absorption and scattering information; different wavelengths of light are absorbed to varying degrees within the material, forming characteristic spectral fingerprints. Polymer binders exhibit characteristic overtone and combination frequency absorption peaks of CH, NH, and OH bonds in the near-infrared region, typically appearing in the 1400-1900 nm and 2000-2400 nm wavelength ranges. Conductive additives such as carbon nanotubes and graphene exhibit broadband absorption characteristics in the near-infrared region, with absorption intensity closely related to their dispersion state and aggregation degree. Inorganic particles such as alumina and boehmite exhibit lattice vibration-related absorption characteristics at specific wavelengths. The collection fiber bundle transmits the reflected light signal back to the spectrometer's detector array, where photoelectric conversion and signal amplification processing yield digital spectral data covering the entire near-infrared band. Spectral data are represented in the form of reflectance or absorbance, with each data point corresponding to a different wavelength. The processor preprocesses the raw spectral data, including spectral smoothing, baseline correction, and normalization, to eliminate the effects of instrument noise and baseline drift. This near-infrared spectroscopy acquisition method can sensitively detect changes in the chemical composition, material distribution, and microstructural characteristics of the coating surface, providing a rich information foundation for accurately assessing the surface condition of the coating.

[0045] This implementation achieves high-precision, non-destructive testing of the surface condition of lithium-ion battery separator coatings through the precise combination of a near-infrared spectrometer and a coaxial optical probe, providing an accurate reference for subsequent complex impedance signal correction. The application of near-infrared spectroscopy fully utilizes the characteristic response of materials in the near-infrared band. By analyzing the absorption intensity and peak position changes at different wavelengths, it is possible to quantitatively assess the polymer content, conductive additive dispersion, inorganic particle distribution, and surface morphology of the coating surface. The coaxial design ensures perfect spatial correspondence between optical and electrical measurements, eliminating the positional deviation problem in traditional separate measurement methods and creating the necessary conditions for accurate signal correction. Intelligent processor control automates and standardizes the spectral measurement process, ensuring high consistency and reliability of measurement results through automatic calibration and real-time quality monitoring. The entire technical solution effectively overcomes the limitations of insufficient sensitivity and low quantification in traditional surface detection methods. By acquiring rich spectral information, it provides a scientific basis for intelligent correction of complex impedance signals, significantly improving the accuracy of detecting the distribution of conductive networks within the coating, and providing an advanced technical means for quality control in lithium-ion battery separator manufacturing processes.

[0046] In one embodiment of this invention, the processor determines surface state characteristic parameters of the target location based on the spectral signal, including: The processor inputs the spectral signal into a pre-trained chemometrics model; The processor uses a chemometric model to reduce the dimensionality of the high-dimensional spectral signal and outputs one or more low-dimensional feature parameters that characterize the surface state, which are then used as surface state feature parameters.

[0047] The processor inputs the acquired near-infrared spectral signals into a pre-trained chemometric model. This model, built upon spectral data from a large number of standard samples and corresponding surface state information, can transform complex spectral information into quantifiable parameters. In practice, the processor first performs standardization preprocessing on the raw spectral data, including spectral range interception, baseline correction, smoothing filtering, and normalization. Spectral range interception typically selects the effective wavelength range of 1000-2500 nm, removing the low signal-to-noise ratio portions at lower and higher wavelengths, while retaining the spectral region containing key material characteristic information. Baseline correction employs polynomial fitting or a rubber band baseline correction algorithm to eliminate baseline shifts caused by instrument drift and environmental factors. The baseline correction formula is: ,in For the corrected spectrum, The original spectrum, This is the baseline function.

[0048] Smoothing filtering employs a Savitzky-Golay filter, typically with a window size of 5-15 data points and a polynomial order of 2-4, effectively reducing random noise while preserving the shape of spectral characteristic peaks. Normalization scales the spectral data to the 0-1 range, eliminating the influence of light intensity differences under varying measurement conditions; this can be achieved using a pre-defined normalization formula. The preprocessed spectral data is input into the chemometrics model as a vector, with the vector dimension corresponding to reflectance or absorbance values ​​at different wavelengths. Upon receiving the input vector, the chemometrics model extracts key information related to surface state based on pre-learned pattern recognition rules, preparing the data for subsequent dimensionality reduction processing and ensuring the input data quality and format meet model requirements.

[0049] The processor effectively reduces the dimensionality of high-dimensional spectral signals using a chemometric model, transforming complex spectral data containing hundreds of wavelengths into several key low-dimensional feature parameters. The chemometric model employs principal component analysis, a statistical method, to identify the main variation patterns in the spectral data and extract the most representative features. By calculating the covariance matrix of the spectral data and performing eigenvalue decomposition, the principal component directions that best explain the data variance are identified. The mathematical expression is: Where C is the covariance matrix, X is the centered spectral data matrix, and n is the number of samples. Eigenvalue decomposition yields the eigenvector matrix P and the eigenvalue diagonal matrix Λ, satisfying... Principal component scores are obtained through projection calculations: The model training process utilizes a large amount of standard sample spectral data with known surface states to establish a quantitative relationship between spectral features and surface state parameters. Training samples cover combinations of different polymer contents, conductive additive dispersion levels, surface roughness, and particle distribution states to ensure the model's generalization ability. After training, the model can project new spectral data into a low-dimensional feature space, outputting 3-8 principal component scores as surface state feature parameters. Each feature parameter has a clear physical meaning; for example, the first principal component may primarily reflect changes in polymer binder content, the second principal component may reflect the uniformity of conductive additive dispersion, and the third principal component may reflect surface morphology features, etc. This dimensionality reduction process significantly simplifies data complexity while retaining the core information most relevant to the surface state, providing accurate and reliable reference parameters for subsequent impedance signal correction.

[0050] This implementation method utilizes a chemometrics model to effectively convert complex spectral signals into concise surface state characteristic parameters, providing a scientific and accurate analytical tool for the quantitative assessment of the surface state of lithium-ion battery separator coatings. Dimensionality reduction, achieved through principal component analysis, successfully extracted the key information most relevant to the surface state from the spectral data, transforming high-dimensional, complex spectral vectors into a few easily understood and processed characteristic parameters, significantly simplifying subsequent data processing. The application of the chemometrics model not only achieves efficient compression of spectral information but, more importantly, establishes a quantitative relationship between spectral features and the surface physicochemical state, elevating surface state assessment from qualitative judgment to precise quantitative analysis. This machine learning-based intelligent analysis method exhibits excellent adaptability and generalization capabilities, handling spectral variations under different material systems and process conditions, providing stable and reliable technical support for coating quality control. It effectively overcomes the limitations of traditional spectral analysis methods, which rely on human experience and have low quantification levels. Through automated feature extraction and parameterized expression, it significantly improves the accuracy, objectivity, and reproducibility of surface state detection, providing a high-quality data foundation for subsequent complex impedance signal correction and conductive network distribution assessment.

[0051] In one embodiment of this invention, the chemometric model is a principal component analysis model; the processor is configured to specifically execute: The spectral signal is projected onto the principal component space of the principal component analysis model to obtain its score on a preset number of principal components. One or more scores are used as surface state feature parameters.

[0052] The processor projects the preprocessed spectral signal onto the principal component space of a pre-trained principal component analysis (PCA) model, obtaining the projection scores of the spectral data along each principal component direction through linear transformation. The core of the PCA model is a set of orthogonal principal component vectors, obtained through covariance analysis of a large amount of training spectral data. Each principal component vector represents a major variation pattern in the spectral data. Specifically, during the projection process, the processor first centers the currently measured spectral data vector, i.e., subtracting the mean of the training data for the corresponding wavelength from the value at each wavelength. The centering formula is as follows: ,in The original spectral vector, Let be the mean vector of the training data. Then, multiply the centered spectral vector by the principal component loading matrix to obtain the principal component score vector, calculated using the following formula: , where t is the principal component score vector, and P is the principal component loading matrix, with each column of the matrix representing a principal component direction. The principal component loading matrix typically has an n×k dimension, where n is the number of wavelength points in the spectral data (usually 500-1500), and k is the number of principal components selected (usually 3-8).

[0053] The projection calculation process employs a matrix operation algorithm, with the processor utilizing parallel computing capabilities to simultaneously calculate the scores of multiple principal components. Each principal component score reflects the projection intensity of the current spectrum along the corresponding principal component direction; the magnitude and sign of the value represent the degree and direction of deviation of the spectrum from the average state of the training dataset. Through this projection transformation, high-dimensional spectral data containing hundreds of wavelength points is effectively compressed into a few key principal component scores. This achieves data dimensionality reduction while preserving the core information in the spectrum most relevant to the surface state, providing a concise and accurate data foundation for subsequent surface state assessment.

[0054] The processor selects one or more of the most representative scores from all calculated principal component scores as surface state characteristic parameters, based on a preset principal component selection strategy. Principal component selection is typically based on the cumulative variance contribution rate criterion, that is, selecting the top few principal components that can explain 85%-95% of the variance of the original spectral data.

[0055] In practice, the processor first analyzes the variance contribution rate of each principal component. The formula for calculating the variance contribution rate is as follows: ; in Let be the variance contribution rate of the i-th principal component. Let be the eigenvalue corresponding to the i-th principal component, and n be the total number of principal components. Typically, the first principal component has the highest variance contribution rate, mainly reflecting the most significant variation pattern in the spectral data; for lithium-ion battery separator coatings, this is usually related to changes in the content of the polymer binder. The second principal component typically has a variance contribution rate of 15%-25%, mainly reflecting the dispersion state and aggregation degree of conductive additives. The third principal component typically has a variance contribution rate of 8%-15%, possibly related to surface roughness or the distribution state of inorganic particles. The processor automatically determines the number of principal components to retain based on the principle that the cumulative variance contribution rate reaches a set threshold. For example, if the cumulative variance contribution rate of the first three principal components reaches 90% or more, the scores of the first three principal components are selected as surface state feature parameters. Each selected principal component score is assigned a clear physical meaning label, such as PC1 representing a polymer content-related parameter, PC2 representing a conductive additive dispersion parameter, etc. The numerical range of these feature parameters is typically between -3 and +3; positive values ​​indicate a positive deviation from the average state of the training data, and negative values ​​indicate a negative deviation; the absolute value reflects the degree of deviation. This principal component selection strategy ensures that the surface state characteristic parameters contain both the main information of the spectral data and have clear physicochemical significance, providing a reliable reference for subsequent impedance signal correction.

[0056] This implementation successfully achieves intelligent transformation from complex high-dimensional spectral data to concise low-dimensional feature parameters through the application of principal component analysis (PCA) models, providing a scientifically reliable technical means for quantitative analysis of the surface state of lithium-ion battery separator coatings. The projection calculation process, through precise linear transformation, completely preserves the distribution characteristics of spectral data in the principal component space, ensuring no loss of key information during dimensionality reduction, while significantly simplifying data complexity and the computational burden of subsequent processing. The principal component selection strategy, based on statistical principles, ensures the representativeness and reliability of the selected feature parameters through variance contribution rate analysis, avoiding bias from subjective human judgment and improving the objectivity and consistency of feature extraction. Each principal component score has a clear physicochemical meaning, intuitively reflecting the specific state characteristics of the coating surface, such as changes in material composition, dispersion uniformity, and surface morphology, providing operable quantitative indicators for process quality control. The entire technical solution effectively solves the technical problems of complex data processing and difficult feature extraction in traditional spectral analysis methods, significantly improving the efficiency and accuracy of surface state detection through automated dimensionality reduction and feature parameterization. The application of principal component analysis not only achieves efficient compression and feature extraction of spectral information, but more importantly, it establishes a quantitative relationship between spectral features and surface physical state, providing an accurate and reliable reference for subsequent complex impedance signal correction, thereby improving the performance and reliability of the entire detection system.

[0057] In one embodiment of this invention, the processor corrects the complex impedance signal based on surface state characteristic parameters, including: Substitute the surface state characteristic parameters into a preset correction model to determine a complex impedance offset caused by the surface state. The corrected impedance signal is obtained by subtracting the complex impedance offset from the complex impedance signal.

[0058] The processor inputs the surface state characteristic parameters obtained from principal component analysis into a pre-established calibration model, and determines the complex impedance measurement offset caused by surface state factors through mathematical modeling methods. The calibration model is built using multiple linear regression or support vector regression algorithms. The model training process uses a large number of standard samples with known internal conductive network distribution characteristics, performing complex impedance measurements under different surface state conditions to establish a quantitative relationship between surface state characteristic parameters and impedance measurement deviation. Specifically, the mathematical expression of the calibration model is: ,in This is the offset of the complex impedance. Scores for the n principal components. For a multiple linear regression model, the specific form is: ,in The regression coefficients are determined from the training data using the least squares method. The complex impedance offset comprises two components: a real offset and an imaginary offset. These correspond to resistance offset and reactance offset, respectively. During model training, hundreds of standard samples with different surface conditions were used. Complex impedance measurements were performed on each sample at multiple frequency points, and the corresponding surface condition characteristic parameters were recorded. Statistical analysis was used to establish the mapping relationship between the characteristic parameters and the impedance offset.

[0059] In practical applications, the processor substitutes the principal component score of the current measurement position into the calibration model and quickly calculates the corresponding complex impedance offset through matrix operations. The calibration model also considers the variation characteristics of the offset at different frequencies, establishing an independent calibration relationship for each frequency point to ensure calibration accuracy across the entire frequency spectrum. This machine learning-based calibration model can accurately predict complex impedance measurement deviations caused by surface factors such as surface roughness, changes in chemical composition, and particle aggregation, providing a scientifically reliable technical basis for eliminating surface state interference.

[0060] The processor subtracts the calculated complex impedance offset from the original complex impedance measurement signal through mathematical operations, achieving precise elimination of surface condition interference and obtaining a corrected impedance signal that truly reflects the characteristics of the conductive network inside the coating. The correction calculation process uses complex number operations to ensure simultaneous processing of real and imaginary components. The correction formula is as follows: When unfolded, it becomes That is, the real part is corrected to Imaginary part correction is The processor performs correction processing on the complex impedance data at each frequency point. Since the offset characteristics differ at different frequencies, the correction model provides a corresponding offset value for each frequency point. During the correction process, the processor automatically initiates a re-measurement procedure when it detects abnormal correction results. The corrected impedance signal maintains the original frequency characteristics and phase relationship, but eliminates systematic deviations caused by changes in surface condition. To verify the correction effect, the processor calculates the statistical characteristics of the impedance data before and after correction, including changes in mean, standard deviation, and spectral characteristics. The corrected impedance signal exhibits better reproducibility and consistency. The difference in corrected impedance values ​​measured under different surface conditions for samples with the same internal conductive network state is typically less than 3%, significantly better than the 10%-20% difference level before correction. The correction process also generates a correction report, recording the changes in key parameters before and after correction, providing detailed technical documentation for subsequent data analysis and quality traceability. This precise mathematical correction method effectively eliminates the systematic interference of surface condition on complex impedance measurement, ensuring the accuracy and reliability of the corrected impedance signal and creating the necessary conditions for accurately evaluating the distribution characteristics of the conductive network inside the coating.

[0061] This implementation method successfully eliminates surface state interference through the synergistic application of a preset calibration model and complex impedance signal calibration, significantly improving the accuracy and reliability of detecting the conductive network inside the coating of lithium-ion battery separators. The calibration model is established based on statistical analysis of a large amount of experimental data. A precise quantitative relationship between surface state characteristic parameters and complex impedance offset is established using machine learning algorithms, providing a predictable calibration basis for measurement deviations under different surface state conditions. The model's high fitting degree and low prediction error ensure the scientific rigor and accuracy of the calibration process, avoiding the subjectivity and uncertainty of traditional empirical calibration methods. The complex impedance signal calibration process, through precise complex number operations, simultaneously processes the real and imaginary parts of the impedance, preserving the frequency domain characteristics and phase information of the original signal, ensuring the integrity and correct physical meaning of the calibrated data. The calibrated impedance signal effectively eliminates measurement deviations caused by surface factors such as surface roughness, chemical composition changes, and uneven particle distribution, enabling the impedance measurement results to truly reflect the distribution state of the conductive network inside the coating. This invention solves the technical problems of traditional impedance measurement methods being susceptible to surface condition interference and having poor measurement result consistency. Through intelligent deviation identification and precise mathematical correction, it achieves accurate detection of the internal electrochemical characteristics of the coating, providing an advanced and reliable technical means for quality control of lithium-ion battery separator manufacturing process, and significantly improving the stability and consistency of product quality.

[0062] In one embodiment of this invention, the processor determines the internal conductive network distribution characteristics of the coating at the target location based on the corrected impedance signal, including: Extract the impedance characteristic value at the preset frequency from the corrected impedance signal; Based on the spatial variation of impedance characteristic values ​​obtained at multiple spatially adjacent target locations, an index for characterizing the distribution characteristics of the internal conductive network is calculated.

[0063] The processor extracts impedance characteristic values ​​at a preset frequency from the corrected complex impedance signal and identifies the key frequency points and corresponding impedance parameters that best reflect the characteristics of the conductive network inside the coating through frequency domain analysis. The selection of the preset frequency is based on the electrochemical characteristic analysis of the lithium-ion battery separator coating material. The high-frequency region (100kHz-1MHz) is usually selected as the target frequency band for feature extraction because the impedance response in this frequency band is mainly determined by the geometry and connection state of the conductive network inside the coating and is less affected by other factors such as ion conduction and electrochemical reactions.

[0064] In practice, the processor first performs frequency domain analysis on the corrected complex impedance spectrum to identify the variation patterns of key parameters such as impedance magnitude |Z|, real part R, imaginary part X, and phase angle θ at different frequencies. Impedance eigenvalues ​​are extracted using several methods: first, directly extracting the impedance value at a specific frequency point, such as the imaginary part X (100kHz) at 100kHz, which sensitively reflects the connectivity of the conductive network; second, extracting characteristic parameters of the impedance spectrum, such as the impedance slope, corner frequency, and relaxation time constant in the high-frequency region; and third, extracting circuit element parameters, such as parallel resistance and capacitance, through equivalent circuit fitting. The processor uses digital signal processing algorithms for eigenvalue calculation, including Fast Fourier Transform (FFT), digital filtering, and interpolation fitting techniques. The calculation formula for the imaginary part of the impedance is as follows: Where f is the preset frequency and Im represents the imaginary part operation. The processor also calculates statistical parameters of the impedance characteristics, such as mean, standard deviation, and coefficient of variation, to evaluate the quality and stability of the measurement data.

[0065] The processor calculates quantitative indicators characterizing the distribution properties of the internal conductive network based on impedance characteristic values ​​obtained at multiple spatially adjacent target locations using spatial statistical analysis methods. The spatial sampling strategy employs a gridded sampling method, performing multi-point measurements on the coating surface according to a regular rectangular or triangular grid. After collecting the impedance characteristic values ​​from each sampling point, the processor constructs a two-dimensional spatial distribution matrix, with matrix elements... This represents the impedance characteristic value at position (i,j). Based on this spatial distribution matrix, the processor calculates various statistical indicators to characterize the distribution characteristics of the conductive network: one is the uniformity indicator, including standard deviation. and coefficient of variation Where N is the number of sampling points, The first is the mean, and the smaller the CV value, the more uniform the distribution; the second is the spatial autocorrelation index, which is calculated using the following formula: ,in The system employs a spatial weighting matrix, where an I value close to 0 indicates random distribution, positive values ​​represent positive correlation clustering, and negative values ​​represent negative correlation dispersion. Thirdly, it uses local clustering indicators to identify high- and low-value clustering areas and pinpoint defects in the conductive network. Fourthly, it uses spatial gradient indicators to calculate the impedance change rate between adjacent points and identify discontinuous regions in the conductive network. The processor also constructs a spatial interpolation surface for impedance eigenvalues, using Kriging interpolation or radial basis function interpolation to generate continuous two-dimensional or three-dimensional distribution maps, visually displaying the spatial distribution pattern of the conductive network. These quantitative indicators provide objective and comparable numerical standards for coating quality assessment, accurately identifying the uniformity, continuity, and integrity of the conductive network distribution.

[0066] This implementation method successfully achieves a precise quantitative assessment of the conductive network distribution characteristics within the coating of lithium-ion battery separators by organically combining impedance eigenvalue extraction and spatial statistical analysis, providing a scientific and reliable technical means for coating quality control. The eigenvalue extraction process accurately identifies the key frequencies and impedance parameters that best reflect the conductive network characteristics through frequency domain analysis, avoiding interference from other electrochemical processes and ensuring a direct correspondence between eigenvalue parameters and the conductive network state. The scientific selection of preset frequencies and the comprehensive application of multiple eigenvalue extraction methods enable the detection results to comprehensively reflect the connectivity, uniformity of distribution, and structural integrity of the conductive network. The application of spatial statistical analysis methods achieves a technological leap from single-point measurement to regional assessment, revealing the distribution patterns and variation characteristics of the conductive network in two-dimensional space through multi-point collaborative analysis. The calculation of multi-dimensional statistical parameters such as uniformity indices, spatial autocorrelation indices, and local aggregation indices provides a comprehensive and in-depth quantitative description of the conductive network distribution characteristics, elevating quality assessment from qualitative judgment to a precise numerical level. The application of spatial interpolation and visualization technologies allows for the intuitive presentation of complex conductive network distribution states, facilitating technicians to quickly identify problem areas and optimize process parameters. It effectively solves the technical problem that traditional testing methods cannot accurately assess the uniformity of the conductive network distribution inside the coating. Through precise feature extraction and scientific spatial analysis, it provides strong technical support for quality control and product performance optimization in lithium-ion battery separator manufacturing processes, and significantly improves the consistency and reliability of battery products.

[0067] In one embodiment of this example, the impedance characteristic value is the imaginary part of the corrected impedance signal in the high-frequency region.

[0068] The impedance eigenvalue is selected as the imaginary part of the corrected impedance signal in the high-frequency region. This selection is based on the electrochemical response mechanism and physical characteristics of the conductive network of the lithium-ion battery separator coating material under high-frequency conditions. Under high-frequency AC excitation, the impedance response of the coating material is mainly determined by its geometric capacitance effect and electronic conduction characteristics, while the influence of low-frequency phenomena such as ion conduction and electrochemical interface processes is significantly suppressed. Specifically, the imaginary part of the impedance... It mainly reflects the capacitance characteristics of the material, where ω is the angular frequency and C is the equivalent capacitance.

[0069] For coatings containing conductive additives, the equivalent capacitance is composed of both geometric capacitance and the distributed capacitance of the conductive network. When the conductive network has good connectivity, the number of conductive paths increases, the equivalent capacitance increases, and the absolute value of the imaginary part decreases. Conversely, when the conductive network is broken or unevenly distributed, the equivalent capacitance decreases, and the absolute value of the imaginary part increases. This correspondence makes the high-frequency imaginary part an ideal indicator for characterizing the distribution state of the conductive network. Measurements under high-frequency conditions also have the advantages of fast response speed and low susceptibility to environmental interference, enabling rapid and accurate detection of the conductive network state. Furthermore, the high-frequency imaginary part is less sensitive to temperature changes, and the measurement results exhibit good stability and reproducibility, making it suitable for online detection applications in industrial production environments. This eigenvalue selection method based on physical mechanisms ensures a direct correspondence between the detection results and the true state of the conductive network inside the coating, providing a reliable technical basis for accurately evaluating the electrochemical performance of coatings.

[0070] The frequency range of the high-frequency region is defined using a scientific method based on the electrochemical impedance spectroscopy characteristics of the material. The frequency range is divided by analyzing the impedance response characteristics of the coating material at different frequencies. Specifically, the processor first performs a broadband impedance spectrum scan on the coating material, ranging from 1 Hz to 10 MHz, to obtain complete impedance spectrum data. Through Nyquist and Bode plot analysis, the dominant mechanisms of impedance response in different frequency domains are identified: the low-frequency region (1 Hz-1 kHz) mainly reflects ion conduction and electrochemical interface processes; the mid-frequency region (1 kHz-100 kHz) reflects the dielectric relaxation process of the polymer matrix; and the high-frequency region (100 kHz-1 MHz) mainly reflects the geometric capacitance effect of the conductive network. The lower limit frequency of the high-frequency region is determined by identifying the turning point where the impedance phase angle begins to stabilize, typically within the range of 50-100 kHz, at which point the influence of ion conduction processes is largely eliminated. The upper limit frequency is determined considering the measurement accuracy of the instrument and the influence of parasitic parameters, and is usually set at 1-2 MHz. Above this frequency, intrinsic effects such as lead inductance and parasitic capacitance begin to appear. The processor employs an adaptive frequency selection algorithm that automatically adjusts the high-frequency range based on the specific electrochemical characteristics of the material. The algorithm is based on the stability criterion of the impedance phase angle. Where θ is the phase angle and f is the frequency. A preset threshold is used. The specific range of the high-frequency region may vary for different material systems. The processor establishes a mapping relationship between material type and the optimal frequency range through machine learning algorithms, achieving intelligent frequency selection. This scientific method of determining the frequency range ensures the accuracy and reliability of the imaginary part extraction, avoids interference from other electrochemical processes on the detection of the conductive network, and improves the correlation between eigenvalues ​​and the state of the conductive network.

[0071] The extraction and calculation of the imaginary part employs digital signal processing techniques to ensure accurate separation and quantitative analysis of the imaginary component from the complex impedance signal. The processor first identifies high-frequency data points from the corrected complex impedance spectrum data, typically containing 20-50 frequency points, ensuring sufficient data density for accurate analysis. The complex impedance data is then processed using... The value is stored in the form of R, where R is the real part, X is the imaginary part, and j is the imaginary unit. The imaginary part is extracted using complex number operations: ,in Let f be the imaginary part value at frequency f, and Im represent the imaginary part extraction operation. Considering that the imaginary part value may vary with frequency in the high-frequency region, the processor employs several strategies to extract representative imaginary part values: one is to directly extract the imaginary part value at a specific frequency point, such as the imaginary part value at 100kHz or 500kHz; the other is to calculate the average value of the imaginary part values ​​in the high-frequency region. The algorithm employs a multiplication table, where n represents the number of data points in the high-frequency region. Thirdly, a weighted average method is used, assigning different weights based on the importance of each frequency point. Fourthly, feature parameters such as slope and intercept are extracted from the imaginary part of the data through curve fitting. The processor also performs sign processing on the imaginary part; since the imaginary part of capacitive impedance is usually negative, its absolute value or sign conversion is often used for ease of subsequent analysis. This precise method for extracting the imaginary part ensures the accuracy and stability of the feature values, providing a reliable data foundation for subsequent conductive network distribution analysis.

[0072] This implementation method successfully characterizes the distribution characteristics of the conductive network inside the coating of a lithium-ion battery separator by selecting the imaginary part of the corrected impedance signal in the high-frequency region as the impedance characteristic value, significantly improving the scientific rigor and accuracy of the detection method. This characteristic value selection strategy is based on in-depth electrochemical theoretical analysis and extensive experimental verification, ensuring a direct correspondence between the detection results and the actual state of the conductive network, and avoiding interference from other electrochemical processes. The scientific definition of the high-frequency region and the adaptive selection algorithm achieve broad applicability to different material systems. Intelligent frequency range determination ensures the targeted and accurate extraction of the imaginary part. The precise imaginary part extraction and calculation technology employs advanced digital signal processing methods, ensuring high accuracy and good reproducibility of the characteristic values ​​through multiple extraction strategies and rigorous quality control. The high sensitivity of the high-frequency imaginary part to the conductivity of the conductive network enables the detection method to identify microscopic changes in the conductive network, providing strong support for the fine-grained control of coating quality. It effectively solves the technical problems of strong subjectivity in the selection of characteristic parameters and poor correspondence with physical mechanisms in traditional detection methods. Through scientific selection based on physical principles and precise numerical extraction, it provides objective and reliable quantitative indicators for the electrochemical performance evaluation of lithium-ion battery separator coatings, significantly improving the scientificity and effectiveness of product quality control.

[0073] Figure 3 This document illustrates a flowchart of a lithium-ion battery separator coating detection method provided in an embodiment of this application. Figure 3 As shown in the embodiments of this application, a method for detecting the coating of a lithium-ion battery separator spraying is also provided. This method may include the following steps: S110. Acquire a first signal from the same target location of the lithium-ion battery separator coating. The first signal is a complex impedance signal characterizing the overall electrical properties of the target location. S120. A second signal is synchronously acquired from the same target location. The second signal is a spectral signal characterizing the surface state of the target location. S130. Determine the surface state characteristic parameters of the target location based on the spectral signal; S140. Based on the surface condition characteristic parameters, the complex impedance signal is corrected to reduce or eliminate the measurement influence caused by the surface condition, thereby obtaining a corrected impedance signal. S150. Based on the corrected impedance signal, determine the internal conductive network distribution characteristics of the lithium-ion battery separator coating at the target location.

[0074] Multi-frequency impedance analysis (MDI) technology was employed to acquire complex impedance signals from the target location of the coating on a lithium-ion battery separator. The overall electrical properties of the coating were accurately measured through the collaborative operation of a microelectrode probe array and a MDI. Specifically, the microelectrode probe array was configured with four terminals to contact the coating surface, eliminating the influence of lead resistance and contact resistance and ensuring measurement accuracy. The MDI scanned within a preset frequency range, applying a small signal excitation at each frequency point to avoid electrochemical damage to the coating. Synchronous detection technology was used for complex impedance measurement to accurately measure the amplitude ratio and phase difference of current and voltage, calculating the real and imaginary parts of the impedance. Multiple averaging processes were automatically performed during data acquisition to improve the signal-to-noise ratio and measurement stability. The acquired complex impedance signal contained the resistive, capacitive, and frequency response characteristics of the coating material, comprehensively reflecting the overall electrical behavior at the target location. This broadband measurement method can capture the complete electrical response spectrum of the coating from low-frequency ion conduction to high-frequency electron conduction, providing a rich information foundation for subsequent data analysis and ensuring a comprehensive characterization of the coating's electrochemical properties.

[0075] Near-infrared spectroscopy is used to synchronously acquire spectral signals from the same target location, achieving complete spatiotemporal correspondence with complex impedance measurements. The optical probe and microelectrode probe array employ a coaxial design to ensure precise overlap between the optical and electrical measurement points, eliminating the influence of spatial position deviations on subsequent corrections. The near-infrared spectrometer transmits light to the coating surface via an optical fiber bundle, creating uniform illumination. The light signal reflected from the coating surface is captured by the collecting optical fiber and transmitted back to the spectrometer for analysis, obtaining reflectance spectral data. The synchronous acquisition mechanism is implemented through a hardware trigger signal, ensuring that complex impedance and spectral measurements occur simultaneously, eliminating measurement errors caused by time differences. Spectral signal acquisition parameters include integration time, number of scans, and spectral resolution, ensuring high-quality spectral data. The acquired spectral signals contain key information such as the coating surface chemical composition, material distribution, and surface morphology, particularly the characteristic absorption peaks of polymer binders, conductive additives, and inorganic particles in the near-infrared band, providing a rich data foundation for quantitative analysis of the surface state. The synchronous acquisition strategy ensures the spatiotemporal consistency of the two signals, creating the necessary conditions for accurate signal correction.

[0076] This study utilizes a chemometric model to process spectral signals and extract characteristic parameters representing the surface state at target locations. The process begins with preprocessing the raw spectral data, including baseline correction, smoothing filtering, and standardization to eliminate the effects of instrument noise and environmental interference. The preprocessed spectral data is then input into a pre-trained principal component analysis (PCA) model. This model, built upon spectral data from a large number of standard samples, identifies the main variation patterns in the spectrum related to surface state. PCA projects the high-dimensional spectral data into a low-dimensional principal component space through eigenvalue decomposition of the covariance matrix, calculating 3-5 principal component scores as surface state characteristic parameters. Each principal component score has a clear physical meaning; for example, the first principal component reflects changes in polymer content, the second reflects the dispersion of conductive additives, and the third reflects surface roughness. The numerical range of the characteristic parameters is typically between -3 and +3, with positive and negative values ​​reflecting the direction of deviation from the standard state, and the absolute value indicating the degree of deviation. This machine learning-based feature extraction method achieves intelligent transformation from complex spectral information to concise characteristic parameters, providing an accurate and reliable description of the surface state for subsequent impedance signal correction, significantly improving the objectivity and quantification of surface state assessment.

[0077] Intelligent correction of complex impedance signals based on surface state characteristic parameters eliminates the interference of surface state changes on impedance measurements. The correction process employs a pre-established correction model, which establishes a quantitative relationship between surface state characteristic parameters and complex impedance offset through extensive experimental data. The mathematical expression of the correction model is detailed below and will not be repeated here. The correction calculation substitutes the surface state characteristic parameters at the current position into the model to obtain the corresponding complex impedance offset, which is then subtracted from the original complex impedance signal. The correction process is performed separately for each frequency point to ensure correction accuracy across the entire frequency spectrum. The corrected impedance signal eliminates systematic deviations caused by surface factors such as surface roughness, chemical composition variations, and uneven particle distribution, enabling the measurement results to accurately reflect the distribution of the conductive network within the coating. The correction effect is verified through statistical analysis; the reproducibility and consistency of the corrected impedance data are significantly improved, providing a reliable data foundation for accurately evaluating the characteristics of the conductive network.

[0078] The distribution characteristics of the conductive network within the coating are determined based on the corrected impedance signal, and a quantitative assessment of the conductive network state is achieved through frequency domain analysis and spatial statistical methods. First, the imaginary part of the impedance in the high-frequency region (100kHz-1MHz) is extracted from the corrected complex impedance signal as a characteristic parameter, which sensitively reflects the connectivity and uniformity of the conductive network. Then, by repeating the above measurement and processing process at multiple spatially adjacent locations, two-dimensional spatial distribution data of the impedance characteristic values ​​are obtained. Based on the spatial distribution data, various statistical indicators are calculated: standard deviation and coefficient of variation reflect distribution uniformity, and spatial autocorrelation coefficient reflects spatial aggregation characteristics. A continuous distribution map is generated using spatial interpolation techniques, visually displaying the spatial distribution pattern of the conductive network. These quantitative indicators provide objective numerical standards for coating quality assessment, accurately identifying defects such as conductive particle agglomeration and network breakage, guiding process parameter optimization and product quality improvement. This step represents a technological leap from single-point measurement to regional assessment, providing a comprehensive and in-depth quantitative description of the conductive network distribution characteristics.

[0079] This embodiment achieves accurate detection and comprehensive evaluation of the electrochemical uniformity of lithium-ion battery separator coatings. Simultaneous dual-signal acquisition ensures spatiotemporal consistency of electrical and optical measurements, eliminating systematic errors inherent in traditional separate measurements. The application of a chemometric model enables intelligent processing of spectral information, transforming complex surface state information into quantifiable characteristic parameters. A machine learning-based correction algorithm effectively eliminates interference from surface states on impedance measurements, significantly improving the accuracy of internal conductive network detection. Spatial statistical analysis methods enable quantitative evaluation of conductive network distribution characteristics, providing a scientific basis for quality control. This entire method not only solves the technical challenge of accurately evaluating coating electrochemical uniformity in existing technologies but also provides an advanced detection method for optimizing lithium-ion battery manufacturing processes, significantly improving product quality stability and consistency.

[0080] In one embodiment of this invention, determining the surface state characteristic parameters of the target location based on the spectral signal includes: S210. Input the spectral signal into a pre-trained chemometrics model to reduce the dimensionality of the spectral signal through the chemometrics model and obtain one or more principal component scores. S220. Use the scores of one or more principal components as surface state characteristic parameters.

[0081] The preprocessed spectral signals are input into a pre-trained chemometrics model. Principal component analysis (PCA) is used to effectively reduce the dimensionality of the high-dimensional spectral data and extract key principal component scores. The training process of the chemometrics model utilizes a large amount of spectral data from standard samples with known surface states. These samples cover combinations of different polymer contents, conductive additive dispersion, surface roughness, and particle distribution, ensuring the model has good generalization ability and representativeness. In practical applications, the input spectral signals undergo the same preprocessing steps as the training data, including spectral range truncation (typically selecting the 1000-2500 nm effective band), baseline correction (using polynomial fitting or rubber band algorithm), smoothing filtering, and standardization (scaling the data to the 0-1 range). The preprocessed spectral vector is centered, and the mean vector of the training data is subtracted. Then, it is multiplied by the principal component loading matrix to obtain the principal component scores. The dimensionality reduction process typically compresses high-dimensional spectral data containing 500-1500 wavelength points into 3-8 principal component scores, significantly simplifying data complexity while retaining the core information most relevant to the surface state.

[0082] Based on the results of principal component analysis and statistical principles, one or more of the most representative scores from the calculated principal component scores are selected as surface state characteristic parameters. The principal component selection strategy is based on the cumulative variance contribution rate criterion, as described above, and will not be repeated here. The number of principal components to be retained is determined by analyzing the explanatory power of each principal component on the variance of the original spectral data.

[0083] In lithium-ion battery separator coating applications, the first principal component typically has the highest variance contribution rate, primarily reflecting changes in the polymer binder content; its positive and negative values ​​represent increases or decreases in polymer content relative to the standard state, respectively. The second principal component mainly reflects the dispersion state of the conductive additives; positive values ​​indicate uniform dispersion, while negative values ​​indicate agglomeration. The third principal component may be related to surface roughness or inorganic particle distribution.

[0084] This implementation successfully achieves intelligent transformation from complex high-dimensional spectral data to concise surface state characteristic parameters through dimensionality reduction processing of the chemometric model and scientific selection of principal component scores, providing an advanced technical means for quantitative analysis of the surface state of lithium-ion battery separator coatings. The dimensionality reduction process is based on rigorous mathematical statistical principles, effectively identifying and extracting the key information most relevant to the surface state from the spectral data through principal component analysis algorithms, avoiding biases from subjective human judgment and significantly improving the objectivity and consistency of feature extraction. The cumulative variance contribution rate criterion ensures the representativeness and reliability of the selected principal component scores, maximizing data compression while retaining core information and reducing the computational burden for subsequent processing. Each principal component score has a clear physicochemical meaning, intuitively reflecting the specific state characteristics of the coating surface, such as key parameters like material composition, dispersion uniformity, and surface morphology, providing operable quantitative indicators for process quality control. Through automated dimensionality reduction and parameterized expression, the efficiency, accuracy, and reproducibility of surface state detection are significantly improved, providing a high-quality data foundation for subsequent complex impedance signal correction and conductive network distribution evaluation.

[0085] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0086] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0087] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0088] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0089] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0090] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0091] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0092] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0093] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A lithium-ion battery separator coating detection system, characterized in that, include: The first signal acquisition unit is used to acquire a first signal from the same target position of the coating sprayed on the lithium-ion battery separator. The first signal is a complex impedance signal that characterizes the overall electrical properties of the target position. The second signal acquisition unit is used to synchronously acquire a second signal from the same target location. The second signal is a spectral signal that characterizes the surface state of the target location. The processor is connected to the first and second acquisition units. The processor is configured to execute: The spectral signal is input into a pre-trained chemometrics model; By using a chemometric model, the high-dimensional spectral signal is reduced in dimensionality, and one or more low-dimensional feature parameters characterizing the surface state are output as surface state feature parameters. Based on the surface condition characteristic parameters, the complex impedance signal is corrected to reduce or eliminate the measurement influence caused by the surface condition, thereby obtaining the corrected impedance signal; Based on the corrected impedance signal, the distribution characteristics of the internal conductive network of the lithium-ion battery separator sprayed at the target location are determined. The chemometric model is the principal component analysis model, and the processor is configured to execute it. The spectral signal is projected onto the principal component space of the principal component analysis model to obtain its score on a preset number of principal components. One or more scores are used as surface state feature parameters.

2. The system according to claim 1, characterized in that, The first signal acquisition unit includes a multi-frequency impedance analyzer and a microelectrode probe array. The processor is configured to drive the microelectrode probe array through the multi-frequency impedance analyzer to acquire complex impedance signals at multiple frequencies.

3. The system according to claim 2, characterized in that, The second acquisition unit includes a near-infrared spectrometer and an optical probe coaxially arranged with the microelectrode probe array. The processor is configured to drive the optical probe through the near-infrared spectrometer to acquire the near-infrared spectral signal at the target location.

4. The system according to claim 1, characterized in that, The processor corrects the complex impedance signal based on surface state characteristic parameters, including: Substitute the surface state characteristic parameters into a preset correction model to determine a complex impedance offset caused by the surface state. The corrected impedance signal is obtained by subtracting the complex impedance offset from the complex impedance signal.

5. The system according to claim 1, characterized in that, Based on the corrected impedance signal, the processor determines the distribution characteristics of the internal conductive network of the coating at the target location, including: Extract the impedance characteristic value at the preset frequency from the corrected impedance signal; Based on the spatial variation of impedance characteristic values ​​obtained at multiple spatially adjacent target locations, an index for characterizing the distribution characteristics of the internal conductive network is calculated.

6. The system according to claim 5, characterized in that, The impedance characteristic value is the imaginary part of the corrected impedance signal in the high-frequency region.

7. A method for detecting the coating of a lithium-ion battery separator, applied to the lithium-ion battery separator coating detection system according to any one of claims 1-6, characterized in that, The method includes: A first signal is acquired from the same target location of the coating sprayed onto the lithium-ion battery separator. The first signal is a complex impedance signal that characterizes the overall electrical properties of the target location. A second signal is simultaneously acquired from the same target location. The second signal is a spectral signal that characterizes the surface state at the target location. The spectral signal is input into a pre-trained chemometric model to reduce the dimensionality of the spectral signal and obtain one or more principal component scores. Use the scores of one or more principal components as surface state feature parameters; Based on the surface condition characteristic parameters, the complex impedance signal is corrected to reduce or eliminate the measurement influence caused by the surface condition, thereby obtaining a corrected impedance signal. Based on the corrected impedance signal, the distribution characteristics of the internal conductive network of the lithium-ion battery separator coating at the target location were determined. The chemometric model is the principal component analysis model, and the processor is configured to execute it. The spectral signal is projected onto the principal component space of the principal component analysis model to obtain its score on a preset number of principal components. One or more scores are used as surface state feature parameters.

Images