Battery lithium precipitation fault detection method and device based on charging curve and impedance spectrum

Abstract

The application relates to the technical field of battery management, and discloses a battery lithium precipitation fault detection method and equipment based on a charging curve and impedance spectrum, which comprises the following steps: acquiring a charging curve time series data set of a battery to be detected, identifying a battery charging polarization inflection point, collecting electrochemical impedance spectrum data with the polarization inflection point as a trigger reference, pre-processing the electrochemical impedance spectrum data to obtain a standardized impedance spectrum data set; performing polarization three-component decoupling processing on the charging curve time series data set to separate real-time change data of ohmic, concentration difference and electrochemistry of the battery; performing relaxation time distribution deconvolution decoupling and equivalent circuit model fitting on the standardized impedance spectrum data set to separate battery impedance response and extract an intrinsic electrochemical feature set; extracting a charging polarization feature set, constructing an electrochemical mechanism coupling mapping model, generating a lithium precipitation identification coupling feature matrix, and quantitatively dividing a lithium fault grade; and the application realizes accurate identification and grade division of lithium precipitation faults by early prediction of the lithium precipitation state.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and more specifically to a method and device for detecting lithium plating faults in batteries based on charging curves and impedance spectra. Background Technology

[0002] Lithium plating in batteries refers to the abnormal deposition of metallic lithium on the surface of the negative electrode during the charging process of lithium-ion batteries. Lithium plating reduces battery capacity and cycle life, and may even threaten battery safety. Therefore, lithium plating detection has become a core technical means to ensure the safe operation of batteries and extend their service life.

[0003] In existing technologies, during the constant current and constant voltage charging process of batteries, charging curves and impedance spectrum data are collected simultaneously or at different times. However, due to the lack of a unified time synchronization benchmark, these two types of data fail to complete the electrochemical mechanism coupling model between the polarization characteristics of the charging curve and the impedance spectrum characteristics. As a result, the lithium plating detection results are mostly qualitative judgments of yes / no lithium plating, and it is impossible to quantitatively analyze the lithium plating fault level.

[0004] Against this backdrop, how to quantitatively classify the lithium plating failure level of batteries has become a key issue. Summary of the Invention

[0005] This invention provides a method and device for detecting lithium plating faults in batteries based on charging curves and impedance spectra, in order to solve the problem of quantitatively classifying the levels of lithium plating faults in batteries.

[0006] In a first aspect, the present invention provides a battery lithium plating fault detection method based on charging curves and impedance spectra, the method comprising: The charging curve time series dataset of the battery under test is obtained, the charging polarization inflection point of the battery is identified, and the electrochemical impedance spectroscopy data at the polarization inflection point is collected with the polarization inflection point as the trigger reference. The electrochemical impedance spectroscopy data is then preprocessed by denoising and operating condition normalization to obtain a standardized impedance spectroscopy dataset. The time-series dataset of charging curves is subjected to polarization three-component decoupling processing to separate the real-time changes of ohmic polarization, concentration polarization and electrochemical polarization of the battery, and obtain the polarization three-component decoupling results; Based on the polarization three-component decoupling results, the standardized impedance spectrum dataset is deconvolved with relaxation time distribution and fitted with an equivalent circuit model to separate the battery impedance response and extract the intrinsic electrochemical feature set specific to the lithium plating reaction. From the real-time change data of the three polarization components obtained by separation, the charging polarization feature set related to the critical behavior of lithium plating is extracted, and an electrochemical mechanism coupling mapping model of the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating is constructed to generate the lithium plating identification coupling feature matrix. The lithium plating identification coupling feature matrix is ​​input into the lithium plating determination threshold system, and the lithium plating state determination result is output. For batteries under test that are determined to have undergone lithium plating, the lithium fault level is quantitatively analyzed by matching the coupling feature matrix with the lithium plating degree calibration database, and the lithium plating fault detection results are output.

[0007] This invention achieves in-situ synchronous acquisition of impedance spectrum data by identifying polarization inflection points, decouples the three polarization components of the charging curve to eliminate cross-interference between ohmic, concentration, and electrochemical polarization, and removes irrelevant interference impedance by performing deconvolution decoupling of the relaxation time distribution and equivalent circuit fitting on the impedance spectrum. A coupled mapping model of charging polarization feature set and intrinsic electrochemical feature set of lithium plating is constructed to generate a lithium plating identification coupled feature matrix, so as to predict the lithium plating state in advance, realize the accurate identification and quantitative classification of lithium plating faults that have occurred, and improve the accuracy of lithium plating fault detection.

[0008] In one alternative implementation, identifying the battery charging polarization inflection point includes: The charging curve time series dataset is processed by moving average filtering to obtain the terminal voltage time series curve; The first and second derivatives of the terminal voltage time-series curve are calculated. The first derivative is used to characterize the polarization growth rate during battery charging, and the second derivative is used to characterize the changing trend of the polarization growth rate. If the number of sampling points where the absolute value of the second derivative continuously exceeds the preset abrupt change threshold reaches the preset number, and the relative error of the terminal voltage time series curve deviating from the linear fitting region is greater than or equal to the error threshold, then the sampling point at the current moment is determined to be the polarization inflection point.

[0009] This invention suppresses high-frequency noise interference by performing moving average filtering on the charging curve time series dataset, calculates the first and second derivatives of the terminal voltage time series curve, and identifies polarization inflection points by combining the polarization inflection point conditions, thus ensuring the accuracy of inflection point identification.

[0010] In an optional implementation, before performing polarization three-component decoupling processing on the charging curve time-series dataset to separate the real-time changes in the battery's ohmic polarization, concentration polarization, and electrochemical polarization, the method further includes: Wavelet adaptive filtering is used to perform secondary noise reduction on the charging curve time series dataset; Using the battery open-circuit voltage and state of charge calibration curve as a reference, the terminal voltage during the charging start and end resting stages is corrected. The synchronously acquired current data is linearly corrected using the calibration coefficient of the charge-discharge tester, the synchronously acquired temperature data is corrected for deviation using the standard temperature of the constant temperature chamber, and the synchronously acquired state-of-charge time series data is corrected using the ampere-hour integral method and the open-circuit voltage and state-of-charge curve, resulting in a corrected full-cycle charging terminal voltage, current, temperature, and state-of-charge time series dataset.

[0011] This invention performs secondary noise reduction on the charging curve dataset through wavelet adaptive filtering, filters out power frequency interference, performs reference correction on the terminal voltage, eliminates the error of the charge and discharge tester, and corrects the timing data of current, temperature, and state of charge, effectively eliminating errors.

[0012] In one optional implementation, the charging curve time-series dataset undergoes polarization three-component decoupling processing to separate the real-time changes in the battery's ohmic polarization, concentration polarization, and electrochemical polarization, including: The total polarization voltage is calculated based on the corrected time-series dataset; During the constant current charging phase, a current interruption operation is performed every preset time interval. The voltage change after the current interruption is collected, and the voltage change is determined as the ohmic polarization voltage drop. The linearly interpolated time series data is determined as the ohmic polarization component. Calculation of electrochemical polarization voltage drop based on the Butler-Volmer equation; Calculation of concentration polarization pressure drop based on Fick's first diffusion law; Using the corrected time-series dataset as input, the recursive least squares method is employed to identify the exchange current density and the lithium-ion concentration on the solid surface in real time, thereby obtaining real-time changes in electrochemical polarization and concentration polarization.

[0013] This invention effectively solves the problem of cross-coupling of the three polarizations in traditional schemes by separating ohmic polarization, electrochemical polarization, and concentration polarization, and provides polarization component timing data for subsequent steps.

[0014] In one optional implementation, based on the polarization three-component decoupling results, the normalized impedance spectrum dataset is subjected to relaxation time distribution deconvolution decoupling and equivalent circuit model fitting, including: Construct a basic model of relaxation time distribution; Solve for the relaxation time distribution function in the basic model of relaxation time distribution, plot the relaxation time distribution spectrum, and identify the relaxation peaks in the relaxation time distribution spectrum; By combining the relaxation peak, an equivalent circuit model corresponding to the electrochemical process is constructed, and the equivalent circuit model is fitted using the complex nonlinear least squares method.

[0015] This invention solves the relaxation time distribution function, plots the relaxation time distribution spectrum, identifies relaxation peaks, separates the impedance responses in different chemical processes, constructs an equivalent circuit model, and completes the quantitative separation and cross-validation with the impedance parameters of the electrochemical process.

[0016] In one alternative implementation, a set of charging polarization features related to the critical behavior of lithium plating is extracted, including: Establish rules for evaluating feature correlation. The relative change rate of ohmic impedance at the polarization inflection point, the proportion of ohmic polarization voltage to the total polarization voltage, and the temperature sensitivity coefficient of ohmic impedance are extracted as relevant characteristics of the ohmic polarization component. The electrochemical polarization voltage abrupt change rate at the polarization inflection point, the proportion of electrochemical polarization voltage to the total polarization voltage, and the polarization response time constant of the charge transfer process are extracted as relevant characteristics of the electrochemical polarization components. The polarization voltage growth rate and the proportion of concentration polarization voltage to total polarization voltage at the polarization inflection point are extracted as relevant features of the concentration polarization component. The extracted ohmic polarization component-related features, electrochemical polarization component-related features, and concentration polarization component-related features are classified according to the polarization components to obtain the charging polarization feature set.

[0017] This invention extracts features related to ohmic polarization, electrochemical polarization, and concentration polarization respectively, classifies them according to polarization components, and constructs a dimensionally unified set of charging polarization features to screen polarization features that are strongly related to lithium plating behavior.

[0018] In one optional implementation, an electrochemical mechanism coupling mapping model is constructed between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, generating a lithium plating identification coupling feature matrix, including: Simultaneous standardization and dimension adaptation processing are performed on the charging polarization feature set and the intrinsic electrochemical feature set; Based on the charging polarization feature set and intrinsic electrochemical feature set after synchronous standardization and dimensional adaptation, an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating is constructed according to the weight allocation rule. Substitute the intrinsic electrochemical eigenvectors of lithium plating and the charging polarization eigenvectors into the electrochemical mechanism coupling mapping model, and use a weight matrix to perform weighted fusion to obtain the coupling eigenvectors. The coupling feature vectors are reconstructed in a preset order to generate a lithium plating identification coupling feature matrix.

[0019] This invention eliminates the dimensional differences between the charging polarization feature set and the intrinsic electrochemical feature set by processing them, providing standardized input for subsequent coupling mapping, constructing a coupling mapping model, realizing mechanism-level coupling of the two feature sets, and generating a lithium plating identification coupling feature matrix, thus providing a data foundation for predicting the critical state of lithium plating and identifying faults.

[0020] In one optional implementation, the lithium plating identification coupling feature matrix is ​​input into the lithium plating determination threshold system, and the lithium plating state determination result is output, including: Compliance verification of the lithium plating identification coupling feature matrix; Dimension matching is performed on the feature matrix after compliance verification to obtain threshold matching results; The threshold matching results are matched with the judgment rules to output a lithium plating status judgment report. The lithium plating status includes no lithium plating, critical lithium plating, and lithium plating has occurred.

[0021] This invention verifies the compliance of the generated lithium plating identification coupling feature matrix to ensure that the feature dimensions are complete, the feature value range is compliant, and the data format is consistent. It performs dimensional matching on the feature matrix to achieve the benchmarking of the coupling features and the threshold system. Combined with the judgment rules, it outputs a lithium plating status judgment report to achieve the prediction of lithium plating status.

[0022] In one optional implementation, the lithium fault level is quantitatively analyzed by matching the coupling feature matrix with a lithium plating degree calibration database, and the lithium plating fault detection results are output, including: Calculate the similarity between the coupling feature matrix and the feature matrix of the lithium-plated calibrated sample; By selecting benchmark samples that meet the similarity criteria, and using similarity as the weight, a weighted average is calculated on the selected benchmark samples to obtain a quantitative assessment value of the lithium plating degree of the battery under test. The lithium plating failure level is obtained by matching the quantitative assessment value of the degree of lithium plating with the grading standard.

[0023] This invention uses a weighted average, with the similarity between the calculated coupling feature matrix and the feature matrix of the lithium-plated calibrated sample as weights, to quantitatively assess the degree of lithium plating in the battery under test, determine the lithium plating fault level, achieve accurate classification of lithium plating fault levels, and provide a decision-making basis for graded early warning.

[0024] In a second aspect, the present invention provides an electronic device, comprising: a memory and a processor, wherein the memory and the processor are communicatively connected to each other, the memory stores computer instructions, and the processor executes the computer instructions to perform the battery lithium plating fault detection method based on charging curve and impedance spectrum described in the first aspect or any corresponding embodiment. Attached Figure Description

[0025] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0026] Figure 1 This is a schematic flowchart of a battery lithium plating fault detection method based on charging curves and impedance spectra according to an embodiment of the present invention. Figure 2 This is a schematic diagram of the polarization inflection point triggering and synchronous acquisition according to an embodiment of the present invention; Figure 3 This is a schematic flowchart illustrating the lithium plating state classification determination according to an embodiment of the present invention; Figure 4 This is a structural block diagram of a battery lithium plating fault detection device based on charging curves and impedance spectra according to an embodiment of the present invention. Figure 5 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0028] It is understood that before using the technical solutions disclosed in the various embodiments of the present invention, users should be informed of the types, scope of use, and usage scenarios of the personal information involved in the present invention in an appropriate manner in accordance with relevant laws and regulations, and their permission should be obtained.

[0029] In related technologies, lithium plating detection in batteries mainly includes the following two methods: The first method is based on macroscopic time-series data such as terminal voltage, current, and temperature in the battery charging curve, and indirectly judges the lithium plating state by identifying features such as abnormal voltage fluctuations and charging platform shifts. The second method introduces electrochemical impedance spectroscopy (EIS) technology, which attempts to characterize the internal electrochemical processes of the battery from a microscopic level by measuring the battery's impedance response at different frequencies, in order to identify lithium plating faults.

[0030] Furthermore, existing technologies combine charging curve data with electrochemical impedance spectroscopy (EIS) data, simultaneously or sequentially acquiring charging curve and impedance spectroscopy data during the constant current and constant voltage charging process of the battery, attempting to achieve multi-dimensional information fusion to diagnose lithium plating faults. However, these two types of data lack a unified time synchronization benchmark and a true in-situ acquisition mechanism, failing to achieve precise decoupling of the three polarization components (ohmic polarization, electrochemical polarization, and concentration polarization), and also failing to establish an electrochemical mechanism coupling model between charging polarization characteristics and impedance spectroscopy characteristics. This results in insufficient accuracy in lithium plating detection, making it impossible to achieve advanced prediction of critical states and quantitative classification of fault levels.

[0031] Existing battery lithium plating detection technologies have the following drawbacks: (1) The detection lag is prominent, making it impossible to predict the critical state of lithium plating in advance: Traditional lithium plating detection methods based on charging curves (terminal voltage, current, temperature, etc.) can only capture the macroscopic features of lag after lithium plating occurs, such as abnormal voltage fluctuations and charging platform shifts. These features usually only appear after irreversible lithium plating has occurred and battery performance has significantly degraded. They cannot provide early warnings in the early stages of the lithium plating reaction evolving from "critical occurrence" to "irreversible lithium plating", resulting in significant lag in the detection results and making it difficult to meet the needs of active battery safety control. (2) The detection results are easily affected by the cross-interference of multiple polarization processes, resulting in insufficient accuracy and stability: During the charging process, there are three physical and chemical processes in the battery at the same time: ohmic polarization, electrochemical polarization and concentration polarization. The time-domain response characteristics of the three processes are coupled and superimposed on the terminal voltage signal. The traditional charging curve analysis method fails to achieve accurate decoupling of the above three polarization processes. The lithium plating fault characteristics are easily submerged by the interference signals of other polarization processes, resulting in a high false alarm rate and false negative rate. The robustness and reliability of the detection results are difficult to guarantee. (3) Electrochemical impedance spectroscopy lacks mechanistic correlation and it is difficult to extract specific features of lithium plating: When electrochemical impedance spectroscopy is used alone, only the overall impedance response spectrum of the battery can be obtained. It does not establish an electrochemical mechanism correlation with the polarization characteristics during the charging process. It cannot effectively eliminate interference impedance components unrelated to lithium plating (such as SEI film impedance, diffusion impedance, ohmic impedance, etc.), and it is also difficult to extract specific intrinsic features of lithium plating reaction from complex impedance spectra, resulting in weak detection specificity and poor feature interpretability. (4) The charging curve and impedance spectrum data lack a unified synchronization reference, and the timing misalignment leads to data distortion: The charging curve data and electrochemical impedance spectrum data are usually collected independently by different devices, with different sampling strategies, and in different time windows. There is a lack of a unified high-precision synchronization clock reference and in-situ acquisition mechanism between the two types of data. The impedance spectrum acquisition process often requires interruption of charging or change of battery conditions, resulting in significant timing misalignment and operating condition differences between the acquired impedance spectrum data and the charging curve data. When data is fused and analyzed, feature mismatch and information distortion are likely to occur. (5) Failure to achieve dual feature set mechanism coupling, making it impossible to quantitatively classify and analyze lithium fault levels: The electrochemical mechanism coupling model between the polarization characteristics of the charging curve and the impedance spectrum characteristics has not been completed, and there is a lack of a unified method for constructing the lithium plating identification coupling feature matrix. As a result, the lithium plating detection results are mostly qualitative judgments (such as "yes / no lithium plating"), lacking the ability to quantitatively assess the degree of lithium plating (slight, moderate, severe). At the same time, the existing scheme cannot output a charging strategy adjustment threshold that matches the degree of lithium plating, making it difficult to achieve active suppression and closed-loop control of lithium plating faults.

[0032] This invention provides a battery lithium plating fault detection method based on charging curves and impedance spectra, aiming to solve the following technical problems: (1) Achieve advanced prediction of critical state of lithium plating and solve the problem of detection lag: By constructing a synchronous acquisition mechanism and a graded threshold judgment system based on the polarization inflection point, the early characteristic signals of the evolution of lithium plating reaction from "critical occurrence" to "irreversible lithium plating" are captured, and the advanced warning of critical state of lithium plating is realized. Before the lithium plating fault causes irreversible damage, an alarm is issued, providing active safety control capability for the battery system. (2) Eliminate cross-interference in polarization process and improve detection accuracy and stability: By performing precise decoupling processing of the three polarization components (ohmic polarization, electrochemical polarization and concentration polarization) on the charging curve data, the real-time change data of each polarization component is separated, interference information unrelated to lithium plating is eliminated, and charging polarization features strongly related to the critical behavior of lithium plating are extracted, thereby reducing the false judgment rate and the missed detection rate, and improving the accuracy of detection results and the adaptability of working conditions. (3) Extracting specific intrinsic features of lithium plating reaction to enhance detection specificity: By performing deconvolution decoupling of the distribution of relaxation times (DRT) and fitting of the equivalent circuit model on the electrochemical impedance spectroscopy data, combined with the polarization three-component constraint boundary, the impedance response of different electrochemical processes inside the battery is separated, and interference impedance components unrelated to lithium plating (such as SEI film impedance, diffusion impedance, ohmic impedance, etc.) are eliminated. Specific intrinsic electrochemical feature set of lithium plating reaction is extracted to enhance the physical interpretability and specificity of detection. (4) Achieve time synchronization and operating condition locking for the two types of data and eliminate data distortion: By introducing a high-precision synchronous clock trigger unit, the polarization inflection point in the charging process is used as a unified trigger reference. At the inflection point, the charging curve feature data and multi-band electrochemical impedance spectroscopy data are collected in situ and synchronously. The timestamps of the two types of data are accurately aligned and the operating conditions are bound together, eliminating data distortion caused by time misalignment and operating condition differences, and providing high-quality and high-fidelity input data for dual feature fusion analysis. (5) Construct a dual-feature set mechanism coupling model to achieve quantitative classification and closed-loop control of lithium plating fault levels: By constructing an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, a standardized lithium plating identification coupling feature matrix is ​​generated. Combined with a preset graded threshold system and a lithium plating degree calibration database, it is possible to predict the critical state of lithium plating, accurately identify lithium plating faults that have occurred, and quantitatively classify the lithium plating fault level (slight / moderate / severe). At the same time, graded early warning signals and charging strategy adjustment thresholds that match the degree of lithium plating are output synchronously, providing executable active control parameters for the battery management system and forming a closed-loop control capability of "detection-early warning-regulation".

[0033] According to an embodiment of the present invention, a battery lithium plating fault detection method based on charging curve and impedance spectrum is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Although the logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0034] This embodiment provides a battery lithium plating fault detection method based on charging curves and impedance spectra. Figure 1 This is a flowchart of a battery lithium plating fault detection method based on charging curves and impedance spectra according to an embodiment of the present invention, as shown below. Figure 1 As shown, the process includes the following steps: Step S101: Obtain the charging curve time series dataset of the battery under test, identify the battery charging polarization inflection point, use the polarization inflection point as the trigger reference, collect the electrochemical impedance spectroscopy data at the polarization inflection point, and perform noise reduction and operating condition normalization preprocessing on the electrochemical impedance spectroscopy data to obtain a standardized impedance spectroscopy dataset.

[0035] In this embodiment of the invention, during the constant current and constant voltage charging process of the new energy battery under test, a time-series dataset of the charging curve of the battery under test is collected in real time. Based on the time-series dataset of the charging curve, the polarization inflection point of the battery charging process is identified. Using the polarization inflection point as the trigger reference, multi-band in-situ electrochemical impedance spectroscopy data at the corresponding inflection point time are collected synchronously, and the timestamp alignment and operating condition locking storage of the two types of data are completed. At the same time, the synchronously collected electrochemical impedance spectroscopy data is preprocessed by denoising and operating condition normalization to obtain a standardized impedance spectroscopy dataset.

[0036] Step S102: Perform polarization three-component decoupling processing on the charging curve time series dataset to separate the real-time change data of ohmic polarization, concentration polarization and electrochemical polarization of the battery, and obtain the polarization three-component decoupling results.

[0037] In this embodiment of the invention, the real-time changes of ohmic polarization, concentration polarization and electrochemical polarization of the battery under test during the entire charging process are separated from the charging curve time series dataset to obtain the polarization three-component decoupling results.

[0038] Step S103: Based on the polarization three-component decoupling results, the standardized impedance spectrum dataset is subjected to relaxation time distribution deconvolution decoupling and equivalent circuit model fitting to separate the battery impedance response and extract the intrinsic electrochemical feature set specific to the lithium plating reaction.

[0039] In this embodiment of the invention, based on the decoupling results of the three polarization components, the standardized impedance spectrum dataset is subjected to relaxation time distribution DRT deconvolution decoupling and equivalent circuit model fitting to separate the impedance responses of different electrochemical processes inside the battery, eliminate interference impedance components unrelated to lithium plating, and extract the intrinsic electrochemical feature set specific to the lithium plating reaction.

[0040] Step S104: Extract the charging polarization feature set related to the critical behavior of lithium plating from the real-time change data of the separated polarization three components, construct an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, and generate a lithium plating identification coupling feature matrix.

[0041] In this embodiment of the invention, a charging polarization feature set strongly correlated with the critical behavior of lithium plating is extracted from the real-time change data of the three polarization components. An electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating is constructed to generate a lithium plating identification coupling feature matrix.

[0042] Step S105: Input the lithium plating identification coupling feature matrix into the lithium plating determination threshold system and output the lithium plating state determination result.

[0043] In this embodiment of the invention, the lithium plating identification coupling feature matrix of the battery under test is input into a preset lithium plating judgment threshold system to complete the advance prediction of the critical state of lithium plating of the battery under test, accurately identify the lithium plating fault that has occurred, and output the lithium plating state judgment result.

[0044] Step S106: For the test battery that is determined to have undergone lithium plating, match it with the lithium plating degree calibration database based on the coupling feature matrix, quantitatively analyze the lithium fault level, and output the lithium plating fault detection result.

[0045] In this embodiment of the invention, for the battery under test that is determined to have undergone lithium plating, the coupling feature matrix is ​​matched with the pre-established lithium plating degree calibration database to complete the quantitative classification of the lithium plating fault level, and the lithium plating fault detection results, graded early warning signals and charging strategy adjustment thresholds are output simultaneously.

[0046] The lithium plating fault detection method based on charging curve and impedance spectrum provided in this embodiment achieves in-situ synchronous acquisition of impedance spectrum data by identifying polarization inflection points. The charging curve is decoupled into three polarization components to eliminate cross-interference between ohmic, concentration, and electrochemical polarization. By performing deconvolution decoupling and equivalent circuit fitting on the impedance spectrum relaxation time distribution, irrelevant interference impedance is eliminated. A coupled mapping model of charging polarization feature set and intrinsic electrochemical feature set of lithium plating is constructed to generate a lithium plating identification coupled feature matrix. This allows for the early prediction of lithium plating status, enabling accurate identification and quantitative classification of lithium plating faults, and improving the accuracy of lithium plating fault detection.

[0047] This embodiment provides a battery lithium plating fault detection method based on charging curves and impedance spectra. The process includes the following steps: Step S201: Obtain the charging curve time series dataset of the battery under test, identify the battery charging polarization inflection point, use the polarization inflection point as the trigger reference, collect the electrochemical impedance spectroscopy data at the polarization inflection point, and perform noise reduction and operating condition normalization preprocessing on the electrochemical impedance spectroscopy data to obtain a standardized impedance spectroscopy dataset.

[0048] like Figure 2 As shown, the first step is to build a data acquisition system to provide the hardware foundation, unified time reference, and accuracy guarantee for data acquisition. This includes the following steps: (1) Pre-calibration of basic parameters of the battery under test: Using the nominal parameters of the battery factory, confirm the nominal capacity of the battery under test as 50Ah, the charging cut-off voltage as 4.2V, the 0.5C constant current charging current as 25A, and the constant voltage cut-off current as 2.5A. After the pre-calibration is completed, fully charge the battery under test and then discharge it to SOC=0%, and let it stand for later use. (2) Synchronous acquisition system construction: The acquisition system consists of a host computer controller, a programmable battery charge and discharge tester, a multi-channel electrochemical workstation, a high-precision temperature acquisition module, and a synchronous clock triggering unit; among them, the programmable battery charge and discharge tester is used to execute the constant current and constant voltage charging process and acquire the charging curve time series dataset, the multi-channel electrochemical workstation is used to acquire electrochemical impedance spectroscopy data, and the synchronous clock triggering unit is electrically connected to all other acquisition devices to provide a 10MHz level unified synchronous clock source for the entire system and eliminate the inherent clock deviation between devices; (3) Accuracy calibration of the acquisition system: Before formal acquisition, the charge-discharge tester was calibrated using a standard resistor and a standard voltage source. After calibration, the voltage acquisition accuracy was ≤ ±0.05%FS and the current acquisition accuracy was ≤ ±0.1%FS. The electrochemical workstation was calibrated using a standard electrochemical impedance calibrator, with the calibration frequency band covering 10 -3 Hz~10 6 Hz, impedance amplitude measurement error ≤1%, phase angle measurement error ≤0.5°; temperature acquisition module after calibration, acquisition accuracy ≤±0.2℃, system parameters are locked after calibration, and midway modification is prohibited.

[0049] Then, the charging curve time-series dataset of the battery under test is collected to achieve uninterrupted, timestamped collection of time-series data for the entire charging process. This includes the following steps: (1) Preparing for charging: Place the battery under test in a high and low temperature constant temperature test chamber, control the ambient temperature to be stable at 25℃±2℃, let it stand for 2 hours to allow the cell body temperature to reach thermal equilibrium with the ambient temperature, and confirm that the cell body temperature is stable at 25℃±1℃. (2) Start of standard constant current and constant voltage charging process: First, constant current charging is performed at a constant current of 25A at a rate of 0.5C until the real-time terminal voltage of the battery under test reaches the charging cutoff voltage of 4.2V. Then, it is switched to constant voltage charging mode to keep the charging voltage constant at 4.2V until the charging current drops to the cutoff current of 2.5A. The charging process ends. The above complete process is the whole process of constant current and constant voltage charging.

[0050] (3) Real-time acquisition of charging curve time series dataset: At the same time as the constant current and constant voltage charging process starts, the charging curve time series dataset of the battery under test is continuously acquired by the charge and discharge tester at a sampling frequency of 50Hz. The acquired parameters include real-time terminal voltage u(t), real-time charging current i(t), charging time t, and cell body temperature T(t). The state of charge SOC(t) of the battery under test is calculated and recorded in real time by the ampere-hour integration method. All acquired data are bound to the real-time timestamp provided by the synchronous clock trigger unit to form a timestamped charging curve time series dataset, which is uploaded to the host computer controller in real time.

[0051] Specifically, identifying the battery charging polarization inflection point in step S201 above includes: Step S2011: Perform moving average filtering on the charging curve time series dataset to obtain the terminal voltage time series curve; Step S2012: Calculate the first and second derivatives of the terminal voltage time-series curve. The first derivative is used to characterize the polarization growth rate during the battery charging process, and the second derivative is used to characterize the changing trend of the polarization growth rate. Step S2013: If the number of sampling points where the absolute value of the second derivative continuously exceeds the preset mutation threshold reaches the preset number, and the relative error of the terminal voltage time series curve deviating from the linear fitting region is greater than or equal to the error threshold, then the sampling point at the current moment is determined to be the polarization inflection point.

[0052] In this embodiment of the invention, the host computer controller performs a moving average filtering process on the real-time collected charging curve time series dataset. The width of the moving window is set to 10 sampling points to eliminate high-frequency noise interference during the acquisition process and obtain a smoothed terminal voltage time series curve U(t).

[0053] Based on the smoothed terminal voltage time-series curve U(t), the first derivative dU / dt and the second derivative dt are calculated in real time. 2 U / dt 2 Wherein, the first derivative dU / dt characterizes the polarization growth rate during battery charging, and the second derivative d... 2 U / dt 2 Characterizes the changing trend of polarization growth rate.

[0054] A dual threshold for polarization inflection points is set. A polarization inflection point is determined to be the moment when both of the following conditions are met: Threshold 1: During the constant current charging stage, the second derivative d 2 U / dt 2 The number of sampling points whose absolute value continuously exceeds a preset mutation threshold reaches a preset number, for example, the preset number is 3, and the preset mutation threshold K = 0.2mV / s. 2 ; Threshold 2: The relative error of the synchronous verification terminal voltage U(t) deviating from the linear fitting interval is greater than or equal to the error threshold. For example, the error threshold is 2%, and the linear fitting interval is the terminal voltage linear growth interval corresponding to the first 60% SOC of the constant current charging stage.

[0055] For example, at a time t0 with a charging time of 32 minutes and a SOC of 78%, if both of the above conditions are met, this time is determined to be the polarization inflection point. The inflection point time t0 is recorded, and the corresponding battery operating parameters are locked simultaneously: SOC(t0)=78%, cell temperature T(t0)=26.1℃, charging current i(t0)=25A, and terminal voltage U(t0)=4.18V.

[0056] By performing moving average filtering on the charging curve time series dataset to suppress high-frequency noise interference, the first and second derivatives of the terminal voltage time series curve are calculated. Combined with the polarization inflection point condition, the polarization inflection point is identified to ensure the accuracy of inflection point identification.

[0057] After identifying the battery charging polarization inflection point, multi-band in-situ electrochemical impedance spectroscopy is simultaneously acquired. Using the polarization inflection point as the sole trigger reference, synchronous acquisition of in-situ impedance spectra is achieved. The specific steps include the following: (1) Synchronous trigger command issuance: When the host computer controller determines the polarization inflection point at time t0, it sends the impedance spectrum acquisition trigger command to the multi-channel electrochemical workstation through the synchronous clock trigger unit. The time delay between command transmission and triggering is ≤1ms, ensuring that the acquisition time is strictly synchronized with the polarization inflection point time t0. (2) In-situ acquisition conditions: During the impedance spectrum acquisition process, the constant current and constant voltage charging process of the battery under test is not interrupted, and the set values ​​of charging current and charging voltage are not changed. The charging conditions and thermodynamic state of the battery are kept stable, so as to realize in-situ acquisition and avoid the distortion of impedance spectrum data caused by changes in operating conditions. (3) Multi-band impedance spectroscopy acquisition parameter settings: The electrochemical workstation uses the sinusoidal potential perturbation method for impedance spectroscopy acquisition. The perturbation signal amplitude is set to 10mV to avoid interference with the battery charging state. The acquisition frequency band is set to full-band logarithmic scan, covering the high-frequency band 10 3 Hz~10 6 Hz, mid-frequency band 10 -1 Hz~10 3 Hz, low frequency band 10 -3Hz~10 -1 Hz, the number of full-band scanning points is set to 80, and they are evenly distributed to each frequency band in a logarithmic distribution; (4) Impedance spectrum data acquisition and timestamp binding: After the electrochemical workstation completes the full-band impedance spectrum scan, it obtains an in-situ electrochemical impedance spectrum dataset containing frequency points, real part of impedance, imaginary part of impedance, and phase angle. The acquisition timestamp provided by the synchronous clock trigger unit is bound synchronously to ensure that the timestamp of the impedance spectrum dataset is consistent with the timestamp of the charging curve data at the polarization inflection point t0. After the acquisition is completed, the dataset is uploaded to the host computer controller.

[0058] Finally, timestamp alignment and condition-locked storage are performed on the two types of data to achieve accurate alignment and structured storage. This includes the following steps: (1) Time stamp precise alignment processing: The host computer controller uses the unified clock source of the synchronous clock trigger unit as the reference to perform time stamp alignment on the charging curve time series dataset and the in-situ electrochemical impedance spectroscopy dataset with timestamps; The alignment method is as follows: Taking the polarization inflection point time t0 as the reference zero point, the time series data in the t0±1s interval of the charging curve data is averaged to obtain the standardized charging curve feature data corresponding to the inflection point time t0; The time stamp reference of the electrochemical impedance spectroscopy dataset is uniformly corrected to t0 to eliminate the timing deviation caused by the scanning time, and the time stamp alignment of the two types of data is completed. After alignment, the timing deviation is ≤8ms; (2) Operating condition locking storage: The aligned charging curve feature data and in-situ electrochemical impedance spectroscopy dataset are uniquely bound to the battery operating condition parameters (SOC=78%, cell temperature 26.1℃, charging current 25A, terminal voltage 4.18V) locked at the inflection point t0 to form a complete dataset corresponding to a single inflection point; the dataset is stored in the local database of the host computer controller according to the preset structured format, and the traceability information such as the battery number, charging rate 0.5C, and test environment temperature 25℃ corresponding to the dataset is recorded to complete the operating condition locking storage.

[0059] It should be noted that only one polarization inflection point is identified during the constant current charging stage. After completing the above process, charging curve data is continuously collected until the entire constant current and constant voltage charging process is terminated. If a new polarization inflection point is identified during subsequent charging, the above triggering acquisition, alignment and storage process is repeated.

[0060] To provide a high-fidelity dataset for subsequent impedance spectroscopy decoupling, the electrochemical impedance spectroscopy data is denoised, specifically including the following steps: (1) Outlier identification and removal: The 3σ criterion combined with the K-means clustering algorithm was used to identify outliers in the in-situ electrochemical impedance spectroscopy data collected in step S201. Based on the Nyquist curve of the impedance spectrum, the impedance data of the whole frequency band was divided into 3 clusters (high frequency inductance region, mid frequency charge transfer region, and low frequency diffusion region). The mean and standard deviation of the real and imaginary parts of the impedance were calculated for the data in each cluster. Outliers exceeding the mean ± 3 times the standard deviation were removed. At the same time, physically invalid data points with an absolute phase angle exceeding 90° were removed to complete the outlier removal. (2) Adaptive noise reduction: For the impedance spectrum data after removing outliers, Savitzky-Golay smoothing filter is used for adaptive noise reduction. The filter window width is set to 7 frequency points and the polynomial order is set to 3. While preserving intrinsic electrochemical information such as impedance spectrum semicircular features and inflection point features, random noise and power frequency interference during the acquisition process are eliminated. (3) Data integrity completion: For the missing frequency point data after removing outliers, cubic spline interpolation is used to complete the data, ensuring the continuity and integrity of the frequency points in the whole frequency band. Finally, the denoised impedance spectrum dataset is obtained, which contains complete and valid data of the frequency points in the whole frequency band, the corresponding real part of the impedance, the imaginary part of the impedance, and the phase angle.

[0061] To eliminate the influence of operating condition differences on impedance spectra, the electrochemical impedance spectroscopy data are preprocessed with operating condition normalization to obtain a standardized impedance spectroscopy dataset. This process includes the following steps: (1) Standard operating condition preset: The preset standard operating conditions are: ambient temperature 25℃, SOC=50%, charging rate 0.5C, and all impedance spectrum data are normalized to this standard operating condition; (2) Multi-factor coupling normalization correction: Temperature Correction: An impedance temperature correction model is established based on the Arrhenius equation to eliminate the influence of temperature on the impedance spectrum. The correction formula is as follows:

[0062] in, Z norm,T This is the temperature-corrected impedance value. Z raw This is the original impedance value after noise reduction. E a This is the activation energy of the electrochemical reaction (35 kJ / mol in this example). R The gas constant is T std The standard operating temperature is 298.15 K. T test The thermodynamic temperature of the battery cell at the time of testing; SOC calibration: Based on the pre-calibrated SOC-impedance reference curve, an SOC calibration coefficient matrix is ​​established. The calibration coefficients of the test SOC and the standard SOC are obtained by linear interpolation, thus completing the correction of the influence of SOC on the impedance spectrum. The impedance spectrum of the battery under test is pre-calibrated in the range of 10%~100% SOC with 5% SOC intervals through standard testing to obtain the impedance reference value for the entire SOC range. Charging current correction: Based on the linear correlation model between concentration polarization and charging current, current correction is performed on the diffusion impedance in the low-frequency band to eliminate the influence of charging current on the impedance of the low-frequency diffusion region. The correction coefficient is the ratio of standard operating current to test current. (3) Output of standardized impedance spectrum dataset: After completing the full-factor correction of temperature, SOC, and charging current, the corrected impedance spectrum data is resampled in the frequency band and unified to the standard frequency band distribution (10). -3 Hz~10 6 (Hz, 80 logarithmically uniformly distributed frequency points), finally obtaining the standardized impedance spectrum dataset under standard operating conditions, completing the entire preprocessing process.

[0063] Step S202: Perform polarization three-component decoupling processing on the charging curve time series dataset to separate the real-time change data of ohmic polarization, concentration polarization and electrochemical polarization of the battery, and obtain the polarization three-component decoupling results.

[0064] Specifically, the input data consists of a time-series dataset of charging curves from a 50Ah ternary lithium-ion battery cell and an in-situ electrochemical impedance spectroscopy dataset corresponding to the polarization inflection point t0. First, the charging curve time-series dataset undergoes noise reduction and benchmark correction to provide an interference-free foundation dataset for subsequent polarization decoupling. This includes the following steps: Step Sa involves using wavelet adaptive filtering to perform secondary noise reduction on the charging curve time series dataset. Step Sb: Using the battery open-circuit voltage and state of charge calibration curve as a reference, perform reference correction on the terminal voltage during the charging start and end resting stages. Step Sc involves linearly correcting the synchronously acquired current data using the calibration coefficient of the charge / discharge tester, correcting the deviation of the synchronously acquired temperature data using the standard temperature of the constant temperature chamber, and performing closed-loop correction on the synchronously acquired state-of-charge time-series data using the ampere-hour integral method and the open-circuit voltage versus state-of-charge curve, thereby obtaining the corrected full-cycle charging terminal voltage, current, temperature, and state-of-charge time-series dataset.

[0065] In this embodiment of the invention, wavelet adaptive filtering is used to perform secondary noise reduction on the collected charging curve time series dataset. The db4 wavelet basis is selected, the number of decomposition layers is set to 5, and a soft thresholding function is used to perform thresholding on the high-frequency detail components. The threshold selection is based on the Stein unbiased risk estimation principle to eliminate power frequency interference and sensor random noise, while fully preserving the voltage change characteristics corresponding to the polarization inflection point, thus avoiding feature distortion caused by filtering.

[0066] Using the battery open-circuit voltage (OCV)-SOC calibration curve as a reference, the terminal voltage at the start and end of the charging resting phases is calibrated. The OCV-SOC curve of the battery under test is pre-calibrated according to the GB / T31485-2015 standard, with a calibration interval of 5% SOC, to obtain the standard OCV reference value for the entire SOC range. The theoretical OCV reference value for each moment of the entire charging cycle is obtained through linear interpolation, eliminating the system zero-point drift error of the charge / discharge tester. The correction formula is:

[0067] in, U corr ( t To correct the back-end voltage, U taw ( t () represents the original acquisition terminal voltage. The zero-point drift error of the system is determined by the difference between the terminal voltage after 1 hour of rest at the start of charging and the standard OCV at 0% SOC.

[0068] Secondary noise reduction of the charging curve dataset is performed by wavelet adaptive filtering to filter out power frequency interference, and the terminal voltage is corrected to eliminate the error of the charge and discharge tester. The timing data of current, temperature and state of charge are also corrected to effectively eliminate errors.

[0069] Synchronous correction is performed on the synchronously acquired current, temperature, and state of charge (SOC) time-series data. Specifically, linear correction is performed on the synchronously acquired current data using the calibration coefficient of the charge-discharge tester, deviation correction is performed on the synchronously acquired temperature data using the standard temperature of the constant temperature chamber, and closed-loop correction is performed on the synchronously acquired SOC data using the ampere-hour integral method and the open-circuit voltage versus SOC curve to eliminate accumulated SOC error. The corrected full-cycle charging time-series dataset of terminal voltage, current, temperature, and SOC is obtained.

[0070] Specifically, in step S202 above, the polarization three-component decoupling processing is performed on the charging curve time-series dataset to separate the real-time changes in the battery's ohmic polarization, concentration polarization, and electrochemical polarization, including: Step S2021: Calculate the total polarization voltage based on the corrected time-series dataset; Step S2022: During the constant current charging stage, a current interruption operation is performed every preset time interval. The voltage change after the current interruption is collected, the voltage change is determined as the ohmic polarization voltage drop, and the linearly interpolated time series data is determined as the ohmic polarization component. Step S2023: Calculate the electrochemical polarization voltage drop based on the Butler-Volmer equation; Step S2024: Calculate the concentration polarization pressure drop based on Fick's first diffusion law; Step S2025: Using the corrected time-series dataset as input, the recursive least squares method is used to identify the exchange current density and the lithium-ion concentration on the solid surface in real time, thereby obtaining real-time change data of electrochemical polarization and concentration polarization.

[0071] In this embodiment of the invention, based on the difference in response time constants of the three types of polarization, a step-by-step identification method is used to complete the full-cycle decoupling of the three polarization components.

[0072] Based on the corrected time-series dataset, the total polarization voltage during the entire charging process is calculated in real time. The calculation formula is as follows:

[0073] in, U pol ( t Let be the total polarization voltage at time t. OCV ( SOC ( t ), T ( t )for t The reference value of the battery open-circuit voltage at the corresponding SOC and temperature.

[0074] For the ohmic polarization component, based on the microsecond-level response characteristics of ohmic polarization, a pulse current interruption method is used to complete the full-cycle identification of ohmic polarization. During the constant current charging phase, a 100ms current interruption operation is performed every 2 minutes. At the moment of interruption, the current drops from 25A to 0A. The voltage jump is collected 100μs after the interruption, and this jump is the ohmic polarization voltage drop. The calculation formula is as follows:

[0075] in, U ohm ( t )for t At any given moment, the ohmic polarization voltage drop. I ( t )for t Constant charging current, R ohm ( t )for t The battery ohmic impedance is constantly identified.

[0076] For the ohmic impedance between two adjacent identification intervals, linear interpolation is used to obtain the total charging cycle. R ohm ( t The time-series data is then used to separate the real-time changes in ohmic polarization throughout the charging process, thus decoupling the ohmic polarization component from the total polarization voltage.

[0077] Based on the difference in response time constants between electrochemical polarization (millisecond level) and concentration polarization (second-minute level), a dual-polarization kinetic model based on the Butler-Volmer equation is adopted, and the parameter identification and decoupling of the two components are completed by combining the recursive least squares method.

[0078] The basic decoupling model is as follows:

[0079] in, U act ( t )for t Electrochemical polarization voltage drop at any time U conc ( t )for t Concentration polarization pressure drop at any given moment.

[0080] The electrochemical polarization voltage drop is calculated based on the Butler-Volmer equation, and the formula is as follows:

[0081] in, R The gas constant is... T The thermodynamic temperature of the battery cell. α is the charge transfer coefficient (value 0.5). n The number of electrons transferred (value 1). F It is Faraday's constant. I 0( t ) represents the exchange current density.

[0082] The concentration polarization pressure drop is calculated based on Fick's first diffusion law, and the formula is as follows:

[0083] in, c s,max This represents the maximum lithium-ion intercalation concentration for the negative electrode active material. c s,surf ( t ) represents the lithium ion concentration on the surface of the negative electrode solid phase.

[0084] Based on the above model, using the corrected time-series dataset as input, the recursive least squares method is used to identify the exchange current density and the lithium-ion concentration on the solid surface in real time, and to calculate the real-time changes in electrochemical polarization and concentration polarization throughout the charging process.

[0085] The three polarization components obtained by separation are superimposed and verified. The relative error between the superimposed value and the total polarization voltage is ≤3% and it is considered that the decoupling is effective. The real-time change time series dataset of ohmic polarization, electrochemical polarization and concentration polarization during the entire charging process is output. The decoupling processing of the three polarization components is completed, which effectively solves the problem of cross-coupling of the three polarizations in the traditional scheme and provides polarization component time series data for subsequent steps.

[0086] The three polarization components obtained by decoupling at the polarization inflection point t0 are: ohmic polarization voltage drop 24.8mV, electrochemical polarization voltage drop 38.2mV, concentration polarization voltage drop 57.5mV, and total polarization voltage 120.5mV. The superposition verification relative error is 1.2%, which meets the decoupling accuracy requirements.

[0087] Step S203: Based on the polarization three-component decoupling results, the standardized impedance spectrum dataset is subjected to relaxation time distribution deconvolution decoupling and equivalent circuit model fitting to separate the battery impedance response and extract the intrinsic electrochemical feature set specific to the lithium plating reaction.

[0088] Specifically, the input data consists of a polarization three-component time-series dataset and a normalized impedance spectrum dataset corresponding to the polarization inflection point t0. First, a DRT deconvolution decoupling preprocessing based on polarization three-component constraints is performed to provide mechanistic constraint boundaries for DRT decoupling and avoid multiple solutions in the decoupling process. This includes the following steps: (1) Polarization-frequency band correspondence matching: Based on the decoupling results of the three polarization components and combined with the relaxation time characteristics of the electrochemical process of lithium-ion batteries, a one-to-one correspondence between polarization components and impedance spectrum frequency bands is established: High frequency band (10) 3 Hz~10 6 Hz): corresponds to the ohmic polarization process, with a relaxation time range of 10. -9 s~10 -6 s; Mid-to-high frequency band (10Hz~10 3 Hz): Corresponds to the SEI film interface process, relaxation time range 10. -4 s~10 -2 s; Mid-frequency band (10 -1 Hz~10Hz): Corresponds to the electrochemical polarization (charge transfer) process, with a relaxation time range of 10 Hz. -2 s~10 1 s, directly coupled with the lithium plating reaction; Low frequency band (10 -3Hz~10 -1 Hz: Corresponds to concentration polarization (lithium-ion solid-state diffusion) process, relaxation time range 10. 1 s~10 3 s; (2) Determination of DRT calculation frequency band and constraint boundary: Using the relaxation time range of the polarization three components as constraints, the calculation relaxation time range of DRT deconvolution is determined to be 10. -9 s~10 3 s, covering the electrochemical process across the entire frequency band; at the same time, based on the decoupling results of the three polarization components, the number of relaxation peaks in each frequency band is constrained: 1 relaxation peak in the high frequency band, 1 relaxation peak in the mid-high frequency band, 12 relaxation peaks in the mid frequency band, and 1 relaxation peak in the low frequency band, providing hard constraint boundaries for DRT deconvolution. (3) Impedance spectrum data preprocessing: Frequency domain resampling is performed on the standardized impedance spectrum dataset to obtain the complex impedance form corresponding to 80 frequency points with a logarithmic uniform distribution:

[0089] Where ω = 2πf is the angular frequency. The real part of the impedance. The imaginary part of the impedance is used to prepare the input data for DRT deconvolution.

[0090] Specifically, step S203 above, based on the polarization three-component decoupling results, involves performing relaxation time distribution deconvolution decoupling and equivalent circuit model fitting on the standardized impedance spectrum dataset, including: Step S2031: Construct a basic model of relaxation time distribution; Step S2032: Solve for the relaxation time distribution function in the basic model of relaxation time distribution, plot the relaxation time distribution spectrum, and identify the relaxation peaks in the relaxation time distribution spectrum; In step S2033, an equivalent circuit model corresponding to the electrochemical process is constructed by combining the relaxation peak, and the equivalent circuit model is fitted by the complex nonlinear least squares method.

[0091] In this embodiment of the invention, to achieve relaxation time domain decoupling of the impedance spectrum and separate the impedance responses of different electrochemical processes, a basic relaxation time distribution (DRT) model is first constructed. Based on linear system theory, the DRT model expression for the battery impedance spectrum is as follows:

[0092] in, For high-frequency ohmic impedance, For relaxation time, The relaxation time distribution function, or DRT spectrum, characterizes the polarization intensity corresponding to different relaxation times.

[0093] The Tikhonov regularization algorithm is used to solve the first type of Fredholm integral equation (i.e., the DRT model expression of the battery impedance spectrum). The number of relaxation peaks and the relaxation time range determined in the previous steps are used as hard constraints. A regularization term is added to suppress oscillations and multiple solutions. The regularization parameter λ is adaptively selected using the Generalized Cross-Validation (GCV) method. For example, the optimal regularization parameter λ is chosen to be 10. -3 .

[0094] Then, the relaxation time distribution function is obtained by solving the problem. The relaxation time distribution spectrum of DRT was plotted, and independent relaxation peaks in the spectrum were identified. Based on the polarization-frequency band correspondence, the electrochemical process and polarization component attribution of each relaxation peak were determined. In this embodiment of the invention, DRT decoupling yielded a total of 5 independent relaxation peaks, and the attribution results are as follows: Peak 1 ( ): High-frequency ohmic processes, corresponding to ohmic polarization; Peak 2 ( ): SEI film interface process, no corresponding polarization component assigned; Peak 3 ( ): The negative electrode charge transfer process corresponds to electrochemical polarization and is directly coupled with the lithium plating reaction; Peak 4 ( ): The lithium plating reaction interface process corresponds to the electrochemical polarization branch and is a specific relaxation peak for the lithium plating reaction; Peak 5 ( ): The solid-phase diffusion process of lithium ions, corresponding to concentration polarization.

[0095] Finally, based on the DRT decoupling results, an equivalent circuit was constructed to complete the quantitative separation and cross-validation of the impedance parameters of each electrochemical process. Combining the five relaxation peaks obtained from DRT decoupling, an equivalent circuit model corresponding to each electrochemical process was constructed. The model topology is as follows:

[0096] The electrochemical meaning and corresponding relaxation peak of each element are as follows: L : High-frequency inductance, corresponding to the parasitic inductance of the test cable, with no corresponding relaxation peak; R s Ohmic impedance, corresponding to peak 1, matching ohmic polarization; R SEI SEI film impedance Q SEI : The constant phase element at the SEI film interface, the two connected in parallel correspond to peak 2; R ct Negative charge transfer impedance Q dl : The electrode interface double-layer constant phase element, the two are connected in parallel and correspond to peak 3, which matches the electrochemical polarization; R Li Lithium plating reaction interface impedance Q Li The lithium plating reaction interface is a constant phase element, and the two elements connected in parallel correspond to peak 4, which is a specific element for lithium plating. W o : A finite-space Warburg diffusion element, corresponding to peak 5, which is matched with concentration polarization.

[0097] The equivalent circuit model was fitted using the complex nonlinear least squares (CNLS) method, with a standardized impedance spectrum dataset as the fitting target and initial values ​​of each component obtained from DRT decoupling as the fitting starting point. The fitting convergence condition was that the sum of squared residuals (SSE) ≤ 10. -4 The maximum number of iterations is set to 1000.

[0098] In addition, cross-validation is performed to fit the results. R s , R ct , W o The parameters were cross-validated with the ohmic polarization, electrochemical polarization, and concentration polarization results obtained from decoupling in step S202, respectively. The validation criterion was the linear correlation coefficient between the two. R 2 ≥0.95; In this embodiment of the invention, the fitted ohmic impedance R s =0.992mΩ, which deviates from the ohmic impedance obtained by decoupling in step S202 by 0.8%; charge transfer impedance R ct =1.528mΩ, correlation coefficient with electrochemical polarization results R 2 =0.982; Correlation coefficient between Warburg diffusion impedance and concentration polarization results R 2 =0.976, which meets the verification requirements, and the fitting results are valid.

[0099] By solving the relaxation time distribution function, plotting the relaxation time distribution spectrum, and identifying relaxation peaks, impedance responses in different chemical processes can be separated, an equivalent circuit model can be constructed, and quantitative separation and cross-validation with the impedance parameters of the electrochemical process can be completed.

[0100] Specifically, to complete the elimination of interfering components and the screening of lithium-deposition-specific features, the intrinsic electrochemical feature set is constructed, and the extraction of intrinsic electrochemical feature set specific to lithium-deposition reaction includes the following steps: (1) Elimination of interference components unrelated to lithium plating: Based on the electrochemical mechanism of lithium plating reaction in lithium-ion batteries and combined with the polarization component attribution, interference components and impedance components that are not directly related to the lithium plating reaction are eliminated. The elimination list is as follows: High-frequency interference components: inductance L Ohmic impedance R s It is only related to the conductive components of the battery and is unrelated to the lithium plating reaction; Interface interference component: SEI film impedance R SEI SEI film constant phase element Q SEI It is only related to the SEI film formation / aging process and is not directly coupled with the lithium plating reaction; Diffusion interference component: Warburg diffusion element W o It is only related to the solid-phase diffusion process of lithium ions and is not an intrinsic feature of lithium plating reaction; (2) Screening of specific features of lithium plating reaction: Based on the retained lithium plating-related elements and DRT relaxation peaks, intrinsic feature parameters directly related to the occurrence and development of the lithium plating reaction were screened and divided into two categories: Category 1: Lithium plating reaction-specific impedance characteristics, corresponding to DRT peak 4 and the equivalent circuit ( R Li -Q Li Parallel branches, including: lithium plating reaction interface impedance R Li Lithium plating reaction interface constant phase element Q Li capacitance value Q Li -T and dispersion coefficient n Li Peak intensity of characteristic relaxation peaks in lithium plating in DRT spectra I Li Peak With half-peak width W Li ; Category 2: Charge transfer characteristics strongly coupled with lithium plating reaction, corresponding to DRT peak 3 and equivalent circuit ( R ct -Q dl Parallel branches, including: negative charge transfer impedance R ct Double-layer constant phase elementQ dl capacitance value Q dl-T With diffusion coefficient n dl Peak intensity of relaxation peak during charge transfer process I ct With peak position offset Construction of intrinsic electrochemical feature set for lithium plating: The 10 feature parameters obtained above are standardized to form a lithium plating reaction-specific intrinsic electrochemical feature set with unified dimensions and clear mechanism, which is then output to the subsequent step S204.

[0101] Step S204: Extract the charging polarization feature set related to the critical behavior of lithium plating from the real-time change data of the separated polarization three components, construct an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, and generate a lithium plating identification coupling feature matrix.

[0102] Specifically, the charging polarization feature set extracted in step S204 above related to the critical behavior of lithium plating includes: Step S2041: Establish feature correlation evaluation rules; Step S2042: Extract the relative change rate of ohmic impedance, the proportion of ohmic polarization voltage to total polarization voltage, and the temperature sensitivity coefficient of ohmic impedance at the polarization inflection point as relevant features of ohmic polarization components. Step S2043: Extract the electrochemical polarization voltage abrupt change rate at the polarization inflection point, the proportion of electrochemical polarization voltage to the total polarization voltage, and the polarization response time constant of the charge transfer process as related features of the electrochemical polarization component. Step S2044: Extract the polarization voltage growth rate and the proportion of concentration polarization voltage to total polarization voltage at the polarization inflection point as relevant features of concentration polarization components. Step S2045: The extracted ohmic polarization component related features, electrochemical polarization component related features, and concentration polarization component related features are classified according to the polarization components to obtain the charging polarization feature set.

[0103] In this embodiment of the invention, based on the lithium plating reaction mechanism of the negative electrode of a lithium-ion battery, a correlation evaluation system between polarization characteristics and critical lithium plating behavior is established. The Pearson correlation coefficient is used as the core evaluation index. Combined with the polarization response characteristics of the lithium plating reaction, the feature screening threshold is determined: features with a correlation coefficient |r|≥0.8 with critical lithium plating behavior are considered strongly correlated features and are included in the charging polarization feature set; features with |r|<0.8 are considered weakly correlated / irrelevant features and are removed.

[0104] Based on the above evaluation rules, strongly correlated feature parameters are extracted from the real-time data of the three types of polarization components obtained by decoupling in step S202: Strong correlation characteristics of ohmic polarization components: The lithium plating reaction, accompanied by the thickening of the SEI film on the negative electrode and the growth of lithium dendrites, will induce a nonlinear increase in ohmic impedance. Therefore, the following three characteristics are extracted: (1) Relative rate of change of ohmic impedance at the polarization inflection point The calculation formula is:

[0105] in, R ohm ( t 0) represents the ohmic impedance at the polarization inflection point. R ohm,ref The reference average ohmic impedance for the first 60% SOC range during the constant current charging phase; (2) The proportion of ohmic polarization voltage to total polarization voltage k ohm ; (3) Ohmic impedance-temperature sensitivity coefficient α ohm .

[0106] Strong correlation characteristics of electrochemical polarization components: The lithium plating reaction is directly coupled with the charge transfer process at the negative electrode. Electrochemical polarization undergoes a sudden change when the lithium plating criticality occurs, which is the most direct time-domain polarization characterization of lithium plating behavior. Therefore, the following three characteristics are extracted: (1) Electrochemical polarization voltage abrupt change rate at the polarization inflection point The calculation formula is:

[0107] in, U act ( t 0) represents the electrochemical polarization voltage at the polarization inflection point. U act,fit ( t 0) The linear fitting curve of electrochemical polarization voltage in the first 60% SOC range of the constant current charging stage. t The fitted value at time 0.

[0108] (2) The proportion of electrochemical polarization voltage to total polarization voltage k act ; (3) Polarization response time constant of charge transfer process .

[0109] Strong correlation characteristics of concentration polarization components: The lithium plating reaction intensifies the concentration polarization of lithium ion solid-phase diffusion at the negative electrode, triggering a nonlinear increase in concentration polarization. Therefore, the following two characteristics are extracted: (1) Concentration polarization voltage growth rate at the polarization inflection point The calculation formula is the change in the slope of the concentration polarization voltage within 1 minute before and after the inflection point; (2) The proportion of concentration polarization voltage to total polarization voltage k conc ; The eight strongly correlated feature parameters obtained from the above screening are categorized and organized according to polarization components to construct a charging polarization feature set. All parameters in the feature set are bound to the polarization inflection point. t The timestamp of 0 is aligned with the intrinsic electrochemical feature set of lithium plating in step S203 to achieve time reference consistency. In this embodiment, the polarization inflection point... t The core parameters of the charging polarization feature set extracted at time 0 are: =4.2%, k ohm =20.6%, =18.7%, k act =31.7%, =12.3%, k conc =47.7%, and the correlation coefficients |r| between all characteristics and critical lithium plating behavior are ≥0.85, which meets the requirements for strong correlation screening.

[0110] By extracting features related to ohmic polarization, electrochemical polarization, and concentration polarization respectively, and classifying them according to polarization components, a unified charging polarization feature set is constructed to screen polarization features strongly related to lithium plating behavior.

[0111] Specifically, step S204 above constructs an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, generating a lithium plating identification coupling feature matrix, including: Step S2046: Simultaneously standardize and adapt the charging polarization feature set and intrinsic electrochemical feature set. Step S2047: Based on the charging polarization feature set and intrinsic electrochemical feature set after synchronous standardization and dimensional adaptation, an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating is constructed according to the weight allocation rule. Step S2048: Substitute the intrinsic electrochemical feature vector of lithium plating and the feature vector of charging polarization into the electrochemical mechanism coupling mapping model, and use a weight matrix to perform weighted fusion to obtain the coupling feature vector; Step S2049: Reconstruct the coupling feature vectors according to a preset order to generate a lithium plating identification coupling feature matrix.

[0112] In this embodiment of the invention, to eliminate the dimensional differences between the two feature sets, complete dimensional adaptation, and provide standardized input for subsequent coupling mapping, the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating are subjected to synchronization standardization processing and dimensional adaptation processing. Specifically: (1) Synchronous standardization: The Min-Max normalization method is used to simultaneously standardize the charging polarization feature set (8 dimensions) and the intrinsic electrochemical feature set of lithium plating obtained in step S203 (10 dimensions), mapping all feature parameters to the [0,1] interval to eliminate the differences in the dimensions and orders of magnitude of different features. (2) Dimensional adaptation processing: The standardized dual feature sets are dimensionally aligned. The 8-dimensional features of the charging polarization feature set and the 10-dimensional features of the lithium plating intrinsic electrochemical feature set are grouped and adapted according to the correspondence between "polarization component - electrochemical process". The grouping rules are as follows: The first group: Ohmic polarization feature group, corresponding to features related to the Ohmic process; The second group: Electrochemical polarization characteristics, corresponding to characteristics related to charge transfer and lithium plating reaction; The third group: concentration polarization feature group, corresponding to lithium-ion diffusion-related features; after completing the grouping, ensure that the grouping of the dual feature sets corresponds one-to-one, providing a suitable dimensional structure for mechanism coupling mapping.

[0113] To achieve mechanistic coupling between the two feature sets, coupling weights are assigned based on the contribution of each feature parameter to the lithium plating reaction. The weight allocation follows the core rules below: The lithium plating reaction-specific feature has the highest weight, ranging from 0.15 to 0.2. The charge transfer characteristic strongly coupled with lithium plating has the second highest weight, ranging from 0.08 to 0.12. The polarization characteristics indirectly related to lithium plating are weighted again, with a weight range of 0.03 to 0.07; The sum of the weights of all features is 1, ensuring the normalization of the coupled model.

[0114] Based on the dual feature sets after grouping and adaptation, and combined with the weight allocation rules, a weighted coupling mapping model with mechanistic constraints is constructed. The model expression is as follows: Fcouple=Welec·Felec+Wpol·Fpol Where Fcouple is the coupled eigenvector, Felec is the normalized intrinsic electrochemical eigenvector of lithium plating, Welec is the weight matrix of electrochemical features, Fpol is the normalized charging polarization eigenvector, and Wpol is the weight matrix of polarization features.

[0115] Among them, the weight matrices Welec and Wpol are assigned based on the lithium plating contribution mechanism. In this embodiment, the weight assignment result is as follows: Lithium plating specific impedance characteristics (6 dimensions): single feature weight 0.1, total weight 0.6; Charge transfer coupling characteristics (4-dimensional): single feature weight 0.05, total weight 0.2; Electrochemical polarization characteristics (3D): Single feature weight 0.04, total weight 0.12; Ohmic polarization + concentration polarization features (5 dimensions): single feature weight 0.016, total weight 0.08; The sum of all weights is 1, which meets the normalization requirement.

[0116] The coupling mapping model was validated using a pre-calibrated full sample set. The validation criterion was that the coupling feature matrix had a discrimination efficiency of ≥95% for the lithium-free, critical lithium-plating, and lithium-plating states, ensuring the effectiveness and reliability of the model. In this embodiment of the invention, the discrimination efficiency for the three states was 98.2%, which meets the validation requirements.

[0117] The standardized intrinsic electrochemical eigenvectors of lithium plating and the charging polarization eigenvectors are substituted into the coupling mapping model, and weighted fusion is performed through the weight matrix to obtain an 18-dimensional coupling eigenvector.

[0118] The feature matrix is ​​reconstructed. Specifically, the coupling feature vectors are reconstructed in the dimensional order of "lithium plating specific features, charge transfer coupling features, and polarization correlation features" to generate a lithium plating identification coupling feature matrix with dimension M×N, where M=1 is the single sample dimension and N=18 is the feature dimension; if it is a multi-inflection point multi-sample scenario, M is the number of samples and N remains unchanged at 18 feature dimensions. Matrix standardization and output: Specifically, the reconstructed coupling feature matrix undergoes secondary normalization to ensure that all elements within the matrix are within the [0, 1] interval, forming the final lithium plating identification coupling feature matrix, which is then output to the lithium plating determination stage in subsequent step S205. In this embodiment, the final generated lithium plating identification coupling feature matrix is ​​a 1×18-dimensional standardized matrix, with each feature dimension having a clear electrochemical mechanism explanation, which can be directly used for lithium plating critical state prediction and fault identification.

[0119] By processing the charging polarization feature set and the intrinsic electrochemical feature set, the dimensional differences between the two feature sets are eliminated, providing standardized input for subsequent coupling mapping. A coupling mapping model is constructed to achieve mechanistic-level coupling of the two feature sets, generating a lithium plating identification coupling feature matrix, thus providing a data foundation for predicting the critical state of lithium plating and identifying faults.

[0120] Step S205: Input the lithium plating identification coupling feature matrix into the lithium plating determination threshold system and output the lithium plating state determination result.

[0121] Specifically, step S205 includes: Step S2051: Perform compliance verification on the lithium plating identification coupling feature matrix; Step S2052: Perform dimensional matching on the feature matrix after compliance verification to obtain the threshold matching result; Step S2053: Match the threshold matching result with the judgment rule and output a lithium plating status judgment report. The lithium plating status includes no lithium plating, critical lithium plating, and lithium plating has occurred.

[0122] In embodiments of the present invention, such as Figure 3 As shown, the input data is the 1×18-dimensional lithium plating identification coupling feature matrix of the battery under test generated in step S204, which provides a standardized threshold benchmark and judgment rule for lithium plating determination, and completes the system construction and calibration in advance.

[0123] First, a full-condition calibration sample library was constructed. One hundred 50Ah ternary lithium-ion cells of the same model and batch as the battery under test were selected to build a calibration sample library covering the entire life cycle and all operating conditions. The samples are grouped as follows: Lithium-free sample group: 30 fresh cells, cycle count ≤50 times, no history of lithium plating, confirmed by dissection analysis and ICP-MS testing that no metallic lithium was deposited; Critical lithium plating sample group: 30 cells with 200-400 cycles. Three-electrode testing confirmed that the negative electrode potential dropped below 0V for the first time. Dissection analysis confirmed that only a trace amount of reversible lithium was deposited on the negative electrode surface, and there were no irreversible lithium dendrites. The lithium-plated sample group consisted of 40 cells with 500-1500 cycles. Dissection analysis confirmed the presence of obvious lithium dendrites on the negative electrode surface, with a capacity decay of ≥5%. Among them, 20 cells had slight lithium plating and 20 cells had severe lithium plating. For all samples, the full-process test of steps S201 to S204 was completed within the full operating range of charging rate from 0.2C to 1C and ambient temperature from -20℃ to 45℃. The lithium plating identification coupling feature matrix of each sample was collected. At the same time, the results of dissection analysis, ICP-MS test and three-electrode test were used as true value labels for lithium plating state to complete the sample library calibration.

[0124] Then, the multi-dimensional threshold boundaries were determined. Based on the calibrated sample library, kernel density estimation and Fisher linear discriminant analysis were used to determine the three-class state classification thresholds corresponding to the 18 feature dimensions of the coupled feature matrix, where: Lithium plating-specific feature dimension: Set two-level boundaries: critical threshold and fault threshold. If the feature value exceeds the critical threshold but does not exceed the fault threshold, it is judged as critical lithium plating. If it exceeds the fault threshold, it is judged as lithium plating has occurred. Strongly coupled feature dimension: Set an auxiliary judgment threshold. When the feature value exceeds the threshold, the confidence of the corresponding judgment result is increased. It is not used as the sole basis for judgment.

[0125] The threshold calibration results of the core features are shown in Table 1.

[0126] Table 1

[0127] Judgment rules are solidified: Hierarchical judgment rules are established, as follows: Lithium-free state: All core feature dimensions have not exceeded the critical threshold, and the confidence level of the judgment is ≥90%; Critical lithium plating state: ≥2 core feature dimensions exceed the critical threshold, and no core feature dimension exceeds the fault threshold, thus completing the advanced prediction of the critical lithium plating state; Lithium plating has occurred: if ≥1 core feature dimension exceeds the fault threshold, or ≥4 core feature dimensions exceed the critical threshold, it is determined that a lithium plating fault has occurred. Threshold system cross-validation: The threshold system was validated using the 5-fold cross-validation method. The validation results showed that the lithium plating state classification accuracy was 98.5%, the false positive rate was 0.8%, and the false negative rate was 0.7%, which met the detection accuracy requirements. The threshold system was then solidified and preset into the detection system.

[0128] To complete the verification and threshold matching calculation of the input data, the lithium plating identification coupling feature matrix is ​​first verified for compliance. The 1×18-dimensional lithium plating identification coupling feature matrix of the battery under test generated in step S204 is received, and the compliance verification is completed first. The verification items include: feature dimension completeness (no missing dimensions), feature value range compliance (all feature values ​​are within the standardized interval [0, 1]), and data format consistency. In this embodiment, the feature matrix of the battery under test is determined to be a valid input if it passes all the verification items.

[0129] After the verification is passed, the 18 feature dimensions of the feature matrix are matched dimension by dimension with the corresponding thresholds in the preset threshold system. The relative deviation of each feature value from the corresponding critical threshold and fault threshold is calculated. At the same time, the judgment contribution of each feature dimension is calculated in combination with the weight coefficient.

[0130] The matching results of the core features of the battery under test are as follows: Lithium plating reaction interface impedance R Li =0.32mΩ, exceeding the critical threshold of 0.25mΩ, but not exceeding the fault threshold of 0.8mΩ; Electrochemical polarization voltage abrupt change rate t =18.7%, exceeding the critical threshold by 10%, but not exceeding the fault threshold by 25%; Lithium plating characteristic peak intensity I Li =0.21 It exceeds the critical threshold of 0.15. The fault threshold of 0.4 was not exceeded. ; Negative charge transfer impedance R ct =1.528mΩ, which does not exceed the critical threshold of 2.0mΩ; Of the remaining auxiliary feature dimensions, three exceeded the auxiliary judgment threshold, while the rest were within the non-lithium plating range.

[0131] Based on the matching results and the judgment rules, the lithium plating state of the battery under test is classified and judged. The threshold matching results are substituted into the preset classification and judgment rules. In this embodiment of the invention, three core feature dimensions exceed the critical threshold, and no core feature dimensions exceed the fault threshold, which fully meets the judgment conditions for the critical lithium plating state.

[0132] Based on the number of features exceeding the threshold, the magnitude of the feature value exceeding the threshold, and the weighting coefficient ratio, the confidence level of this judgment is calculated using the following formula:

[0133] in, C To determine the confidence level, w i For the first i The weight coefficients of each feature x i For eigenvalues, T th,i The confidence level is set to 100% to correspond to the critical threshold.

[0134] The calculated confidence level of this judgment is 97.2%, which is ≥90%, and the judgment result is valid.

[0135] Grading and determination conclusion: The battery under test was ultimately determined to be in the critical state of lithium plating, completing the advanced prediction of the critical state of lithium plating, and no irreversible lithium plating fault that had occurred was detected.

[0136] Standardized Judgment Report Generation: Based on the graded judgment conclusion, a standardized judgment report is generated. The report content includes: the battery number under test, nominal capacity of 50Ah, test conditions (0.5C charging rate, 25℃ ambient temperature), polarization inflection point time and corresponding conditions, details of threshold matching of coupling feature matrix, lithium plating status label (critical lithium plating), judgment confidence level of 97.2%, risk warnings and suggestions.

[0137] Output results: The lithium plating status determination result (critical lithium plating) is synchronously output to the subsequent step S206 for quantitative classification and graded early warning of lithium plating fault level; at the same time, the determination report, feature matching data, and original collected data are synchronously stored in the local database to complete data traceability and archiving.

[0138] By performing compliance verification on the generated lithium plating identification coupling feature matrix, we ensure that the feature dimensions are complete, the feature value range is compliant, and the data format is consistent. We then perform dimensional matching on the feature matrix to achieve the benchmarking of the coupling features and the threshold system. Combined with the judgment rules, we output a lithium plating status judgment report to achieve the prediction of lithium plating status.

[0139] Step S206: For the test battery that is determined to have undergone lithium plating, the lithium fault level is quantitatively analyzed by matching the coupling feature matrix with the lithium plating degree calibration database, and the lithium plating fault detection result is output.

[0140] Specifically, in step S206 above, the lithium fault level is quantitatively analyzed by matching the coupling feature matrix with the lithium plating degree calibration database, and the output lithium plating fault detection results include: Step S2061: Calculate the similarity between the coupling feature matrix and the feature matrix of the lithium-plated calibration sample; Step S2062: Select benchmark samples that meet the similarity criteria, and use similarity as the weight to perform a weighted average on the selected benchmark samples to obtain a quantitative evaluation value of the degree of lithium plating of the battery under test. Step S2063: Match the quantitative assessment value of lithium plating degree with the grading standard to obtain the lithium plating failure level.

[0141] In this embodiment of the invention, a standardized benchmark database is provided for quantitative assessment of the degree of lithium plating. The database is constructed and calibrated in advance, and the true value of the sample library is calibrated. The true value of the degree of lithium plating is calibrated for 40 samples that have been plating lithium. The mass of metallic lithium deposited on the negative electrode of each sample is accurately measured by inductively coupled plasma mass spectrometry (ICP-MS) to calculate the degree of lithium plating (lithium plating mass fraction). At the same time, the capacity decay rate and DC internal resistance growth rate of each sample are tested simultaneously to complete the true value calibration of the samples. The calibration results are divided into three level ranges: Slight lithium plating: Lithium plating degree 0.5%~2%, capacity decay rate 5%~10%, low safety risk; Moderate lithium plating: Lithium plating degree 2%~5%, capacity decay rate 10%~20%, medium safety risk; Severe lithium plating: Lithium plating degree >5%, capacity decay rate >20%, high safety risk; A database mapping relationship is established. For all calibrated samples, the lithium plating identification coupling feature matrix of each sample is obtained through the full-process test of steps S201 to S204 of this invention. A one-to-one mapping relationship between the 18-dimensional feature matrix and the true value of lithium plating degree, fault level, and capacity decay rate is established to form a lithium plating degree calibration database. At the same time, the feature weight coefficients determined in step S204 are synchronously embedded into the database to provide weight constraints for subsequent matching calculations.

[0142] Database cross-validation and solidification: The leave-one-out method is used to cross-validate the database. The validation results show that the average error of quantitative prediction of lithium plating degree is ≤4.2% and the accuracy of fault level classification is ≥99%. After meeting the accuracy requirements, the database is solidified and preset into the detection system and vehicle BMS (Battery Management System).

[0143] To complete the matching of the feature matrix to be tested with the database and obtain a quantitative assessment value of the degree of lithium plating, the 1×18-dimensional lithium plating identification coupling feature matrix of the battery to be tested that has been determined to have undergone lithium plating in step S205 is received. The core weighted feature dimension is extracted and aligned with the feature dimension in the database to ensure that the matching dimension corresponds one-to-one.

[0144] The weighted cosine similarity algorithm is used to calculate the similarity between the feature matrix to be tested and the feature matrices of all lithium-plated calibrated samples in the database. The calculation formula is as follows:

[0145] in, Sim ( X , Y ) is the matrix to be tested X With the calibration sample matrix Y The similarity score ranges from 0 to 1, with values ​​closer to 1 indicating higher similarity. w i For the first i The weight coefficients of the dimensional features are consistent with the coupling weights in step S204. x i , y i The first and second parts of the test matrix and the calibration sample matrix are respectively the first and second parts of the calibration sample matrix. i Standardized eigenvalues.

[0146] To quantitatively calculate the degree of lithium plating, the top three benchmark samples with the highest similarity were selected. Using similarity as the weight, the true values ​​of the lithium plating degree of the three samples were weighted and averaged to obtain the quantitative assessment value of the lithium plating degree of the battery under test. The calculation formula is as follows:

[0147] in, Li ratio This is a quantitative assessment value of the degree of lithium plating in the battery under test. Sim k For the first k Similarity between the benchmark samples Li ratio,k For the first k True values ​​of lithium deposition degree of a benchmark sample.

[0148] For example, the highest similarity obtained by matching the battery under test was 0.982, and the quantitative assessment value of lithium plating degree obtained by weighted calculation of the first 3 benchmark samples was 2.87%.

[0149] To determine the matching level, the quantitative assessment value of the lithium plating degree of the battery under test is substituted into the preset level classification standard. In this embodiment, the lithium plating degree of 2.87% is in the range of 2% to 5%, which meets the level judgment standard of moderate lithium plating, corresponding to medium safety risk, and the control priority is level two.

[0150] The verification level was reviewed by combining the capacity decay rate test results of the battery under test (the capacity decay rate in this test was 12.8%) and the extent to which the core feature dimensions exceeded the fault threshold. The fault level was reviewed, and the review results were consistent with the quantitative classification results, so the level was determined to be valid.

[0151] Fault information is archived by synchronously archiving the quantitative value of lithium plating degree, fault level, capacity decay rate, and core characteristic data of the battery under test to the calibration database, completing the iterative optimization of the database, and further improving the accuracy of subsequent matching.

[0152] The system completes standardized result output, graded early warning, and charging strategy adjustment threshold generation. First, it outputs standardized lithium plating fault detection results, generating a standardized test report that includes the battery number under test, nominal capacity, test conditions, polarization inflection point information, quantitative assessment value of lithium plating degree (2.87%), lithium plating fault level (moderate lithium plating), capacity decay rate, and safety risk level. The report is then stored in the local database to complete the traceability and archiving of test results.

[0153] Output graded early warning signals. Based on the lithium plating fault level, match the corresponding early warning signal. The early warning grading rules are as follows: Slight lithium plating: A yellow warning signal is issued, indicating that the battery has slight lithium plating and the development trend of lithium plating needs to be monitored. Moderate lithium plating: An orange warning signal is output, indicating that the battery has moderate lithium plating and the charging strategy needs to be optimized, and high-rate charging should be restricted. Severe lithium plating: A red warning signal is output, indicating that the battery has severe lithium plating and a high safety risk. High-rate charging and discharging should be stopped immediately, and battery replacement is recommended. In this embodiment, an orange secondary warning signal is output simultaneously, which can be directly transmitted to the vehicle warning system of the vehicle BMS to alert the user and the vehicle controller through the instrument panel.

[0154] Charging strategy adjustment threshold output: Based on the lithium plating fault level and degree of lithium plating, combined with the electrochemical mechanism, a charging strategy adjustment threshold adapted to the battery under test is generated. While suppressing the development of lithium plating, the battery's usability performance is also taken into account. The core adjustment threshold output in this embodiment is shown in Table 2.

[0155] Table 2

[0156] The aforementioned adjustment thresholds can be directly written into the charging control module of the vehicle's BMS, automatically executing the optimized charging strategy to suppress further aggravation of lithium plating on the negative electrode from the root cause, thus achieving proactive prevention and control of lithium plating faults.

[0157] By using the similarity between the calculated coupling feature matrix and the feature matrix of the lithium-plated calibrated sample as weights, a weighted average is performed to quantitatively assess the degree of lithium plating in the battery under test, determine the lithium plating fault level, achieve accurate classification of lithium plating fault levels, and provide a decision-making basis for graded early warning.

[0158] The battery lithium plating fault detection method based on charging curves and impedance spectra provided in this embodiment has the following advantages: (1) Using the polarization inflection point as a reference, the in-situ synchronous acquisition and precise timing alignment of the charging curve and electrochemical impedance spectrum are achieved to eliminate data distortion; (2) For the first time, the three components of ohmic polarization, electrochemical polarization and concentration polarization were precisely decoupled during the charging process, effectively eliminating cross-interference; (3) Based on the deconvolution of relaxation time distribution and equivalent circuit fitting, a 10-dimensional intrinsic electrochemical feature set specific to lithium plating reaction is extracted to enhance the physical interpretability of detection. (4) Construct a mechanism coupling mapping model of charging polarization characteristics and intrinsic electrochemical characteristics of lithium plating with weighted allocation, and generate an 18-dimensional normalized coupling feature matrix; (5) Based on the graded threshold system, the critical state of lithium plating is predicted in advance and the fault is accurately identified. The calibration database and weighted similarity matching are combined to quantitatively classify the mild / moderate / severe lithium plating levels.

[0159] It should be noted that this battery lithium plating fault detection method based on charging curves and impedance spectra can be extended to various battery systems (such as lithium iron phosphate, sodium ion, and solid-state batteries) and charging strategies (pulse charging, multi-stage constant current, etc.).

[0160] This embodiment also provides a battery lithium plating fault detection device based on charging curves and impedance spectra. This device is used to implement the above embodiments and preferred embodiments, and details already described will not be repeated. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0161] This embodiment provides a battery lithium plating fault detection device based on charging curves and impedance spectra, such as... Figure 4 As shown, it includes: The acquisition module 401 is used to acquire the charging curve time series dataset of the battery under test, identify the charging polarization inflection point of the battery, and acquire the electrochemical impedance spectroscopy data at the polarization inflection point as the trigger reference. The electrochemical impedance spectroscopy data is then preprocessed by denoising and operating condition normalization to obtain a standardized impedance spectroscopy dataset. The separation module 402 is used to perform polarization three-component decoupling processing on the charging curve time series dataset, separating the real-time change data of the battery's ohmic polarization, concentration polarization and electrochemical polarization, and obtaining the polarization three-component decoupling results; Extraction module 403 is used to perform relaxation time distribution deconvolution decoupling and equivalent circuit model fitting on the standardized impedance spectrum dataset based on the polarization three-component decoupling results, separate the battery impedance response, and extract the intrinsic electrochemical feature set specific to the lithium plating reaction. The model building module 404 is used to extract the charging polarization feature set related to the critical behavior of lithium plating from the real-time change data of the separated polarization three components, construct the electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, and generate the lithium plating identification coupling feature matrix. The first output module 405 is used to input the lithium plating identification coupling feature matrix into the lithium plating determination threshold system and output the lithium plating state determination result. The second output module 406 is used to match the lithium fault level of the test battery that is determined to have undergone lithium plating based on the coupling feature matrix and the lithium plating degree calibration database, quantitatively analyze the lithium fault level, and output the lithium plating fault detection result.

[0162] In some alternative implementations, the acquisition module 401 includes: The moving average filtering processing unit is used to perform moving average filtering processing on the charging curve time series dataset to obtain the terminal voltage time series curve. The first calculation unit is used to calculate the first and second derivatives of the terminal voltage time-series curve. The first derivative is used to characterize the polarization growth rate during the battery charging process, and the second derivative is used to characterize the changing trend of the polarization growth rate. The determination unit is used to determine the current sampling point as the polarization inflection point if the number of sampling points where the absolute value of the second derivative continuously exceeds the preset mutation threshold reaches a preset number, and the relative error of the terminal voltage time series curve deviating from the linear fitting region is greater than or equal to the error threshold.

[0163] In some alternative embodiments, the device further includes: The noise reduction module is used to perform secondary noise reduction processing on the charging curve time series dataset using wavelet adaptive filtering; The first calibration module is used to perform reference calibration on the terminal voltage during the charging start and end resting stages, based on the battery open-circuit voltage and state of charge calibration curve. The second calibration module is used to perform linear calibration on the synchronously acquired current data using the calibration coefficient of the charge-discharge tester, to perform deviation correction on the synchronously acquired temperature data using the standard temperature of the constant temperature chamber, and to perform closed-loop correction on the synchronously acquired state-of-charge time-series data using the ampere-hour integral method and the open-circuit voltage and state-of-charge curve, so as to obtain the corrected full-cycle charging terminal voltage, current, temperature, and state-of-charge time-series dataset.

[0164] In some alternative implementations, the separation module 402 includes: The second computing unit is used to calculate the total polarization voltage based on the corrected time-series dataset; The linear interpolation unit is used to perform a current interruption operation at preset intervals during the constant current charging stage, collect the voltage change after the current interruption, determine the voltage change as the ohmic polarization voltage drop, and determine the linearly interpolated timing data as the ohmic polarization component. The third calculation unit is used to calculate the electrochemical polarization voltage drop based on the Butler-Volmer equation; The fourth calculation unit is used to calculate the concentration polarization pressure drop based on Fick's first diffusion law; The identification unit is used to identify the exchange current density and the lithium ion concentration on the solid surface in real time using the corrected time-series dataset as input, and to obtain real-time change data of electrochemical polarization and concentration polarization.

[0165] In some alternative implementations, the extraction module 403 includes: Building units are used to construct the basic model of relaxation time distribution; The identification unit is used to solve the relaxation time distribution function in the basic model of relaxation time distribution, draw the relaxation time distribution spectrum, and identify the relaxation peaks in the relaxation time distribution spectrum. The fitting unit is used to combine the relaxation peak to construct an equivalent circuit model corresponding to the electrochemical process, and the equivalent circuit model is fitted using the complex nonlinear least squares method.

[0166] In some alternative implementations, the model building module 404 includes: The rule-establishing unit is used to establish rules for evaluating feature correlation. The first extraction unit is used to extract the relative change rate of ohmic impedance, the proportion of ohmic polarization voltage to total polarization voltage, and the temperature sensitivity coefficient of ohmic impedance at the polarization inflection point, as relevant features of ohmic polarization components. The second extraction unit is used to extract the electrochemical polarization voltage abrupt change rate at the polarization inflection point, the proportion of electrochemical polarization voltage to the total polarization voltage, and the polarization response time constant of the charge transfer process, as relevant features of the electrochemical polarization components. The third extraction unit is used to extract the polarization voltage growth rate and the proportion of concentration polarization voltage to total polarization voltage at the polarization inflection point, as relevant features of concentration polarization components. The classification unit is used to classify the extracted ohmic polarization component-related features, electrochemical polarization component-related features, and concentration polarization component-related features according to the polarization component to obtain the charging polarization feature set.

[0167] In some alternative implementations, the model building module 404 further includes: The preprocessing unit is used to perform simultaneous standardization and dimension adaptation processing on the charging polarization feature set and the intrinsic electrochemical feature set; The mapping model construction unit is used to construct an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set after synchronous standardization and dimensional adaptation, according to the weight allocation rule. The weighted fusion unit is used to substitute the intrinsic electrochemical feature vector of lithium plating and the feature vector of charging polarization into the electrochemical mechanism coupling mapping model, and perform weighted fusion using a weight matrix to obtain the coupling feature vector. The matrix generation unit is used to reconstruct the coupling feature vectors in a preset order to generate a lithium plating identification coupling feature matrix.

[0168] In some alternative implementations, the first output module 405 includes: The compliance verification unit is used to perform compliance verification on the lithium plating identification coupling feature matrix; The dimension matching unit is used to perform dimension matching on the feature matrix after compliance verification to obtain the threshold matching result; The first output unit is used to match the threshold matching result with the judgment rule and output a lithium plating status judgment report. The lithium plating status includes no lithium plating, critical lithium plating, and lithium plating has occurred.

[0169] In some alternative implementations, the second output module 406 includes: The similarity calculation unit is used to calculate the similarity between the coupling feature matrix and the feature matrix of the lithium-plated calibrated sample; The weighted average unit is used to select benchmark samples that meet the similarity criteria. The selected benchmark samples are weighted and averaged using similarity as the weight to obtain a quantitative assessment value of the lithium plating degree of the battery under test. The second output unit is used to match the quantitative assessment value of the degree of lithium plating with the grading standard to obtain the lithium plating failure level.

[0170] The battery lithium plating fault detection device based on charging curves and impedance spectra provided in this invention can execute the battery lithium plating fault detection method based on charging curves and impedance spectra provided in any embodiment of this invention, and has the corresponding functional modules and beneficial effects of the method. Further functional descriptions of the above modules and units are the same as in the corresponding embodiments described above, and will not be repeated here.

[0171] Figure 5 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present invention.

[0172] The following is a detailed reference. Figure 5 The diagram illustrates a structural schematic suitable for implementing an electronic device according to embodiments of the present invention. The electronic device may include a processor (e.g., a central processing unit, graphics processor, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from memory 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device. The processor 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0173] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; memory devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown, and more or fewer devices may be implemented or have instead.

[0174] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a memory 508, or installed from a ROM 502. When the computer program is executed by the processor 501, it performs the functions defined in the battery lithium plating fault detection method based on charging curves and impedance spectra of the embodiments of the present invention.

[0175] Figure 5 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0176] This invention also provides a computer-readable storage medium. The methods described above according to embodiments of the invention can be implemented in hardware or firmware, or implemented as computer code that can be recorded on a storage medium, or implemented as computer code downloaded via a network and originally stored on a remote storage medium or a non-transitory machine-readable storage medium and then stored on a local storage medium. Thus, the methods described herein can be processed by software stored on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. The storage medium can be a magnetic disk, optical disk, read-only memory, random access memory, flash memory, hard disk, or solid-state drive, etc.; further, the storage medium can also include combinations of the above types of memory. It is understood that the computer, processor, microprocessor controller, or programmable hardware includes storage components capable of storing or receiving software or computer code. When the software or computer code is accessed and executed by the computer, processor, or hardware, the battery lithium plating fault detection method based on charging curves and impedance spectra shown in the above embodiments is implemented.

[0177] A portion of this invention can be applied as a computer program product, such as computer program instructions, which, when executed by a computer, can invoke or provide the methods and / or technical solutions according to the invention through the operation of the computer. Those skilled in the art will understand that the forms in which computer program instructions exist in a computer-readable medium include, but are not limited to, source files, executable files, installation package files, etc. Correspondingly, the ways in which computer program instructions are executed by a computer include, but are not limited to: the computer directly executing the instructions, or the computer compiling the instructions and then executing the corresponding compiled program, or the computer reading and executing the instructions, or the computer reading and installing the instructions and then executing the corresponding installed program. Here, the computer-readable medium can be any available computer-readable storage medium or communication medium accessible to a computer.

[0178] Although embodiments of the invention have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of the invention, and all such modifications and variations fall within the scope defined by the appended invention.

Claims

1. A battery lithium plating fault detection method based on charging curves and impedance spectra, characterized in that, The method includes: Obtain the charging curve time series dataset of the battery under test, identify the battery charging polarization inflection point, use the polarization inflection point as the trigger reference, collect the electrochemical impedance spectroscopy data at the polarization inflection point, and perform noise reduction and operating condition normalization preprocessing on the electrochemical impedance spectroscopy data to obtain a standardized impedance spectroscopy dataset. The charging curve time series dataset is subjected to polarization three-component decoupling processing to separate the real-time change data of ohmic polarization, concentration polarization and electrochemical polarization of the battery, and obtain the polarization three-component decoupling result; Based on the polarization three-component decoupling results, the standardized impedance spectrum dataset is subjected to relaxation time distribution deconvolution decoupling and equivalent circuit model fitting to separate the battery impedance response and extract the intrinsic electrochemical feature set specific to the lithium plating reaction. From the real-time change data of the three polarization components obtained by separation, the charging polarization feature set related to the critical behavior of lithium plating is extracted, and an electrochemical mechanism coupling mapping model of the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating is constructed to generate the lithium plating identification coupling feature matrix. The lithium plating identification coupling feature matrix is ​​input into the lithium plating determination threshold system, and the lithium plating state determination result is output. For batteries under test that are determined to have undergone lithium plating, the lithium fault level is quantitatively analyzed by matching the coupling feature matrix with the lithium plating degree calibration database, and the lithium plating fault detection results are output.

2. The method according to claim 1, characterized in that, The identification of the battery charging polarization inflection point includes: The charging curve time series dataset is subjected to moving average filtering to obtain the terminal voltage time series curve; Calculate the first and second derivatives of the terminal voltage time-series curve. The first derivative is used to characterize the polarization growth rate during battery charging, and the second derivative is used to characterize the changing trend of the polarization growth rate. If the number of sampling points where the absolute value of the second derivative continuously exceeds the preset abrupt change threshold reaches the preset number, and the relative error of the terminal voltage time series curve deviating from the linear fitting region is greater than or equal to the error threshold, then the sampling point at the current moment is determined to be the polarization inflection point.

3. The method according to claim 1, characterized in that, Before performing polarization three-component decoupling processing on the charging curve time-series dataset to separate the real-time changes in the battery's ohmic polarization, concentration polarization, and electrochemical polarization, the method further includes: The charging curve time-series dataset is subjected to secondary noise reduction processing using wavelet adaptive filtering. Using the battery open-circuit voltage and state of charge calibration curve as a reference, the terminal voltage during the charging start and end resting stages is corrected. The synchronously acquired current data is linearly corrected using the calibration coefficient of the charge-discharge tester, the synchronously acquired temperature data is corrected for deviation using the standard temperature of the constant temperature chamber, and the synchronously acquired state-of-charge time series data is corrected using the ampere-hour integral method and the open-circuit voltage and state-of-charge curve, resulting in a corrected full-cycle charging terminal voltage, current, temperature, and state-of-charge time series dataset.

4. The method according to claim 3, characterized in that, The process of decoupling the three polarization components of the charging curve time-series dataset to separate the real-time changes in the battery's ohmic polarization, concentration polarization, and electrochemical polarization includes: The total polarization voltage is calculated based on the corrected time-series dataset; During the constant current charging phase, a current interruption operation is performed every preset time interval. The voltage change after the current interruption is collected, and the voltage change is determined as the ohmic polarization voltage drop. The linearly interpolated time series data is determined as the ohmic polarization component. Calculation of electrochemical polarization voltage drop based on the Butler-Volmer equation; Calculation of concentration polarization pressure drop based on Fick's first diffusion law; Using the corrected time-series dataset as input, the recursive least squares method is employed to identify the exchange current density and the lithium-ion concentration on the solid surface in real time, thereby obtaining real-time changes in electrochemical polarization and concentration polarization.

5. The method according to claim 1, characterized in that, The step of performing relaxation time distribution deconvolution decoupling and equivalent circuit model fitting on the standardized impedance spectrum dataset based on the polarization three-component decoupling results includes: Construct a basic model of relaxation time distribution; Solve for the relaxation time distribution function in the basic model of relaxation time distribution, plot the relaxation time distribution spectrum, and identify the relaxation peaks in the relaxation time distribution spectrum; Based on the relaxation peak, an equivalent circuit model corresponding to the electrochemical process is constructed, and the equivalent circuit model is fitted using the complex nonlinear least squares method.

6. The method according to claim 1, characterized in that, The extracted charging polarization feature set related to the critical behavior of lithium plating includes: Establish rules for evaluating feature correlation. The relative change rate of ohmic impedance at the polarization inflection point, the proportion of ohmic polarization voltage to the total polarization voltage, and the temperature sensitivity coefficient of ohmic impedance are extracted as relevant characteristics of the ohmic polarization component. The electrochemical polarization voltage abrupt change rate at the polarization inflection point, the proportion of electrochemical polarization voltage to the total polarization voltage, and the polarization response time constant of the charge transfer process are extracted as relevant characteristics of the electrochemical polarization components. The polarization voltage growth rate and the proportion of concentration polarization voltage to total polarization voltage at the polarization inflection point are extracted as relevant features of the concentration polarization component. The extracted ohmic polarization component-related features, electrochemical polarization component-related features, and concentration polarization component-related features are classified according to the polarization components to obtain the charging polarization feature set.

7. The method according to claim 1, characterized in that, The electrochemical mechanism coupling mapping model for constructing the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating, and generating the lithium plating identification coupling feature matrix, includes: Simultaneous standardization and dimension adaptation processing are performed on the charging polarization feature set and the intrinsic electrochemical feature set; Based on the charging polarization feature set and intrinsic electrochemical feature set after synchronous standardization and dimensional adaptation, an electrochemical mechanism coupling mapping model between the charging polarization feature set and the intrinsic electrochemical feature set of lithium plating is constructed according to the weight allocation rule. Substitute the intrinsic electrochemical eigenvectors of lithium plating and the charging polarization eigenvectors into the electrochemical mechanism coupling mapping model, and use a weight matrix to perform weighted fusion to obtain the coupling eigenvectors. The coupling feature vectors are reconstructed in a preset order to generate a lithium plating identification coupling feature matrix.

8. The method according to claim 1, characterized in that, The step of inputting the lithium plating identification coupling feature matrix into the lithium plating determination threshold system and outputting the lithium plating state determination result includes: Compliance verification of the lithium plating identification coupling feature matrix; Dimension matching is performed on the feature matrix after compliance verification to obtain threshold matching results; The threshold matching results are matched with the judgment rules to output a lithium plating status judgment report, which includes no lithium plating, critical lithium plating, and lithium plating has occurred.

9. The method according to claim 1, characterized in that, The process involves matching the coupled feature matrix with a lithium plating degree calibration database to quantitatively classify and analyze lithium fault levels, and outputting lithium plating fault detection results, including: Calculate the similarity between the coupling feature matrix and the feature matrix of the lithium-plated calibrated sample; By selecting benchmark samples that meet the similarity criteria, and using similarity as the weight, a weighted average is calculated on the selected benchmark samples to obtain a quantitative assessment value of the lithium plating degree of the battery under test. The lithium plating degree quantitative assessment value is matched with the grading standard to obtain the lithium plating failure level.

10. An electronic device, characterized in that, include: The system includes a memory and a processor, which are interconnected. The memory stores computer instructions, and the processor executes the computer instructions to perform the battery lithium plating fault detection method based on charging curves and impedance spectra as described in any one of claims 1 to 9.

Images