A lithium ion battery capacity sorting process anomaly diagnosis method and system
Patent Information
- Application Number
- CN202610805373.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-05
- Publication Date
- 2026-09-18
AI Technical Summary
由于恒流、恒压充电阶段物理驱动模式与电池响应机制存在本质差异,恒流充电末期积累的极化效应会产生残留极化电流,与电池本征充电电流叠加,造成恒压充电初期电信号响应混杂
[0009]In summary, this application provides a method and system for diagnosing abnormalities during the capacity grading process of lithium-ion batteries. By collecting the electrical signal response sequence from the end of constant current charging to the beginning of constant voltage charging, and extracting the constant current response information and the mixed response information superimposed by multiple components in a layered manner, it overcomes the shortcomings of traditional battery capacity grading detection, which cannot distinguish between polarization response and intrinsic charging response and has a single dimension of state identification. By designing signal separation, dual response component parameter solving, and characteristic parameter benchmark deviation measurement for the mixed response information, it avoids the shortcomings of traditional detection methods, which are difficult to accurately distinguish between charging polarization abnormalities and charging acceptance abnormalities and have low defect identification accuracy. It can effectively identify two different charging abnormal conditions during the battery capacity grading process, accurately identify subtle performance deviations in the battery charging stage, ensure the accuracy and pertinence of battery capacity grading anomaly judgment, and effectively improve the accuracy of battery capacity grading detection and the reliability of cell performance screening.
Smart Images

Figure CN122776084A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of battery fault diagnosis technology, and more specifically, to a method and system for diagnosing abnormalities in the capacity grading process of lithium-ion batteries. Background Technology
[0002] In the mass production process of lithium-ion batteries, the capacity grading process is a core and critical step in screening cell performance, calibrating battery capacity, and determining the quality of cells, directly determining the battery's final quality and batch consistency. The battery capacity grading process mainly involves collecting electrical signal data of the battery throughout the entire charging and discharging process through standardized procedures to complete cell capacity calculation and performance status determination. Among these steps, the state transition from constant current charging to constant voltage charging is the most typical and critical condition conversion node in the battery capacity grading process.
[0003] In a standard capacity grading process, the battery first enters a constant current charging phase, where a constant current continuously charges the battery and drives its voltage to rise steadily. Once the battery voltage reaches a preset threshold, the system automatically switches to a constant voltage charging phase, using a constant voltage to constrain the battery's charging state, causing the charging current to gradually decrease as the battery polarization saturates. Existing capacity grading testing solutions typically rely on overall voltage and current data from the complete charge and discharge process, combined with fixed thresholds and general statistical models, to perform batch judgments on conventional parameters such as battery capacity and charging stability, achieving preliminary screening of good and defective cells.
[0004] In conventional testing processes, existing judgment methods have significant inherent technical limitations. Their core logic assumes that the battery's state is continuous and stable across each charging stage, with no residual interference during stage transitions, neglecting the unique operating conditions at the critical node of constant current charging transitioning to constant voltage charging. Due to the fundamental differences between the physical driving modes and battery response mechanisms of constant current and constant voltage charging stages, the polarization effect accumulated at the end of constant current charging generates residual polarization current, which superimposes on the battery's intrinsic charging current, resulting in a mixed electrical signal response at the beginning of constant voltage charging. Current detection technologies cannot effectively separate the two current components for this mixed signal, making it difficult to distinguish whether signal fluctuations originate from abnormal cell polarization or abnormal battery charging acceptance. They can only rely on overall signal deviation to make a general judgment of cell status, failing to accurately classify the type of abnormality. This easily leads to missed detection of cells with hidden defects such as excessive residual polarization or diminished charging acceptance, causing defective products to flow into subsequent processes. It is also susceptible to polarization interference, misjudging normal cells as abnormal, causing incorrect rejection of good products, significantly reducing the accuracy of capacity testing and production efficiency. Meanwhile, traditional testing solutions only use a single, fixed overall characteristic parameter to determine the state, without setting differentiated testing logic for the switching conditions between charging and discharging stages, and without considering the temporal impact of the continuity of charging state and polarization residue. They cannot achieve abnormal timing location and mechanism tracing, and have poor ability to identify subtle stage performance deviations during battery charging. This not only makes it difficult to meet the high-precision and high-consistency mass production testing requirements of power batteries, but also lacks good technical scalability and cannot support subsequent in-depth applications such as refined cell state analysis, defect tracing, and production process optimization.
[0005] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention
[0006] To address the shortcomings of existing technologies, this application aims to provide a method and system for diagnosing abnormalities during the capacity grading process of lithium-ion batteries. This method can accurately separate the mixed current response components in the initial stage of constant voltage charging, distinguish between two types of defects: abnormal battery polarization and abnormal charging acceptance performance. It can effectively identify latent performance deviations in battery cells, improve the accuracy of battery capacity grading abnormality diagnosis, and enhance the batch consistency control of batteries.
[0007] Firstly, this application provides a method for diagnosing abnormalities during the capacity grading process of a lithium-ion battery. This method is used to diagnose abnormalities when the battery transitions from a constant current charging stage to a constant voltage charging stage during the capacity grading process. The method includes: Obtain the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging; The constant current response information at the end of the constant current charging stage and the mixed response information at the beginning of the constant voltage charging stage are extracted from the electrical signal response sequence. The mixed response information is at least composed of a first response component characterizing the residual polarization at the end of the constant current charging stage and a second response component characterizing the inherent charging characteristics at the beginning of the constant voltage charging stage. The mixed response information is processed by signal separation to obtain the first response component and the second response component. Then, the parameters of the separated first response component and the second response component are solved according to the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component. The first set of characteristic parameters is a set of parameters characterizing the polarization release current characteristics, and the second set of characteristic parameters is a set of parameters characterizing the intrinsic charging current characteristics. The deviation of the first set of feature parameters from a preset first benchmark and the deviation of the second set of feature parameters from a preset second benchmark are calculated respectively. The abnormal conditions in the battery capacity grading process are determined based on the calculation results of the deviation degree.
[0008] Secondly, this application provides a lithium-ion battery capacity grading process anomaly diagnosis system for performing the aforementioned lithium-ion battery capacity grading process anomaly diagnosis method. The system includes: The acquisition module is used to acquire the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging. The data extraction module is used to extract the constant current response information at the end of the constant current charging stage and the mixed response information at the beginning of the constant voltage charging stage based on the electrical signal response sequence. The mixed response information is at least composed of a first response component characterizing the residual polarization at the end of the constant current charging stage and a second response component characterizing the inherent charging characteristics at the beginning of the constant voltage charging stage. The parameter solving module is used to perform signal separation processing on the mixed response information to obtain the first response component and the second response component, and to perform parameter solving on the mixed response information according to the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component; wherein, the first set of characteristic parameters is a set of parameters characterizing the polarization release current characteristics, and the second set of characteristic parameters is a set of parameters characterizing the intrinsic charging current characteristics; The deviation calculation module is used to calculate the degree of deviation of the first feature parameter set relative to a preset first benchmark, and the degree of deviation of the second feature parameter set relative to a preset second benchmark, respectively. An anomaly detection module is used to determine abnormal conditions in the battery capacity testing process based on the calculation results of the deviation degree.
[0009] In summary, this application provides a method and system for diagnosing abnormalities during the capacity grading process of lithium-ion batteries. By collecting the electrical signal response sequence from the end of constant current charging to the beginning of constant voltage charging, and extracting the constant current response information and the mixed response information superimposed by multiple components in a layered manner, it overcomes the shortcomings of traditional battery capacity grading detection, which cannot distinguish between polarization response and intrinsic charging response and has a single dimension of state identification. By designing signal separation, dual response component parameter solving, and characteristic parameter benchmark deviation measurement for the mixed response information, it avoids the shortcomings of traditional detection methods, which are difficult to accurately distinguish between charging polarization abnormalities and charging acceptance abnormalities and have low defect identification accuracy. It can effectively identify two different charging abnormal conditions during the battery capacity grading process, accurately identify subtle performance deviations in the battery charging stage, ensure the accuracy and pertinence of battery capacity grading anomaly judgment, and effectively improve the accuracy of battery capacity grading detection and the reliability of cell performance screening. Attached Figure Description
[0010] Figure 1 This is a flowchart illustrating an abnormality diagnosis method for the capacity grading process of a lithium-ion battery, provided in an embodiment of this application.
[0011] Figure 2 This is a schematic diagram of a lithium-ion battery capacity grading process abnormal diagnosis system provided in an embodiment of this application.
[0012] Labeling Explanation: 1. Acquisition Module; 2. Data Extraction Module; 3. Parameter Solving Module; 4. Deviation Calculation Module; 5. Anomaly Detection Module. Detailed Implementation
[0013] The technical solutions of this application will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are merely some embodiments of this application, and not all embodiments. The components of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely to illustrate selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.
[0014] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.
[0015] refer to Figure 1This application proposes a method for abnormal diagnosis during the capacity grading process of lithium-ion batteries. This method is used to diagnose abnormalities when the battery transitions from a constant current charging stage to a constant voltage charging stage during the capacity grading process. The method includes: A1. Obtain the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging; A2. Extract the constant current response information of the end of the constant current charging and the mixed response information of the beginning of the constant voltage charging based on the electrical signal response sequence. The mixed response information is at least composed of a first response component characterizing the residual polarization of the end of the constant current charging and a second response component characterizing the inherent charging characteristics of the beginning of the constant voltage charging. A3. Perform signal separation processing on the mixed response information to obtain the first response component and the second response component, and solve the parameters of the separated first response component and the second response component according to the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component; wherein, the first set of characteristic parameters is a set of parameters characterizing the polarization release current characteristics, and the second set of characteristic parameters is a set of parameters characterizing the intrinsic charging current characteristics; A4. Calculate the degree of deviation of the first feature parameter set from the preset first benchmark, and the degree of deviation of the second feature parameter set from the preset second benchmark, respectively. A5. Determine the abnormal conditions in the battery capacity grading process based on the calculation results of the deviation degree.
[0016] In this application, the electrical signal response sequence refers to the timing data obtained during the period when the battery switches from constant current charging to constant voltage charging. It includes at least the timing data of the battery terminal voltage and the timing data of the loop current, and records the electrical response process during the switching of the battery charging excitation mode. Specifically, high-frequency sampling equipment, such as a data acquisition card combined with a high-precision sensor, can be used to ensure that the acquired data can accurately reflect the electrical signal characteristics of the battery during this stage.
[0017] Constant current response information refers to the characteristic information that reflects the battery state at the end of constant current charging, which can be represented by the battery voltage change rate and internal resistance information. The voltage change rate reflects the trend of battery voltage change over time at the end of constant current charging, while the internal resistance information reflects the impedance characteristics inside the battery.
[0018] Hybrid response information refers to the current response characteristics that reflect the battery state in the initial stage of constant voltage charging, which can be represented by the corresponding current response curve for this stage. Physically, it is the superposition of two processes: first, the polarization effect accumulated during the constant current charging stage begins to relax and release, generating a decaying polarization current, i.e., the first response component; second, the battery continues to charge under a constant voltage, generating an intrinsic charging current that matches its own charging acceptance capability, i.e., the second response component.
[0019] The first set of characteristic parameters refers to the set of parameters that characterize the polarization release current. Examples include the polarization current decay rate and the steady-state value of the polarization current.
[0020] The second set of characteristic parameters refers to the set of parameters that characterize the intrinsic charging current characteristics. Examples include the initial value of the charging current during the constant voltage period and the current decay coefficient.
[0021] The preset first benchmark refers to the standard reference data used to evaluate the polarization release current characteristics, which are obtained by statistically analyzing normal batteries of the same model and specification under standard charging conditions. Examples include: standard polarization current decay coefficient, standard polarization release peak current, standard polarization dissipation time, and standard residual polarization amplitude.
[0022] The preset second benchmark refers to the reference data used to evaluate the intrinsic charging current characteristics of the battery, which is obtained by collecting and statistically analyzing qualified batteries of the same model under standard charging conditions. Examples include: standard intrinsic charging current value, standard current change rate, standard intrinsic charging response amplitude, and standard cell charging steady-state parameters.
[0023] The degree of deviation refers to the magnitude of the deviation between the current polarization release current characteristics and intrinsic charging current characteristics of the battery and the corresponding electrical characteristics of a normal, qualified battery.
[0024] Abnormal conditions during the battery capacity grading process refer to the deviation of the battery's signal response characteristics from the normal batch standard at the end of constant current charging and the beginning of constant voltage charging. Specifically, they are divided into two categories: abnormal polarization response at the end of constant current charging and abnormal charging acceptance response at the beginning of constant voltage charging.
[0025] The electrical signal response sequence of the battery during the transition from the end of constant current charging to the beginning of constant voltage charging is obtained. Specifically, the transition period from the end of constant current charging to the beginning of constant voltage charging is first extracted from the complete battery charging process, and then various electrical signals generated by the battery during this stage are acquired in real time using an electrical signal acquisition device. Specifically, this can be achieved in the following ways: One method is to measure the voltage across the battery terminals and compare it with a preset cutoff voltage. Because the battery's terminal voltage gradually approaches the cutoff voltage at the end of constant current charging, and when the terminal voltage reaches the cutoff voltage, the battery switches from constant current charging mode to constant voltage charging mode. The comparison between the battery's terminal voltage and the preset cutoff voltage can determine whether the battery is within this transition period. Another method is to monitor changes in the circuit current and compare them with preset current thresholds and drop amplitudes. Because the circuit current remains constant at the end of constant current charging, but a slight drop trend occurs near the change in operating conditions, the comparison between monitoring changes in the circuit current and preset constant current thresholds and drop amplitudes can determine whether the battery is within this transition period. If the battery circuit current drops beyond a preset amplitude, this solution can lock the end of constant current charging and initiate electrical signal acquisition. The electrical signal acquisition device in this solution can be any tool with data acquisition capabilities, such as a battery testing system, data acquisition card, high-precision multimeter, current and voltage acquisition module, etc.
[0026] This method extracts constant current response information at the end of constant current charging and mixed response information at the beginning of constant voltage charging based on the electrical signal response sequence. Specifically, it identifies and separates the battery response characteristics at the end of constant current charging and the mixed response characteristics formed by the superposition of multiple responses at the beginning of constant voltage charging from the acquired electrical signal response sequence. Specifically, this can be achieved by using methods such as voltage threshold mutation judgment, wavelet decomposition algorithm, or Kalman filtering to first divide the signal intervals at the end of constant current charging and the beginning of constant voltage charging. The constant current response information can be extracted from the current stability parameters, voltage fluctuation amplitude, and other features within this interval. The mixed response information can be extracted by using a signal decomposition algorithm to remove the superposition effect of polarization release current and intrinsic charging current, and then extracting the feature parameters of the two types of responses separately. This information extraction method can initially distinguish the battery characteristics at different stages, laying the foundation for subsequent refined analysis.
[0027] The mixed response information is processed by signal separation to obtain a first response component and a second response component. Then, based on the constant current response information, the separated first and second response components are solved for parameters, yielding a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component. Specifically, the mixed response information can be decomposed into a superposition of polarization release current and intrinsic charging current using a signal decomposition algorithm. This is then combined with information related to the polarization release current at the end of the constant current charging phase to obtain the first and second sets of characteristic parameters. Specifically, wavelet packet decomposition or Kalman filtering can be used to separate the mixed response information into a first response component characterizing the polarization release current and a second response component characterizing the intrinsic charging current. Using the constant current response information as a baseline constraint for parameter solving, the first set of characteristic parameters corresponding to the first response component and the second set of characteristic parameters corresponding to the second response component are solved through data fitting, trend extrapolation, or error calibration. Through signal separation and parameter solving, the complex mixed signal is decomposed into characteristic parameters with clear physical meaning. These parameters can quantify the polarization and intrinsic charging characteristics of the battery.
[0028] The deviation of the first set of characteristic parameters from a preset first benchmark and the deviation of the second set of characteristic parameters from a preset second benchmark are calculated respectively. Specifically, multiple groups of standard batteries of the same model and batch, in a healthy and normal state, are selected. Under the same charging conditions, ambient temperature, and charging rate, the data of the entire constant current to constant voltage charging process is collected sequentially. The first benchmark related to the polarization release current and the second benchmark related to the intrinsic charging current of each group of standard batteries are extracted. The deviation can be calculated in the following ways: one way is to calculate the absolute difference between each parameter in the first set of characteristic parameters and the preset first benchmark, and take the average or maximum value of all absolute differences as the deviation of the first set of characteristic parameters from the corresponding parameter of the first benchmark; similarly, the deviation of the second set of characteristic parameters from the second benchmark is calculated. Another way is to calculate the relative deviation between each parameter in the first set of characteristic parameters and the corresponding parameter of the first benchmark, and take the average or maximum value of all relative deviations as the deviation of the first set of characteristic parameters from the first benchmark; similarly, the deviation of the second set of characteristic parameters from the second benchmark is calculated.
[0029] The abnormal conditions in the battery capacity testing process are determined based on the calculation results of the deviation degree. Specifically, based on the deviation degree, it is determined whether there are any abnormalities in the battery during the capacity testing process, and the type and cause of the abnormality can be further identified, thereby achieving effective monitoring and diagnosis of the battery capacity testing process.
[0030] Through the above technical solution, this application can solve the problem of abnormal diagnosis during the transition from constant current charging to constant voltage charging in the capacity grading process of lithium-ion batteries. This method, through refined analysis of the battery's electrical signal response from the end of constant current charging to the beginning of constant voltage charging, can effectively distinguish between the battery's polarization characteristics and intrinsic charging characteristics, and quantify their deviation degree, thereby achieving accurate judgment of abnormal conditions during the battery capacity grading process.
[0031] In some preferred embodiments, the step of obtaining the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging includes: During the final stage of constant current charging, the difference between the battery terminal voltage and the preset cutoff voltage threshold is continuously monitored. When the difference is within the preset sampling range, the preset sampling frequency is used to collect the battery terminal voltage and loop current during the period covering the end of constant current charging, the switching node between constant current and constant voltage charging, and the initial stage of constant voltage charging. The collected battery terminal voltage sequence and loop current sequence are used as electrical signal response sequences.
[0032] In this application, the preset acquisition interval refers to a pre-set difference range. When the difference between the battery terminal voltage and the cutoff voltage threshold falls within this range, the electrical signal during the battery charging process will be acquired.
[0033] The system continuously monitors the difference between the battery's terminal voltage and a preset cutoff voltage threshold. Specifically, during constant current charging, the battery's terminal voltage gradually increases. In the final stage of constant current charging, the voltage across the battery is measured in real time and compared to the cutoff voltage threshold, which is the battery manufacturer's specified upper limit for charging. This process allows for real-time monitoring of the battery's charging status, providing a basis for subsequent accurate data acquisition.
[0034] When the difference falls within the preset sampling interval, a preset sampling frequency is used to collect battery terminal voltage and loop current during the period covering the end of constant current charging, the transition node from constant current to constant voltage charging, and the initial stage of constant voltage charging. Specifically, the preset sampling interval is designed to ensure data collection occurs during the transition from constant current charging to constant voltage charging, avoiding inaccurate data due to premature or late collection. From the end of constant current charging to the beginning of constant voltage charging, both battery terminal voltage and current flowing through the battery are collected simultaneously. Based on the battery's rated capacity and standard charging rate, the standard transition time for the corresponding stage is obtained from a table and used as the sampling duration. For example, when the difference is greater than 0.05V and less than 0.1V, the sampling interval is entered, and a sampling frequency of 100Hz and a sampling duration of 60 seconds can be used to collect battery terminal voltage and loop current.
[0035] The collected battery terminal voltage sequence and loop current sequence are used as electrical signal response sequences. Specifically, the collected voltage and current data points are organized into time series, and these two sequences are combined. The collected data can be saved as timestamps, forming two ordered arrays.
[0036] Through the above technical solution, this application can obtain the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging more specifically and accurately, thereby improving the accuracy and reliability of abnormal diagnosis of lithium-ion battery capacity grading process.
[0037] In some preferred embodiments, after the step of obtaining the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging, the method includes: Outlier removal is performed on the electrical signal response sequence to remove singular outliers and obtain a preliminary screening sequence. The initial screening sequence was smoothed and denoised to obtain the optimized electrical signal response sequence.
[0038] In this application, outlier removal refers to removing data points that deviate from the normal range in an electrical signal response sequence. This can be achieved using statistical methods, such as the 3σ criterion based on standard deviation, box plots, Z-scores, or machine learning-based anomaly detection algorithms. For example, the 3σ criterion considers a data point deviating from the mean by more than three times the standard deviation as an outlier.
[0039] Outliers are data points in a dataset that significantly deviate from the normal range. These points may be caused by measurement errors, sensor malfunctions, or transient interference.
[0040] The initial screening sequence refers to the electrical signal response sequence after outlier removal, which has eliminated singular outliers. Although this sequence removes obvious erroneous data, it may still contain random noise.
[0041] Smoothing and noise reduction refers to processing the initially screened sequence to reduce or eliminate random noise, making the data smoother. This can be achieved using digital filtering techniques such as moving average filtering, Savitzky-Golay filtering, wavelet transform denoising, or Kalman filtering. For example, moving average filtering smooths the data by calculating the average value of the data points' neighborhoods.
[0042] The optimized electrical signal response sequence refers to an electrical signal response sequence whose data quality has been significantly improved after outlier removal and smoothing / denoising. This sequence has lower noise and fewer outliers, and can more accurately reflect the true electrical signal response characteristics of the battery.
[0043] Outlier removal and smoothing / denoising in electrical signal response sequences are interconnected. Outlier removal first eliminates extreme errors in the data, providing a cleaner data foundation for subsequent denoising. Smoothing / denoising, building upon outlier removal, further filters out random noise, making the data more realistic and reliable. This process ensures improved data quality from raw acquisition to final optimized data, providing high-quality input data for subsequent extraction of constant current and mixed response information, signal separation, and parameter solving. This significantly improves the accuracy and reliability of anomaly diagnosis, addressing the impact of outliers and noise in the original electrical signal response sequence on diagnostic accuracy.
[0044] Through the above technical solution, this application addresses the impact of outliers and noise in electrical signal response sequences on the accuracy of anomaly diagnosis. Outlier removal from the electrical signal response sequence eliminates singular outliers that severely distort the true distribution of the data, affecting the accuracy of subsequent analysis. While the initial screening sequence obtained after outlier removal eliminates obvious erroneous data, it may still contain random noise. Therefore, further smoothing and denoising the initial screening sequence effectively filters out this random noise, making the electrical signal response sequence smoother and more realistic, thus obtaining an optimized electrical signal response sequence. This optimized sequence more accurately reflects the electrical signal response characteristics of the battery from the end of constant current charging to the beginning of constant voltage charging, providing high-quality input data for subsequent extraction of constant current and mixed response information, signal separation, and parameter solving, thereby improving the accuracy and reliability of anomaly diagnosis.
[0045] In some preferred embodiments, the step of performing signal separation processing on the mixed response information to obtain a first response component and a second response component includes: Using a preset decoupling model, the constant current response information is used as a constraint condition for the decoupling model. Based on this constraint condition, the mixed response information is decoupled and separated into components, and the first response component and the second response component are output.
[0046] In this application, the preset decoupling model refers to the mathematical model used to realize signal decomposition and analysis. A multi-exponential decay hybrid model can be selected, which can decompose the superimposed response data into multiple independent response components.
[0047] Specifically, characteristic parameters such as response amplitude, attenuation coefficient, and time constant are extracted from the constant current response information at the end of constant current charging. These characteristic parameters are then used as constraint thresholds and loaded into the decoupling model. During the solution process, a reasonable numerical benchmark range is defined using the constraint threshold as the core reference, limiting the parameters of the decomposed components to fall within this range. The mixed response information is iteratively fitted and separated according to the constrained model. After verifying that the component characteristics conform to the physical constraint standards, the first and second response components are output. For example, if the time constant threshold is 0.8s and the benchmark range is 0.75-0.85s, the time constant parameters of the first and second response components must fall within this benchmark range. By using the constant current response information at the end of constant current charging to extract characteristic parameters, the decoupling model is constrained and guided, making the signal separation process more closely resemble the actual physical process and improving the accuracy and robustness of the separation. Implementing component decoupling and separation of the mixed response information according to this constraint condition means that, under the guidance of the constant current response information, the decoupling model can more effectively decompose the mixed response information into a first response component characterizing the residual polarization at the end of constant current charging and a second response component characterizing the inherent charging characteristics at the beginning of constant voltage charging.
[0048] Through the above technical solution, this application solves the problem that inaccurate signal separation results affect the accuracy of subsequent parameter solving and anomaly diagnosis. By using a preset decoupling model and taking the constant current response information as a constraint condition for the decoupling model, the accuracy of mixed response information signal separation can be improved, thereby obtaining the first and second response components more accurately and providing a reliable data foundation for subsequent anomaly diagnosis.
[0049] In some preferred embodiments, the decoupling model is a multi-exponential decay hybrid mathematical model, the expression of which is: ; in, This refers to the cumulative charging time during the initial stage of constant voltage charging. This represents the total current during the initial stage of constant voltage charging. Represents the initial amplitude of polarization release. The time constant representing polarization release, The first response component is used to characterize the polarization release current. The initial amplitude representing the intrinsic charge component. The time constant representing intrinsic charge decay, This is the second response component, used to characterize the intrinsic charging current. This represents the steady-state leakage current component.
[0050] In this application, The timer can be started from the moment constant voltage charging begins, accumulating in seconds. A current sensor can be used to collect the total current value of the battery in real time during the initial stage of constant voltage charging. The solution can be obtained from the mixed response information using a nonlinear fitting algorithm, and its numerical range is typically in the milliampere range. The solution can be obtained from the mixed response information using a nonlinear fitting algorithm, and its numerical range is typically on the order of seconds. Based on the solution and Combined with cumulative charging time Calculations show that The solution can be obtained from the mixed response information using a nonlinear fitting algorithm, and its numerical range is typically in the milliampere range. The solution can be obtained from the mixed response information using a nonlinear fitting algorithm, and its numerical range is typically on the order of seconds. Based on the solution and Combined with cumulative charging time Calculations show that It can be obtained from the mixed response information using a nonlinear fitting algorithm, and its numerical range is usually in the microampere range.
[0051] Through the above technical solution, this application can accurately separate the first response component and the second response component in the mixed response information, thereby providing accurate input for subsequent anomaly diagnosis. It solves the problem that when performing signal separation processing, a suitable decoupling model is needed to accurately separate these components to ensure the accuracy of subsequent parameter solving, and thus more accurately diagnose the abnormal conditions of the battery capacity grading process.
[0052] In some preferred embodiments, the constant current response information includes the rate of change of voltage and internal resistance information of the battery at the end of constant current charging. The steps of solving for the parameters of the separated first and second response components based on the constant current response information to obtain the first set of characteristic parameters corresponding to the first response component and the second set of characteristic parameters corresponding to the second response component include: Voltage change rate and internal resistance information are used as parameter constraints for the multi-exponential decay hybrid mathematical model. Using the first and second response components as fitting samples, a nonlinear optimization algorithm is used to fit the parameters of the multi-exponential decay hybrid mathematical model, resulting in the first set of characteristic parameters corresponding to the first response component and the second set of characteristic parameters corresponding to the second response component.
[0053] In this application, voltage change rate and internal resistance information are used as parameter constraints for the multi-exponential decay hybrid mathematical model. Specifically, when fitting the parameters of the multi-exponential decay hybrid mathematical model, these constant current response information needs to be introduced into the model as constraints. These constraints can be reflected in the model's boundary conditions or parameter ranges to ensure that the model fitting results conform to the actual physical characteristics of the battery at the end of constant current charging. For example, reasonable value ranges for certain parameters in the model can be set based on the voltage change rate and internal resistance information, making the fitting results closer to reality.
[0054] The first and second response components are used as fitting samples. Specifically, the first and second response components obtained after signal separation processing are used as input data to train and optimize the multi-exponential decay hybrid mathematical model. This takes into account the actual physical characteristics of the battery at the end of constant current charging, so that the solution of the model parameters is more in line with the actual situation and avoids the errors that may be caused by blind fitting.
[0055] Nonlinear optimization algorithms are employed to fit the parameters of a multi-exponentially decaying mixture mathematical model. Specifically, nonlinear optimization algorithms are used to find the optimal parameters of the multi-exponentially decaying mixture mathematical model. These algorithms can include, but are not limited to, least squares, gradient descent, and the Levenberg-Marquardt algorithm. These algorithms iteratively calculate and continuously adjust the model parameters to minimize the error between the model output and the fitted sample, while simultaneously satisfying preset parameter constraints. In this way, the first set of characteristic parameters corresponding to the first response component and the second set of characteristic parameters corresponding to the second response component can be obtained.
[0056] Through the above technical solution, this application solves the problems of how to select a suitable decoupling model and how to use constant current response information to solve for the parameters of the separated components, ensuring the accuracy and reliability of the parameter solution, when performing signal separation processing on mixed response information. By introducing constant current response information as a parameter constraint condition for the multi-exponential decay mixed mathematical model and using a nonlinear optimization algorithm for parameter fitting, the solution accuracy and reliability of the first and second characteristic parameter sets are improved, thereby improving the reliability of anomaly diagnosis.
[0057] In some preferred embodiments, the method further includes a step of temperature compensation for the internal resistance information: The first ambient temperature at the end of constant current charging and the second ambient temperature at the beginning of constant voltage charging are obtained. Calculate the temperature difference between the first ambient temperature and the second ambient temperature; based on the temperature difference and the preset temperature compensation relationship, calculate the compensation factor corresponding to the internal resistance information; The internal resistance information is numerically corrected using a compensation factor.
[0058] In this application, the temperature compensation relationship refers to the matching correspondence between the temperature difference and the target control parameter (i.e., internal resistance information). Based on this relationship, the target control parameter is adjusted according to the temperature change to achieve temperature adaptive compensation. Specifically, it can be established and implemented using linear formulas, piecewise functions, or compensation lookup tables.
[0059] The compensation factor refers to the correction coefficient used to quantify the degree of influence of temperature deviation on the target control parameters.
[0060] The system acquires the first ambient temperature at the end of constant current charging and the second ambient temperature at the beginning of constant voltage charging. Specifically, during the end of constant current charging, when the voltage across the battery approaches a preset cutoff voltage, the ambient temperature where the battery is located is measured by a temperature sensor and used as the first ambient temperature. At the beginning of constant voltage charging, i.e., when the charging mode has just switched from constant current to constant voltage, the ambient temperature where the battery is located is measured by a temperature sensor and used as the second ambient temperature. This can be achieved using devices such as thermistors, thermocouples, or infrared temperature sensors, which are placed near the battery to monitor the surface temperature of the battery or the ambient temperature in real time.
[0061] Calculate the temperature difference between the first ambient temperature and the second ambient temperature. Specifically, subtract the first ambient temperature from the second ambient temperature to obtain the temperature difference. For example, if the first ambient temperature is 25℃ and the second ambient temperature is 28℃, then the temperature difference is 3℃.
[0062] Based on the temperature difference, a compensation factor corresponding to the internal resistance information is calculated using a pre-established temperature compensation relationship. Specifically, a pre-established mathematical model or lookup table is used to map the calculated temperature difference into a compensation factor for correcting the internal resistance information. These models are established through a large amount of experimental data, reflecting the relationship between temperature changes and internal resistance changes. For example, the pre-established temperature compensation relationship might be a function f(ΔT) = k*ΔT + b, where ΔT is the temperature difference, k and b are pre-established coefficients, and the calculated f(ΔT) is the compensation factor.
[0063] A compensation factor is used to numerically correct the internal resistance information. Specifically, by matching the corresponding compensation coefficient to the current temperature difference, the measured original internal resistance data is proportionally converted to offset the deviation in the cell's ohmic and polarization internal resistance values caused by changes in ambient temperature, thus eliminating measurement errors caused by temperature factors. The correction calculation logic can be: Corrected internal resistance = Original internal resistance × Compensation factor. The internal resistance data obtained after correction accurately reflects the actual internal resistance characteristics of the battery, providing accurate and reliable data for subsequent parameter calculations and fault anomaly diagnosis.
[0064] By employing the aforementioned technical solution, this application addresses the problem of inaccurate internal resistance information caused by variations in ambient temperature, thereby improving the accuracy of the internal resistance information. Consequently, the accuracy of subsequent parameter calculations and anomaly diagnosis is significantly enhanced, enabling more precise identification of abnormal conditions during the battery capacity grading process.
[0065] In some preferred embodiments, the steps of calculating the deviation of the first set of feature parameters from a preset first reference and the deviation of the second set of feature parameters from a preset second reference include: Obtain the feature parameter set of batteries in the same batch, and extract the first reference feature parameter set and the second reference feature parameter set from the feature parameter set respectively; A first benchmark is established based on the first set of reference feature parameters; A second benchmark is established based on the second set of reference feature parameters; Calculate the first Mahalanobis distance between the first set of feature parameters and the preset first reference, as the degree of deviation relative to the first reference; Calculate the second Mahalanobis distance between the second set of feature parameters and the preset second reference, as the degree of deviation relative to the second reference.
[0066] In this application, a set of characteristic parameters of batteries from the same batch is obtained. Specifically, relevant data from multiple batteries in the same production batch during the capacity grading process are collected. This data may include electrical signal response sequences such as voltage, current, and temperature, as well as characteristic parameters extracted from these sequences. This can be achieved by deploying a data acquisition device on the battery capacity grading line, which can monitor and record various electrical parameters of each battery in real time during the charging phase.
[0067] The first and second reference feature parameter sets are extracted from the feature parameter sets respectively. Specifically, from the complete feature parameter set of batteries in the same batch, a specific subset of parameters is selected to establish the first and second benchmarks according to preset rules or algorithms. This can be implemented using data processing algorithms. For example, the first reference feature parameter set can be defined as parameters characterizing the polarization release current characteristics, and the second reference feature parameter set can be defined as parameters characterizing the intrinsic charging current characteristics. These parameters may include Ap, τaup, Ac, τauc, etc., from a multi-exponential decay hybrid mathematical model.
[0068] Based on the first set of reference feature parameters, a first benchmark is established. Specifically, using the first set of reference feature parameters extracted from batteries in the same batch, a reference system representing normal batteries in that feature dimension is constructed through statistical methods or machine learning algorithms. This can be achieved by calculating the mean and covariance matrix of the first set of reference feature parameters, with the mean serving as the center of the first benchmark, and the covariance matrix describing the shape and direction of the data distribution.
[0069] A second benchmark is established based on the second set of reference feature parameters. Specifically, using the second set of reference feature parameters extracted from batteries in the same batch, a reference system representing normal batteries in this feature dimension is constructed through statistical methods or machine learning algorithms. This can be achieved by calculating the mean and covariance matrix of the second set of reference feature parameters, with the mean serving as the center of the second benchmark, and the covariance matrix describing the shape and direction of the data distribution.
[0070] The first Mahalanobis distance between the first set of feature parameters and a preset first benchmark is calculated as the degree of deviation relative to the first benchmark; the second Mahalanobis distance between the second set of feature parameters and a preset second benchmark is calculated as the degree of deviation relative to the second benchmark. Specifically, the first set of feature parameters of the battery to be diagnosed is compared with the established first benchmark, and the second set of feature parameters of the battery to be diagnosed is compared with the established second benchmark. The degree of deviation is quantified by calculating the Mahalanobis distance between the two. This can be achieved using the Mahalanobis distance calculation formula, which considers the correlation between different dimensions of the data and can more accurately reflect the degree of deviation of multi-dimensional feature parameters.
[0071] This solution obtains a set of characteristic parameters from batteries in the same batch and extracts a first reference set and a second reference set of characteristic parameters from them, providing a reliable data foundation for subsequent benchmark establishment. Based on these reference sets of characteristic parameters, a first benchmark and a second benchmark are established. These benchmarks represent the typical behavior patterns of normal batteries during the capacity testing process, providing a reference system for anomaly identification. By calculating the first Mahalanobis distance between the first set of characteristic parameters of the battery to be diagnosed and the first benchmark, and the second Mahalanobis distance between the second set of characteristic parameters and the second benchmark, the difference in behavior patterns between the battery to be diagnosed and normal batteries can be quantified. Mahalanobis distance, as a distance metric that considers data correlation, can more accurately reflect the degree of deviation of multi-dimensional characteristic parameters, thereby effectively identifying abnormal conditions during the battery capacity testing process.
[0072] Through the above technical solution, this application provides a specific and accurate method for calculating the degree of deviation, thereby improving the accuracy of abnormal diagnosis in the battery capacity grading process. By establishing a reliable benchmark and using Mahalanobis distance to quantify the degree of deviation, potential problems in the battery capacity grading process can be effectively identified, avoiding decreased production efficiency and product quality risks caused by inaccurate diagnosis.
[0073] In some preferred embodiments, determining the abnormal occurrence stage of the battery capacity grading process based on the calculation results of the deviation degree includes: If the first Mahalanobis distance is less than or equal to the preset polarization deviation threshold, and the second Mahalanobis distance is less than or equal to the preset charging deviation threshold, the battery capacity grading process is determined to be operating normally. If the first Mahalanobis distance is greater than the polarization deviation threshold and the second Mahalanobis distance is less than or equal to the charging deviation threshold, it is determined that there is a polarization anomaly at the end of the constant current charging stage. If the first Mahalanobis distance is less than or equal to the polarization deviation threshold and the second Mahalanobis distance is greater than the charging deviation threshold, it is determined that there is an abnormality in charging acceptance during the initial stage of constant voltage charging. If the first Mahalanobis distance is greater than the polarization deviation threshold and the second Mahalanobis distance is greater than the charging deviation threshold, it is determined that there is both polarization anomaly at the end of constant current charging and charging acceptance anomaly at the beginning of constant voltage charging.
[0074] In this application, Mahalanobis distance refers to an effective measure for calculating the similarity between two or more samples, taking into account the covariance structure of the data, thereby more accurately reflecting the true distance between the data.
[0075] The preset polarization deviation threshold refers to the threshold used to judge polarization anomalies. It can be determined based on historical data analysis, expert experience, or statistical methods. For example, it can be set as a certain percentile or standard deviation multiple of the normal battery polarization deviation from the Mahalanobis distance.
[0076] The preset charging deviation threshold is a threshold used to determine abnormal charging acceptance. It can be determined using a method similar to the polarization deviation threshold.
[0077] This solution, by comparing the first and second Mahalanobis distances with their respective preset critical values, can precisely diagnose the abnormality type and stage of the battery capacity grading process, thus solving the problem of insufficient specificity in the abnormality diagnosis results of existing technologies. Specifically, by comparing the first Mahalanobis distance with a preset polarization deviation critical value, and the second Mahalanobis distance with a preset charging deviation critical value, four different combinations can be obtained, each corresponding to a specific abnormal condition or normal operating state. When the first Mahalanobis distance is less than or equal to the polarization deviation critical value, and the second Mahalanobis distance is less than or equal to the charging deviation critical value, it indicates that the battery's polarization characteristics and charging acceptance characteristics are both within the normal range, thus determining that the battery capacity grading process is operating normally. When the first Mahalanobis distance is greater than the polarization deviation critical value, and the second Mahalanobis distance is less than or equal to the charging deviation critical value, this indicates that the battery's polarization characteristics have deviated significantly at the end of constant current charging, but the charging acceptance characteristics are still normal at the beginning of constant voltage charging, thus determining that there is a polarization abnormality at the end of constant current charging. This judgment helps to pinpoint the problem to the polarization process, providing direction for further analysis of the cause of the polarization abnormality. When the first Mahalanobis distance is less than or equal to the polarization deviation threshold, while the second Mahalanobis distance is greater than the charging deviation threshold, this indicates that the battery's polarization characteristics are normal, but the charging acceptance characteristics show a significant deviation in the initial stage of constant voltage charging. Therefore, it is determined that there is an abnormality in charging acceptance during the initial stage of constant voltage charging. This judgment helps to pinpoint the problem to charging acceptance capability and provides direction for further analysis of the cause of the abnormality. When the first Mahalanobis distance is greater than the polarization deviation threshold, and the second Mahalanobis distance is greater than the charging deviation threshold, this indicates that the battery's polarization characteristics at the end of constant current charging and the charging acceptance characteristics at the beginning of constant voltage charging both show significant deviations. Therefore, it is determined that there is a simultaneous polarization abnormality at the end of constant current charging and a charging acceptance abnormality at the beginning of constant voltage charging. This comprehensive diagnosis can identify more complex anomalies and avoid omissions that may be caused by a single judgment.
[0078] Through the above technical solution, this application can precisely diagnose the abnormal types and stages of the battery capacity grading process, solving the problem of insufficient specificity in the abnormal diagnosis results of existing technologies. This solution, by comparing the first and second Mahalanobis distances with their respective preset critical values, can further refine abnormal conditions into four scenarios: normal operation, polarization abnormalities at the end of constant current charging, charging acceptance abnormalities at the beginning of constant voltage charging, and the simultaneous presence of two abnormalities. This detailed diagnostic result provides more specific guidance for subsequent fault location and handling, improves the accuracy of diagnosis, and avoids omissions that may be caused by a single judgment.
[0079] refer to Figure 2 This application provides a fault diagnosis system for the capacity grading process of lithium-ion batteries, the system comprising: Module 1 is used to acquire the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging (the specific process can be found in the previous text). Data extraction module 2 is used to extract the constant current response information at the end of the constant current charging stage and the mixed response information at the beginning of the constant voltage charging stage based on the electrical signal response sequence. The mixed response information is at least composed of a first response component characterizing the residual polarization at the end of the constant current charging stage and a second response component characterizing the inherent charging characteristics at the beginning of the constant voltage charging stage (the specific process can be referred to above). The parameter solving module 3 is used to perform signal separation processing on the mixed response information to obtain the first response component and the second response component, and to solve the parameters of the mixed response information according to the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component; wherein, the first set of characteristic parameters is a set of parameters characterizing the polarization release current characteristics, and the second set of characteristic parameters is a set of parameters characterizing the intrinsic charging current characteristics (the specific process can be referred to above). Deviation calculation module 4 is used to calculate the degree of deviation of the first feature parameter set relative to the preset first benchmark, and the degree of deviation of the second feature parameter set relative to the preset second benchmark (the specific process can be referred to above). The anomaly determination module 5 is used to determine the abnormal conditions of the battery capacity grading process based on the calculation results of the deviation degree (the specific process can be referred to above).
[0080] The above are merely embodiments of this application and are not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.
Claims
1. A method for diagnosing abnormalities during the capacity grading process of a lithium-ion battery, characterized in that, This method is used for anomaly diagnosis during the battery capacity grading process, specifically when the battery charging stage transitions from constant current charging to constant voltage charging. The steps of this method include: Obtain the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging; The constant current response information at the end of the constant current charging stage and the mixed response information at the beginning of the constant voltage charging stage are extracted from the electrical signal response sequence. The mixed response information is at least composed of a first response component characterizing the residual polarization at the end of the constant current charging stage and a second response component characterizing the inherent charging characteristics at the beginning of the constant voltage charging stage. The mixed response information is processed by signal separation to obtain the first response component and the second response component. Then, the parameters of the separated first response component and the second response component are solved according to the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component. The first set of characteristic parameters is a set of parameters characterizing the polarization release current characteristics, and the second set of characteristic parameters is a set of parameters characterizing the intrinsic charging current characteristics. The deviation of the first set of feature parameters from a preset first benchmark and the deviation of the second set of feature parameters from a preset second benchmark are calculated respectively. The abnormal conditions in the battery capacity grading process are determined based on the calculation results of the deviation degree.
2. The method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 1, characterized in that, The steps for obtaining the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging include: During the final stage of constant current charging, the difference between the battery terminal voltage and the preset cutoff voltage threshold is continuously monitored. When the difference is within the preset sampling range, the battery terminal voltage and loop current are collected during the period covering the end of constant current charging, the switching node between constant current and constant voltage charging, and the initial stage of constant voltage charging, using a preset sampling frequency. The acquired battery terminal voltage sequence and loop current sequence are used as the electrical signal response sequence.
3. The method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 2, characterized in that, The step following the acquisition of the battery's electrical signal response sequence from the end of constant current charging to the beginning of constant voltage charging includes: Outlier removal is performed on the electrical signal response sequence to remove singular outliers and obtain a preliminary screening sequence. The initial screening sequence is smoothed and denoised to obtain an optimized electrical signal response sequence.
4. The method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 1, characterized in that, The step of performing signal separation processing on the mixed response information to obtain the first response component and the second response component includes: Using a preset decoupling model, the constant current response information is used as a constraint condition of the decoupling model. Based on the constraint condition, the mixed response information is decoupled and separated into components, and the first response component and the second response component are output.
5. The battery anomaly diagnosis method according to claim 4, characterized in that, The decoupling model is a multi-exponential decay hybrid mathematical model, and its expression is as follows: ; in, This refers to the cumulative charging time during the initial stage of constant voltage charging. This represents the total current during the initial stage of constant voltage charging. Represents the initial amplitude of polarization release. The time constant representing polarization release, The first response component is used to characterize the polarization release current. The initial amplitude representing the intrinsic charge component. The time constant representing intrinsic charge decay, This is the second response component, used to characterize the intrinsic charging current. This represents the steady-state leakage current component.
6. The method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 5, characterized in that, The constant current response information includes the voltage change rate and internal resistance information of the battery when it is in the final stage of constant current charging. The step of solving for the parameters of the separated first response component and second response component based on the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component includes: The voltage change rate and the internal resistance information are used as parameter constraints for the multi-exponential decay hybrid mathematical model. Using the first response component and the second response component as fitting samples, a nonlinear optimization algorithm is used to fit the parameters of the multi-exponential decay hybrid mathematical model to obtain the first set of characteristic parameters corresponding to the first response component and the second set of characteristic parameters corresponding to the second response component.
7. The method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 6, characterized in that, The method also includes a step of temperature compensation for the internal resistance information: The first ambient temperature at the end of the constant current charging phase and the second ambient temperature at the beginning of the constant voltage charging phase are obtained. Calculate the temperature difference between the first ambient temperature and the second ambient temperature; Based on the temperature difference, a compensation factor corresponding to the internal resistance information is calculated using a preset temperature compensation relationship. The internal resistance information is numerically corrected using the compensation factor.
8. The method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 1, characterized in that, The steps of calculating the deviation of the first feature parameter set from a preset first benchmark and the deviation of the second feature parameter set from a preset second benchmark include: Obtain the feature parameter set of batteries in the same batch, and extract the first reference feature parameter set and the second reference feature parameter set from the feature parameter set respectively; A first benchmark is established based on the first set of reference feature parameters; A second benchmark is established based on the second set of reference feature parameters; Calculate the first Mahalanobis distance between the first set of feature parameters and the preset first reference, as the degree of deviation relative to the first reference; Calculate the second Mahalanobis distance between the second set of feature parameters and the preset second reference, as the degree of deviation relative to the second reference.
9. A method for diagnosing abnormalities in the capacity grading process of a lithium-ion battery according to claim 8, characterized in that, The determination of the abnormal occurrence stage in the battery capacity grading process based on the calculation results of the deviation degree includes: If the first Mahalanobis distance is less than or equal to the preset polarization deviation threshold, and the second Mahalanobis distance is less than or equal to the preset charging deviation threshold, the battery capacity grading process is determined to be operating normally. If the first Mahalanobis distance is greater than the polarization deviation threshold and the second Mahalanobis distance is less than or equal to the charging deviation threshold, it is determined that there is a polarization anomaly at the end of the constant current charging. If the first Mahalanobis distance is less than or equal to the polarization deviation threshold, and the second Mahalanobis distance is greater than the charging deviation threshold, it is determined that there is an abnormality in the charging acceptance during the initial stage of constant voltage charging. If the first Mahalanobis distance is greater than the polarization deviation threshold and the second Mahalanobis distance is greater than the charging deviation threshold, it is determined that there is both polarization abnormality at the end of constant current charging and charging acceptance abnormality at the beginning of constant voltage charging.
10. A lithium-ion battery capacity grading process abnormal diagnosis system, used to perform the steps of the method according to any one of claims 1 to 9, characterized in that, include: The acquisition module is used to acquire the electrical signal response sequence of the battery from the end of constant current charging to the beginning of constant voltage charging. The data extraction module is used to extract the constant current response information at the end of the constant current charging stage and the mixed response information at the beginning of the constant voltage charging stage based on the electrical signal response sequence. The mixed response information is at least composed of a first response component characterizing the residual polarization at the end of the constant current charging stage and a second response component characterizing the inherent charging characteristics at the beginning of the constant voltage charging stage. The parameter solving module is used to perform signal separation processing on the mixed response information to obtain the first response component and the second response component, and to perform parameter solving on the mixed response information according to the constant current response information to obtain a first set of characteristic parameters corresponding to the first response component and a second set of characteristic parameters corresponding to the second response component; wherein, the first set of characteristic parameters is a set of parameters characterizing the polarization release current characteristics, and the second set of characteristic parameters is a set of parameters characterizing the intrinsic charging current characteristics; The deviation calculation module is used to calculate the degree of deviation of the first feature parameter set relative to a preset first benchmark, and the degree of deviation of the second feature parameter set relative to a preset second benchmark, respectively. An anomaly detection module is used to determine abnormal conditions in the battery capacity testing process based on the calculation results of the deviation degree.