A method for predicting the residual life of a retired power cell

Abstract

The application provides a method for predicting the residual life of a retired power battery, which comprises: performing impedance tests at different frequencies on a to-be-tested battery cell in a retired power battery within a preset test frequency range, obtaining impedance response data at different frequencies, and obtaining impedance spectrum data; obtaining a utilization scenario condition corresponding to different steps, wherein the utilization scenario condition comprises a temperature range and a charge / discharge rate, and a target step utilization scenario is matched in combination with a step grade to which the to-be-tested battery cell belongs; storing the target step utilization scenario in association with retired battery cell model, capacity, internal resistance and impedance spectrum data information, and forming a retired battery step utilization scheme; and if an abnormal state of the battery is detected, triggering a warning mechanism, and taking corresponding safety protection measures according to a preset processing strategy, including cutting off the power supply and starting a cooling system.

Description

Technical Field

[0001] This invention relates to the field of information technology, and in particular to a method for predicting the remaining lifespan of retired power battery cells. Background Technology

[0002] In the scenario of tiered utilization of retired power batteries, it is necessary to assess the remaining life of the cells in order to make reasonable tiered classification. Firstly, obtaining impedance spectrum data of the cells in retired power batteries is a crucial step in predicting the remaining life of the cells. However, impedance spectrum data typically contains multiple characteristic parameters, such as ohmic internal resistance, polarization resistance, and diffusion resistance. Different characteristic parameters have varying degrees of influence on the remaining life of the cells, requiring the selection of appropriate characteristic parameters as input to the life prediction model. Simultaneously, the choice of measurement frequency during impedance spectrum data acquisition also affects the data quality and the accuracy of subsequent models. Therefore, how to acquire high-quality impedance spectrum data under complex operating conditions and extract effective characteristic parameters from it is a key technical problem in assessing the remaining life of retired power batteries. Furthermore, how to classify the batteries into tiers based on their remaining life is also a significant technical challenge in this field, and matching the tiered classification with corresponding utilization scenarios is another technical difficulty to be solved. In addition, ensuring the safety and reliability of utilization during the actual tiered utilization process is also one of the technical issues that this solution needs to consider. Summary of the Invention

[0003] This invention provides a method for predicting the remaining life of retired power battery cells, mainly including:

[0004] Obtain basic parameter information of the cell to be tested in the retired power battery, including cell model, capacity, and internal resistance. Based on the cell model, obtain the standard impedance spectrum data of the cell model from the preset cell parameter database.

[0005] For the cells under test in retired power batteries, impedance tests are performed at different frequencies within a preset test frequency range to obtain impedance response data at different frequencies and obtain impedance spectrum data.

[0006] The acquired impedance spectrum data is preprocessed, including removing outliers, smoothing the data, extracting key feature parameters of the impedance spectrum curve, including arc radius, center coordinates, and slope, and constructing impedance spectrum feature vectors.

[0007] The impedance spectrum feature vector is input into the pre-built lifetime prediction model, and the predicted remaining lifetime data of the cell under test is output. Combined with the preset ladder utilization division rules, the ladder level to which the cell under test belongs is obtained.

[0008] Obtain the utilization scenario conditions corresponding to different tiers, including temperature range and charge / discharge rate, and match the target tier utilization scenario based on the tier level to which the current test cell belongs.

[0009] The target tiered utilization scenarios are associated with and stored with retired battery cell model, capacity, internal resistance and impedance spectrum data to form a tiered utilization scheme for retired batteries;

[0010] In the actual tiered utilization process, according to the established tiered utilization plan for retired batteries, real-time working data of retired power batteries in the corresponding utilization scenarios are collected, including voltage, current, and temperature. Through a preset anomaly detection model, it is determined whether the battery is in an abnormal state.

[0011] If an abnormal state of the battery is detected, an early warning mechanism is triggered, and corresponding safety protection measures are taken according to the preset handling strategy, including cutting off the power supply and starting the cooling system.

[0012] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0013] This invention discloses a method for predicting the remaining lifespan of retired power battery cells. This method acquires the basic parameters and impedance spectrum data of retired battery cells, and combines this with a pre-set lifespan prediction model and tiered utilization classification rules to accurately assess the remaining lifespan of retired batteries and classify them into tiered levels. Furthermore, this invention matches corresponding utilization scenarios based on the battery's tiered level to form a tiered utilization scheme for retired batteries. In actual utilization, this invention also monitors battery operating data in real time, and combines this with an anomaly detection model to promptly identify abnormal battery states and trigger corresponding safety protection measures. This method not only improves the utilization efficiency of retired power batteries but also ensures the safety of the tiered utilization process, providing an effective solution for the full lifecycle management of power batteries. Attached Figure Description

[0014] Figure 1 This is a flowchart of a method for predicting the remaining life of retired power battery cells according to the present invention.

[0015] Figure 2 This is a schematic diagram of a method for predicting the remaining life of retired power battery cells according to the present invention.

[0016] Figure 3 This is another schematic diagram of a method for predicting the remaining life of retired power battery cells according to the present invention. Detailed Implementation

[0017] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this specification, and not all embodiments. Based on the embodiments in this specification, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of this specification.

[0018] like Figure 1-3 This embodiment of a method for predicting the remaining life of retired power battery cells may specifically include:

[0019] S101. Obtain the basic parameter information of the cell to be tested in the retired power battery, including cell model, capacity, and internal resistance. Obtain the standard impedance spectrum data of the cell model from the preset cell parameter database according to the cell model.

[0020] Design specifications, including nominal capacity, rated voltage, and standard internal resistance, are obtained from a pre-set database based on the identification of retired power cells. The retired power cells are charged to their rated voltage using a constant current and constant voltage method, and then left to stand for a preset time to obtain a stable state. An electrochemical impedance spectroscopy (EIS) is used to perform frequency sweep measurements on the stable state cells to obtain raw impedance data. A sequence of real and imaginary impedance values ​​is generated using a fast Fourier transform. The impedance value sequence is then validated for data validity, outlier data points are removed, and cubic spline interpolation is used to complete the missing data, generating a continuous impedance curve.

[0021] Specifically, the design specifications of the retired power battery cells are retrieved from a pre-defined database based on their identification. Data lookup and matching are performed on the cell model code and production batch number. The nominal capacity, rated voltage, and standard internal resistance of the cell model recorded at the time of manufacture are obtained from the parameter database. The ambient temperature and humidity are controlled to maintain a constant temperature of 25 degrees Celsius for the test cell. The cell is charged to its rated voltage using a constant current and constant voltage method and allowed to stand for a preset time to reach a stable state. An electrochemical impedance spectroscopy (EIS) is used to perform frequency sweep measurements on the cell within a preset frequency range to obtain raw impedance data. The measured data is then analyzed using a fast Fourier transform to generate a sequence of real and imaginary impedance values. The impedance value sequence is validated for data validity, outlier data points are removed, and missing data is filled in using cubic spline interpolation to generate a continuous impedance curve. The locations of charge transfer impedance points and double-layer capacitance points are marked on the spectrum. The measured impedance curve of the battery cell is compared with the standard impedance curve of the same model. The relative offset of the feature point position is calculated. If the offset exceeds a preset threshold range, a pre-trained convolutional model is used to fit the impedance spectrum data. The least squares method is used to calculate the matching degree difference rate between the measured impedance curve and the standard impedance curve. The difference rate data is normalized using the hyperbolic tangent function to generate a standardized feature vector. The standardized feature vector is input into a pre-trained support vector regression model. This model uses a Gaussian kernel function, and the input parameters include the impedance feature point offset, the internal resistance difference rate, and the voltage response curve feature value. The output is the value of the battery cell capacity decay. Retired power battery cells will experience performance degradation due to cyclic charging and discharging and environmental factors during use. Electrochemical impedance spectroscopy can be used to measure the changes in the internal impedance characteristics of the battery cell. Taking a lithium iron phosphate battery cell as an example, the nominal capacity is 60 amp-hours, the rated voltage is 3.2 volts, and the standard internal resistance at the factory is 0.8 milliohms. After a certain period of use, the measured internal resistance of the battery cell may rise to 1.2 milliohms, indicating that performance degradation has occurred within the cell. Before impedance measurement, the cell needs to be pre-charged to ensure it is in a suitable testing state. For lithium iron phosphate cells, a constant current charge of 0.5 times the rated current is used to charge to 3.2 volts, followed by constant voltage for 2 hours to stabilize the internal electrochemical reaction. The test environment temperature is maintained at 25 degrees Celsius, and the relative humidity is controlled at 45% to avoid interference from environmental factors. The impedance spectrum measurement frequency range is set between 0.01 Hz and 100 kHz, and impedance data at different frequencies are obtained through a logarithmic frequency sweep. After fast Fourier transform processing, typical semi-circular impedance spectrum characteristics can be observed on the Nyquist plot. The high-frequency region reflects the electrolyte resistance characteristics, the semi-circular diameter in the mid-frequency region corresponds to the charge transfer impedance, and the low-frequency region reflects the diffusion characteristics of lithium ions in the electrode material.The handling of outlier data employs a three-standard-deviation principle, identifying and removing data points that deviate from the mean by more than three standard deviations. For example, an impedance value of 2.5 milliohms measured at 50 Hz, while the mean at that frequency is 1.5 milliohms and the standard deviation is 0.2 milliohms, is removed because it deviates from the mean by more than three standard deviations. Data is then supplemented using cubic spline interpolation. The shift of characteristic points on the impedance curve reflects the degree of degradation of different components within the cell. A shift to the right of the charge transfer impedance point exceeding 30% indicates a decrease in the activity of the electrode material. By analyzing the changing trends of parameters such as charge transfer impedance and double-layer capacitance, combined with a support vector regression model, the capacity decay of the cell can be assessed. When the capacity decay exceeds 20%, the cell is no longer suitable for use in the power battery field but can be downgraded for use in energy storage and other fields. This impedance-based assessment method provides an important basis for the graded utilization of retired power batteries.

[0022] S102. For the cell under test in the retired power battery, perform impedance tests at different frequencies within the preset test frequency range, obtain impedance response data at different frequencies, and obtain impedance spectrum data.

[0023] The retired power battery cell is charged with constant current and constant voltage until the charging cutoff voltage value is reached. Then, a constant current power supply is used to apply the rated current to the charged cell to discharge it until the discharge cutoff voltage value is reached. After discharge, the open-circuit voltage of the cell is monitored. If the open-circuit voltage fluctuation value is less than a preset voltage fluctuation threshold, the cell is determined to be in a stable state. For the cell in a stable state, a sinusoidal AC excitation signal is applied using an electrochemical impedance spectroscopy (EIS). The response to the excitation signal is subjected to Fourier transform to obtain an impedance value sequence. Cubic spline interpolation is used to perform point supplementation on the impedance value sequence to obtain an impedance data curve. Based on the impedance data curve, the positions of semi-circular feature points are identified to obtain an impedance spectrum feature dataset.

[0024] Specifically, the retired power battery cells are charged using constant current and constant voltage based on their rated voltage. The surface temperature of the battery cells is collected by a temperature sensor. When the temperature is between 20 and 30 degrees Celsius, a constant current power supply is used to apply a current of 0.2 times the rated current to the battery cells for constant current discharge until the battery cell voltage reaches the discharge cutoff voltage. The battery cells are then left to stand, and the open-circuit voltage is continuously monitored by a voltage acquisition module. When the open-circuit voltage fluctuation is less than a preset threshold, the battery cell is determined to be in a stable state. An electrochemical impedance spectroscopy (EIS) instrument is used to apply a sinusoidal AC excitation signal with an amplitude of 5% of the charging current to the stable battery cells. Starting from a kilohertz frequency, the frequency is swept down to the millihertz level, decreasing by an order of magnitude every ten frequency points. The voltage response signal and current excitation signal at each frequency point are recorded using a high-precision data acquisition device. The signal-to-noise ratio (SNR) of the obtained response signal is calculated, and frequency points with SNR below a preset threshold are removed. Fourier spectrum transformation is performed on the remaining response signal to calculate the voltage amplitude and phase value at each frequency sampling point. The real and imaginary parts of the impedance are obtained through complex number operations. Cubic spline interpolation is used to densify and supplement the frequency intervals in the impedance value sequence, generating a continuous impedance data curve. Wavelet multi-scale decomposition is used to smooth and reduce noise on the curve. Feature extraction is performed on the smoothed impedance data curve to identify the start, vertex, and end points of the semicircle, and the frequency and impedance values ​​corresponding to each feature point are recorded to generate a complete impedance spectrum feature dataset. Electrochemical impedance spectroscopy (EIS) measurement requires high control over the cell's state, necessitating preprocessing of the cell before measurement. Taking a ternary lithium battery as an example, when the rated voltage is 3.7 volts, constant current charging is performed at 0.2 times the rated current to 4.2 volts, followed by constant voltage charging at 4.2 volts until the charging current drops to 0.05 times the rated current, at which point the cell reaches full charge. During charging, the cell surface temperature rises and needs to be allowed to cool down to around 25 degrees Celsius. During this cooling period, the open-circuit voltage will fluctuate slightly. When the voltage fluctuation is less than 2 millivolts over 30 minutes of continuous monitoring, it indicates that the electrochemical reaction inside the cell is approaching equilibrium. The measured open-circuit voltage at this point is 4.186 volts, within a reasonable range from the theoretical value of 4.2 volts, indicating that the cell is suitable for impedance measurement. Impedance measurement uses a small-signal excitation method, sweeping the frequency with an AC signal at 5% of the rated charging current. For a 60 Ah battery, the excitation current amplitude is 3 Amps. The frequency sweep range starts at 1000 Hz, decreasing by an order of magnitude every 10 frequency points until reaching 0.01 Hz, for a total of 61 frequency sampling points. Data is collected for 16 cycles at each frequency point. The acquisition time for a single frequency point is shorter at higher frequencies, while at 0.01 Hz, a single point acquisition requires 1600 seconds. The original acquired signal contains noise interference. Data is filtered by calculating the signal-to-noise ratio (SNR). Data with an SNR below 20 dB is considered invalid.Impedance data obtained through Fourier transform showed an impedance of 0.8 milliohms at 1000 Hz, primarily reflecting the electrolyte resistance. A semicircular arc formed between 100 Hz and 1 Hz, corresponding to the charge transfer process, with a diameter of 1.2 milliohms. Below 0.1 Hz, it appeared as a sloping line, reflecting the diffusion characteristics of lithium ions in the electrode material. Cubic spline interpolation was performed on the impedance curve, adding 9 data points at intervals to the original 61 frequency points, resulting in a continuous curve of 541 points. A db4 wavelet was used to decompose the curve into four levels, removing high-frequency noise before reconstructing a smooth curve. On the Nyquist plot, the starting point of the semicircular arc was the point of minimum real resistance, and the ending point was the point of maximum real resistance. The position of the semicircular arc's apex was determined through curvature calculation. The positions and values ​​of these characteristic points reflect the dynamic characteristics of different physical processes within the battery and are important indicators for evaluating battery performance.

[0025] S103. Preprocess the acquired impedance spectrum data, including removing outliers, smoothing the data, extracting key feature parameters of the impedance spectrum curve, including arc radius, center coordinates, and slope, and constructing impedance spectrum feature vectors.

[0026] The local variance of the impedance spectrum data sequence is calculated using a sliding window. If the variance exceeds the threshold of the mean variance of all frequencies, the marked outlier data points are removed from the impedance spectrum data sequence. The interval between adjacent frequency points is calculated based on the impedance spectrum data sequence after removing outliers. If the interval exceeds a preset frequency step size, linear interpolation is used to obtain supplementary frequency point data. For the impedance spectrum sequence after supplementing the frequency point data, high-frequency noise components are removed using a Butterworth low-pass filter and four-layer discrete wavelet decomposition. A continuous and smooth impedance spectrum curve is obtained using a cubic spline function. Circular and linear fitting are performed on the complex plane based on the continuous and smooth impedance spectrum curve. The coordinates of the center and the radius are obtained using iterative least squares. The real and imaginary parts of the low-frequency slope are obtained through complex number operations. The eigenvectors are then normalized.

[0027] Specifically, based on the fluctuation amplitude between adjacent data points in the original impedance spectrum data sequence, a sliding window of five data points is used to calculate the local variance. If the variance exceeds three times the mean variance of all variances within that frequency range, the data point is marked as an outlier and removed from the impedance spectrum data sequence. The impedance spectrum data sequence after outlier removal is then validated for data integrity by calculating the interval between adjacent frequency points. If the interval exceeds a preset frequency step size, linear interpolation is used to supplement the missing frequency data. For the impedance spectrum sequence after data supplementation, a Butterworth low-pass filter with a cutoff frequency one-twentieth of the sampling frequency is used for processing. High-frequency noise components are removed through four-layer discrete wavelet decomposition. A cubic spline function is used to fit the data points, resulting in a continuous and smooth impedance spectrum curve. Circular arc fitting is performed on the impedance spectrum data in the frequency range of 10 Hz to 1 kHz on the complex plane. Iterative least squares is used to calculate the center coordinates and radius of the arc. Iteration stops when the fitting residual is less than a preset threshold. The least squares method was used to linearly fit impedance spectrum data with frequencies less than 10 Hz, calculating the slope of the impedance curve in the low-frequency region. Complex number operations were then used to obtain the real and imaginary components of this slope. A five-dimensional feature vector was constructed by combining the abscissa of the arc center, the ordinate of the arc center, the arc radius, the real part of the low-frequency slope, and the imaginary part of the low-frequency slope in a predetermined order, recording the physical meaning of each feature component. Principal component analysis was used to reduce the dimensionality of the five-dimensional feature vector, selecting principal components with a cumulative contribution rate exceeding a preset threshold as new feature vectors. The feature vectors were then normalized by calculating the Mahalanobis distance from the data points to the centroid. Preprocessing of impedance spectrum data is a crucial step in extracting effective features, as the original data often contains various anomalies and noise. In actual measurements, taking a 60 Ah ternary lithium battery as an example, the impedance measured at 1000 Hz was 0.8 milliohms, while the impedance values ​​at adjacent frequency points of 999 Hz and 1001 Hz were 0.79 milliohms and 2.1 milliohms, respectively, clearly indicating an anomaly in the data at 1001 Hz. Using a 5-point sliding window to calculate the variance, the local variance at this anomaly point was 0.42, far exceeding the average variance of 0.005 within this frequency band; therefore, it was marked as an anomaly and removed. Data integrity verification showed that there were excessively large frequency intervals in the 100 Hz to 10 Hz range, with intervals between adjacent frequency points reaching 15 Hz, exceeding the preset 10 Hz step size requirement. Data at intermediate frequency points such as 12 Hz and 14 Hz were added through linear interpolation to make the frequency distribution more uniform. The supplemented data was processed using a 4th-order Butterworth low-pass filter with a cutoff frequency set to 50 Hz, effectively removing high-frequency interference during the measurement process. In the complex plane, the impedance data in the 100 Hz to 1 Hz range exhibits a typical semi-circular shape, and the iterative least squares method is used for circular arc fitting.The initial center coordinates were set to a real part of 1.2 mΩ, an imaginary part of 0.6 mΩ, and a radius of 0.8 mΩ. After 20 iterations, the fitting residual decreased to below 0.01, resulting in a center coordinate of 1.25 mΩ, an imaginary part of 0.62 mΩ, and a radius of 0.82 mΩ. In the low-frequency region below 1 Hz, the impedance curve exhibits an approximately 45-degree slope, reflecting the characteristics of the diffusion process. Least squares fitting yielded a slope with a real part of 0.71 and an imaginary part of 0.68, close to the 45-degree angle of the ideal diffusion process. Combining the arc feature and low-frequency feature into a five-dimensional vector [1.25, 0.62, 0.82, 0.71, 0.68], and reducing it to three dimensions through principal component analysis, the cumulative contribution rate reached 95%. These feature components reflect the characteristics of electrolyte resistance, charge transfer resistance, and diffusion impedance, respectively, and are important indicators for evaluating battery performance. The Mahalanobis distance of all sample feature vectors is calculated and normalized to obtain standardized feature vectors, which facilitates subsequent battery performance evaluation.

[0028] S104. Input the impedance spectrum feature vector into the pre-built lifetime prediction model, and output the predicted remaining lifetime data of the cell under test. Combine the preset ladder utilization division rules to obtain the ladder level to which the cell under test belongs.

[0029] The mean and standard deviation of each feature component are calculated based on the impedance spectrum feature vector, and a standardized feature vector is obtained by using the Gaussian distribution standardization method. Singular value decomposition is performed on the standardized feature vector, and the feature components whose cumulative contribution rate exceeds the contribution rate threshold are retained to obtain a dimensionality-reduced feature vector. This dimensionality-reduced feature vector is then input into a preset deep neural network, and a nonlinear transformation is performed on the output values ​​of each layer node to obtain the predicted remaining life of the battery cell. Fuzzy clustering is used to calculate the membership degree of the predicted remaining life of the battery cell to each level. The grading results are verified by calculating the Mahalanobis distance between the battery cell feature vector and the sample centers of each level, thus obtaining the final grading data for the battery cell.

[0030] Specifically, the impedance spectrum feature vector is preprocessed by normalization, and the feature data is scaled using a Gaussian distribution standardization method. The mean and standard deviation of each feature component are calculated to generate a standardized feature vector. Singular value decomposition is used to denoise the standardized feature vector, and the feature components with a cumulative contribution rate exceeding a preset threshold are retained, generating a dimensionality-reduced feature vector. Training weight parameters are read from a preset five-layer deep neural network containing three hidden layers with sixteen, eighteen, and four nodes respectively. The input layer corresponds to the dimension of the dimensionality-reduced feature vector, and the output layer is the remaining life value corresponding to a single node. The dimensionality-reduced feature vector is input into the deep neural network, and the output value of each layer node is calculated sequentially using the forward propagation algorithm. A nonlinear transformation is performed using the sigmoid activation function to obtain the predicted remaining life value of the battery cell. The prediction result is verified using a five-fold cross-validation method. The mean and variance of the five prediction results are calculated, and the confidence level of the prediction result is evaluated using the root mean square error. If the confidence level is higher than a preset threshold, the prediction result is recorded. Based on the recorded predicted remaining lifespan of the battery cells, and combined with a pre-defined four-tiered grading standard, fuzzy clustering is used to calculate the membership degree of the predicted value to each tier. The tier with the highest membership degree is the tier to which the battery cell belongs. The grading results are then verified by calculating the Mahalanobis distance between the cell's feature vector and the sample centers of each tier. The tier with the smallest distance should be consistent with the fuzzy clustering result, generating the final grading data for the battery cells. The prediction of the lifespan of retired power batteries begins with the preprocessing of impedance feature vector data. Taking a certain type of ternary lithium battery as an example, its impedance feature vector includes five feature components: arc radius, center coordinates, low-frequency slope, etc. The original feature data has significant differences in dimensions and numerical ranges; for example, the arc radius is 0.82 milliohms, while the low-frequency slope is 45.3 degrees. After Gaussian standardization, the feature means were calculated to be 1.25, 0.62, 0.82, 0.71, and 0.68, with standard deviations of 0.15, 0.08, 0.11, 0.09, and 0.08, transforming the original data into a standard normal distribution with a mean of 0 and a standard deviation of 1. Singular value decomposition (SVD) was then performed on the standardized feature vectors, yielding five singular values: 2.85, 1.42, 0.76, 0.35, and 0.12. The feature vectors corresponding to the top three singular values ​​with a cumulative contribution rate of 95% were selected to achieve dimensionality reduction and noise reduction. The deep neural network employs a three-layer hidden layer structure. The input layer corresponds to a 3-dimensional feature vector, with 16 nodes in the first hidden layer, 8 nodes in the second hidden layer, 4 nodes in the third hidden layer, and 1 node in the output layer corresponding to the remaining iteration count. In the trained neural network, the standardized feature vector of a certain battery cell [-0.56, 0.82, 0.33] is input. After calculation by each layer node and processing by the sigmoid activation function, the predicted remaining number of iterations is 453.Using five-fold cross-validation, the sample was randomly divided into five parts. Four parts were selected as the training set and one part as the validation set each time, resulting in five predictions: 453, 448, 460, 442, and 456, with a mean of 452 and a standard deviation of 6.8. The root mean square error (RMSE) was 1.5%, lower than the preset threshold of 2%, indicating a high confidence level in the predictions. Based on a preset tiered utilization classification standard, cells with more than 800 remaining cycles were classified as Level 1, 500-800 as Level 2, 200-500 as Level 3, and less than 200 as Level 4. The membership degrees for each level were calculated for the predicted value of 452, yielding membership degrees of 0.05, 0.15, 0.75, and 0.05, respectively, classifying the cell as Level 3. By calculating the Mahalanobis distance between the feature vector of the battery cell and the sample centers of each level, the distance values ​​were found to be 8.6, 4.2, 1.8, and 5.3, respectively. The minimum distance corresponds to the third level, which is consistent with the fuzzy clustering results and verifies the reliability of the grading results. At this point, it can be determined that the battery cell is suitable for low-rate energy storage and other fields.

[0031] S105. Obtain the utilization scenario conditions corresponding to different steps. The utilization scenario conditions include temperature range and charge / discharge rate. Combine the step level to which the current test cell belongs to and match the target step utilization scenario.

[0032] A tiered utilization scenario table is retrieved from a preset scenario parameter database. This table includes temperature range limits, charge / discharge rate limits, voltage range limits, cycle depth limits, and operating time limits. A judgment matrix is ​​constructed based on the limit parameters in the tiered utilization scenario table, and the weight coefficients of the limit parameters are calculated using the analytic hierarchy process (AHP). Operating parameters for energy storage, power, and backup power scenarios are extracted based on these weight coefficients, and scenario adaptability scores are obtained using statistical methods. The membership degree of the tested battery cell to each scenario is calculated based on the scenario adaptability scores, and the performance parameters of the tested battery cell are verified using the Monte Carlo method to determine whether they meet the scenario requirements.

[0033] Specifically, a tiered utilization scenario table is retrieved from a preset scenario parameter database. This table includes upper and lower limits for temperature range, charging rate limits, discharging rate limits, upper and lower limits for operating voltage, cycle depth limits, and continuous operating time limits. A judgment matrix is ​​constructed using the analytic hierarchy process (AHP). The relative importance of five parameters—temperature range, charging / discharging rate, voltage range, cycle depth, and operating time—is calculated through pairwise comparisons, and weighting coefficients are obtained by calculating eigenvalues ​​and eigenvectors. Operating parameters are extracted for energy storage, power, and backup power scenarios, including ambient temperature change curves, load power curves, charging / discharging sequences, and operating time distributions. Statistical methods are used to obtain the characteristic parameter values ​​for each scenario. Based on the parameter limits of the tiered level to which the tested cell belongs, the difference between the tested cell and the characteristic parameters of each application scenario is calculated. A weighted summation method is used to calculate the comprehensive difference score, generating a scenario adaptability score. The scenario adaptability score is normalized, and a fuzzy comprehensive evaluation method is used to calculate the membership value of the tested cell to each scenario, generating a scenario matching matrix. Based on the principle of maximum membership, the application scenario with the highest matching degree is selected. The Monte Carlo method is used to analyze the fluctuation of cell performance parameters in this scenario to verify whether the cell parameters meet the scenario requirements. According to the performance parameter fluctuation analysis results, if the fluctuation range of key parameters is within the scenario constraints, the parameters of this application scenario are recorded as a tiered utilization scheme for the cell, generating recommended tiered utilization scenario data. The selection of tiered utilization scenarios for retired power batteries needs to consider the matching degree of multiple performance parameters with application conditions. Taking ternary lithium batteries as an example, the tiered utilization scenario table sets strict parameter limits for different levels of cells. The operating temperature range of Level 1 cells is -20 to 55 degrees Celsius, the charging rate limit is 1, the discharging rate limit is 2, the operating voltage range is 3.0 to 4.2 volts, the cycle depth limit is 80%, and the continuous working time limit is 4 hours. When determining the weight of scenario parameters, a 5th-order judgment matrix is ​​constructed using the analytic hierarchy process (AHP). The temperature range has an importance of 2 compared to the charging / discharging rate, 3 compared to the voltage range, 4 compared to the cycle depth, and 5 compared to the working time. By calculating the maximum eigenvalue and eigenvector of the judgment matrix, the weights of temperature range, charge / discharge rate, voltage range, cycle depth, and operating time are 0.42, 0.26, 0.16, 0.10, and 0.06, respectively. For the operating characteristics of energy storage scenarios, the ambient temperature fluctuates between 15 and 35 degrees Celsius, the load power curve shows two peaks in the morning and evening, the charge / discharge sequence is highly regular, and the single operating time is 2 to 6 hours. In power scenarios, temperature changes drastically, load power fluctuates greatly, charge / discharge is highly random, and operating time is not fixed. In backup power scenarios, the temperature is relatively stable, the load power is stable, the number of charge / discharge cycles is low, and the single operating time is short.The matching degree between cell parameter limits and scenario characteristic parameters was calculated. Temperature adaptability was scored using the overlap range ratio, charge / discharge rate was scored using the margin ratio, voltage range was scored using coverage, and cycle depth and operating time were scored using the satisfaction rate. The calculation results for a certain level 3 cell showed that the temperature score for energy storage scenarios was 0.85, the rate score was 0.92, the voltage score was 0.88, the cycle depth score was 0.78, and the operating time score was 0.82. After weighted summation, the comprehensive score for energy storage scenarios was 0.86, for power scenarios 0.65, and for backup power scenarios 0.72. 1000 parameter fluctuation simulations were conducted using the Monte Carlo method to analyze the cell's performance in energy storage scenarios. The results showed that temperature fluctuations were within the limits in 95% of the operating conditions, the charge / discharge rate met the requirements in 98% of the operating conditions, and the cycle depth did not exceed the limit in 93% of the operating conditions. The comprehensive evaluation indicates that this cell is suitable for energy storage scenarios, specifically for residential or commercial / industrial energy storage.

[0034] S106. Link and store the target tiered utilization scenarios with the retired battery cell model, capacity, internal resistance and impedance spectrum data to form a tiered utilization scheme for retired batteries.

[0035] An original identifier string is generated by concatenating model identification, capacity value, internal resistance value, and impedance spectrum data. A scheme identification code is obtained through a secure hash algorithm. Based on the scheme identification code, parameter information table, impedance spectrum data table, grade information table, and scenario information table are established, with the scheme identification code serving as the primary key to establish the relationship between the data tables. A null value detection method is used to identify missing data items in the data tables, and anomaly data markers are obtained through data type verification methods. A data quality report is generated based on the anomaly data markers. A structured query statement is used to extract complete data from the data tables, and a retired battery tiered utilization scheme document containing a basic information area, a feature data area, and a scenario parameter area is generated based on the complete data.

[0036] Specifically, based on the model identifier, capacity value, internal resistance value, and impedance spectrum data of the retired battery cells, an original identifier string is generated using information concatenation. A 256-bit scheme identifier code is then generated using a secure hash algorithm, and the current timestamp is recorded as the scheme creation time. Four data tables are established for the retired battery cell data: a parameter information table, an impedance spectrum data table, a grade information table, and a scenario information table. The scheme identifier code is used as the primary key to establish relationships, and secondary indexes are created for scheme time, battery cell model, and scenario type. The data items in the four data tables are subjected to integrity verification. A null value detection method is used to identify missing data items, and a data format verification method is used to check the data format. Data items that do not meet the specifications are marked. A data quality report is generated for the marked abnormal data items, recording the location, type, and severity of the abnormal data. A data completion method is used to repair the missing data, generating a data repair record. A structured query statement is used to extract complete data from the four data tables, generating a scheme document containing a basic information area, a feature data area, and a scenario parameter area. The document version number and generation time are recorded. A retrieval mapping table is created for the generated solution document, containing five fields: solution identifier, document version number, timestamp, data source table identifier, and index type identifier. This mapping table is stored in the database index area. A distributed storage mechanism is used to back up the solution document to three copies, storing the copies on storage nodes in different physical locations, and recording the storage location identifier and synchronization timestamp of each copy. Data management of retired power batteries involves the establishment and maintenance of multiple related tables. Taking a certain model of ternary lithium battery as an example, its basic information includes the model identifier LIR18650-30, capacity value of 3000 mAh, and internal resistance value of 20 milliohms. The original string "LIR18650-30_3000_20" is generated by concatenating the information, and a fixed-length identifier code "7a8b9c0d" is generated using the SHA-256 hash algorithm, while simultaneously recording the timestamp "2024-01-08-14:30:25". The database design employs a relational structure. The parameter information table contains basic parameter fields such as model, capacity, and internal resistance. The impedance spectrum data table stores frequency points, real and imaginary impedance values. The grade information table records the grade level and evaluation time. The scenario information table contains scenario parameters such as temperature range and rate limit. A one-to-one relationship is established between tables using the scheme identifier code as the primary key. A timestamp index is created for the scheme time, a B-tree index for the cell model, and a hash index for the scenario type. Data integrity verification revealed an anomaly in the impedance spectrum data table: a missing impedance value at the 1000 Hz frequency point. The data quality report records the anomaly location as "Impedance Spectrum Data Table - Frequency 1000 Hz," the anomaly type as "Missing Value," and the severity as "Medium." Linear interpolation was used to fill in the missing data based on the impedance values ​​at adjacent frequency points of 999 Hz and 1001 Hz. The data repair record shows that the filled value is an imaginary part of 0.82 milliohms and a real part of 1.25 milliohms.The solution document employs a partitioned storage structure. The basic information area records the cell model, capacity, and internal resistance; the characteristic data area stores impedance spectrum curve parameters; and the scenario parameter area contains the operating condition requirements of the utilization scenario. The document version number uses a three-segment format "1.0.0", where the first digit represents the major version number, the second digit represents the feature update version number, and the last digit represents the revision version number. The retrieval mapping table records the solution identifier "7a8b9c0d", version number "1.0.0", timestamp "20240108143025", data table identifier "EIS_DATA", and index type "TIME_INDEX". Data backup utilizes a distributed storage mechanism, creating document copies on three different storage nodes. The primary node resides on the local storage server and is identified as "NODE_001". The two backup nodes are located in remote data centers and are identified as "NODE_002" and "NODE_003". Each copy records a synchronization timestamp for tracking data consistency. When the data on the primary node changes, the data content on the backup nodes is updated through an asynchronous replication mechanism.

[0037] S107. In the actual tiered utilization process, according to the established tiered utilization plan for retired batteries, real-time working data of retired power batteries under the corresponding utilization scenarios are collected, including voltage, current, and temperature. Through a preset anomaly detection model, it is determined whether the battery is in an abnormal state.

[0038] The battery management module uses an acquisition unit to obtain the operating parameters of the retired battery, including positive and negative electrode voltages, charge / discharge currents, surface temperature, and ambient temperature. The battery state of charge (SOC) value is obtained using the current integration method based on the charge / discharge current. A Butterworth low-pass filter is used to obtain filtered data based on the operating parameters, and this filtered data is then processed by a Kalman filter to obtain a smoothed data sequence. A sliding window method is used to calculate three indicators on the smoothed data sequence: voltage change rate, temperature change rate, and current fluctuation rate. Based on these three indicators and the battery SOC value, a weighted summation method is used to obtain the thermal runaway risk, overcharge / over-discharge risk, and temperature anomaly risk. These risk levels are then input into a pre-trained long short-term memory (LSTM) network to obtain risk prediction values. If the risk prediction values ​​exceed a preset threshold, an isolated forest algorithm is used to analyze the abnormal parameter sequence and obtain abnormal alarm data.

[0039] Specifically, the battery management module's acquisition unit collects the operating parameters of retired batteries at a fixed sampling period, including positive and negative electrode voltages, charge / discharge currents, surface temperature, and ambient temperature. The battery's state of charge (SOC) is calculated using the current integration method, and a Butterworth low-pass filter is used to reduce noise in the collected data. The validity of the filtered data is verified by setting a detection interval based on battery specifications, eliminating invalid data outside the interval, calculating the parameter change rate between adjacent sampling points, and smoothing the data sequence using a Kalman filter. A sliding window method is used to calculate three indicators: voltage change rate, temperature change rate, and current fluctuation rate. Combined with the SOC value, a weighted summation method is used to calculate three risk indicators: thermal runaway risk, overcharge / over-discharge risk, and temperature anomaly risk. These risk indicators are arranged in a time series to construct a feature sequence, which is then input into a pre-trained long short-term memory (LSTM) network. This network includes an input layer, hidden layers, and an output layer, with each output layer node corresponding to a predicted risk value. Risk indicator thresholds are determined based on historical anomaly data statistics. If the predicted risk value exceeds the threshold, anomaly detection is triggered. An exponentially weighted average method is used to perform time-series analysis on parameter data prior to the anomaly. An isolated forest algorithm is employed to analyze the anomaly parameter sequence, calculate the anomaly score for each parameter point, compare the anomaly score with a preset threshold, locate key anomalies, and generate alarm data containing anomaly type, anomaly parameters, and anomaly time. Based on the anomaly type in the alarm data, the corresponding parameter backtracking time interval is selected, and the original parameter data within that time interval is extracted. A data dimensionality reduction method is used to extract features from the multi-dimensional parameter data, generating anomaly tracing data. During the cascade utilization of retired power batteries, their operating status needs to be monitored in real time. Taking a certain model of ternary lithium battery as an example, the sampling period is set to 100 milliseconds, collecting parameters such as voltage, current, and temperature. Noise interference exists in the original collected data; for example, the voltage value fluctuates around 3.8 volts with a fluctuation amplitude reaching 50 millivolts. A fourth-order Butterworth low-pass filter is used for noise reduction, with a cutoff frequency set to 5 Hz. After filtering, the voltage fluctuation is reduced to within 5 millivolts. Data validity verification is based on battery specification parameters, setting the detection range as follows: effective voltage range of 2.5 to 4.2 volts, current limit of twice the rated value, and temperature range of -20 to 60 degrees Celsius. Data outside these ranges are discarded; for example, a collected temperature value of 85 degrees Celsius is significantly outside the normal range. A Kalman filter is used to smooth the data sequence, with a prediction error covariance of 0.1 and a measurement error covariance of 0.2, resulting in a continuously smooth parameter curve. Risk indicators are calculated using a 60-second sliding window, within which the voltage change rate, temperature change rate, and current fluctuation rate are calculated. When the temperature rises by 8 degrees Celsius within 10 minutes, the voltage drops by 0.2 volts, and the current fluctuation exceeds 0.5 times the rated value, it indicates a potential risk of thermal runaway. Similarly, when the voltage exceeds 4.15 volts or falls below 2.8 volts, it indicates an overcharge / over-discharge risk. Temperature anomaly judgment is based on the temperature mean and variance.The Long Short-Term Memory (LSTM) network employs a hidden layer containing 64 neurons, with input features being a sequence of risk indicators from the most recent 30 minutes. Network training data is derived from historical anomaly cases, including typical fault modes such as overcharging, over-discharging, and thermal runaway. Anomaly detection is triggered when the predicted risk value exceeds 0.8, and parameter backtracking analysis is performed using a 30-minute time window. The Isolation Forest algorithm locates anomalous parameters, with a sampling size of 256 and a tree depth of 8, calculating anomaly scores for the parameter sequences. Parameter points with scores exceeding 0.6 are marked as anomalies, and the specific time and parameter value of the anomaly are recorded. For example, if a detection reveals a rapid 10-degree Celsius increase in battery temperature within 5 minutes, accompanied by a 0.3-volt drop in voltage and drastic current fluctuations, the system identifies this as a risk of thermal runaway and generates an alarm message containing the anomaly type, time, and parameters. Backtracking analysis extracts parameter data from the 60 minutes prior to the anomaly, and principal component analysis (PCA) is used to reduce the dimensionality of multi-dimensional data such as voltage, current, and temperature, retaining feature components with a contribution rate exceeding 95%. By restoring the data, it was found that 30 minutes before the anomaly occurred, the battery had already shown signs of a slow temperature rise and increased internal resistance.

[0040] S108. If an abnormal state of the battery is detected, an early warning mechanism is triggered, and corresponding safety protection measures are taken according to the preset processing strategy, including cutting off the power supply and starting the cooling system.

[0041] Based on the abnormality type, abnormal parameter value, and abnormality severity recorded in the battery abnormality alarm data, the abnormality level is determined by referring to the preset abnormality level classification table, and the corresponding protective measure instruction sequence is read from the processing strategy database. The main controller generates a standard format control instruction package according to the protective measure instruction sequence, and converts the control instruction package into an execution unit control signal. The execution unit implements temperature control and overcharge / over-discharge protection measures, and the status feedback acquisition module acquires the execution unit action status data stream, which includes relay opening / closing feedback and the working status of the cooling system. Based on the action status data stream, a Bayesian network node is constructed to calculate the probability of successful execution of the protective measures, and an abnormality handling report is generated using a lossless compression algorithm and stored in the archive database.

[0042] Specifically, based on the abnormality type, abnormal parameter value, and abnormality severity recorded in the battery abnormality alarm data, the abnormality level is determined by comparing it with a preset abnormality level classification table. The corresponding protective measure instruction sequence is read from the processing strategy database, generating a processing strategy list containing instruction priorities. A master-slave control structure is used to distribute control instructions to each execution unit. The master controller generates a standard format control instruction package according to the processing strategy list, and the control instructions are converted into control signals for the execution units through a protocol conversion module. For temperature abnormality protection measures, a cooling control parameter sequence is generated according to the cooling power level, cooling fan speed, and cooling cycle duration. The parameter sequence is converted into pulse width modulation signals by a digital signal processor. For overcharge and over-discharge protection measures, relay disconnection signals and equalization circuit conduction signals are generated. Opto-isolation modules are used to electrically isolate the control signals, and execution signals are output through a power driver. A status feedback acquisition module records the action status of the execution units, including relay opening / closing feedback, cooling system operating status, cooling fan speed value, and actuator action time, generating an execution status data stream. A Bayesian network of nodes is constructed based on the execution status data stream, including nodes for execution action completion, execution timing compliance, and execution effect achievement. The probability of successful implementation of protective measures is calculated through probabilistic reasoning. The execution data of these protective measures is recorded, including anomaly triggering conditions, control command content, execution action parameters, and execution effect data. An anomaly handling report is generated using a lossless compression algorithm and stored in an archive database. The execution of anomaly protection measures for retired power batteries requires strict control timing and reliable execution verification. Taking a ternary lithium battery pack in an energy storage power station as an example, when the battery temperature rises by 12 degrees Celsius within 5 minutes, accompanied by a rapid voltage drop of 0.3 volts, the system determines a risk of thermal runaway, with an anomaly level of Level 1. At this time, a protection command sequence is read from the processing strategy database, including three commands: disconnecting the charging / discharging circuit, initiating forced cooling, and switching to backup power, with priorities of 1, 2, and 3, respectively. The main controller adopts a multi-task parallel processing architecture, converting control commands into standard format control frames. Each control frame includes the command type, target address, action parameters, and checksum. The control frames are converted into the dedicated protocols of each execution unit via a protocol conversion module; for example, the refrigeration controller uses the Modbus protocol, and the relay controller uses the CAN protocol. To address the risk of thermal runaway, the control parameters of the refrigeration system include compressor power, coolant flow rate, and cooling fan speed. The compressor power is set to 90% of its rated value, corresponding to an output frequency of 45 Hz; the coolant flow rate is set to 20 liters per minute; and the cooling fan speed is set to 3000 revolutions per minute. These parameters are converted into pulse-width modulated signals with an 80% duty cycle and a frequency of 20 kHz by a digital signal processor. Power cutoff uses a dual-relay series structure, with the drive signals of the two relays achieving 4000 volt electrical isolation via an optocoupler.The relay coil drive voltage is 12 volts, and the contact current is 50 amps. The relay operation is controlled by the gate drive signal output by the power driver. To prevent arcing upon disconnection, a resistor-capacitor (RC) snubber circuit is connected in parallel with the relay contacts. The execution status feedback acquisition module records the operation status of each execution unit. The relay status is fed back through auxiliary contacts, the cooling system through temperature, pressure, and flow sensors, and the cooling fan through a speed sensor. A Bayesian network uses a three-layer structure to evaluate the execution effect. The bottom layer nodes correspond to the completion status of specific execution actions, the middle layer nodes represent the degree of compliance with the execution timing, and the top layer nodes reflect the overall protection effect. When all thermal runaway protection measures are executed, the battery temperature drops by 8 degrees Celsius within 10 minutes, and the voltage tends to stabilize. The system calculates the success probability of the protection measures to be 0.95. The anomaly handling report records data such as triggering conditions, control process, and execution effect. The report is compressed to 40% of its original size using Huffman coding before being archived. This data provides an important basis for subsequent optimization of protection strategies.

[0043] Based on the embodiments of the present invention described above, and through the above description, those skilled in the art can make various changes and modifications without departing from the technical concept of the present invention. The technical scope of the present invention is not limited to the contents of the specification, but must be determined according to the scope of the claims.

Claims

1. A method for predicting the remaining life of retired power battery cells, characterized in that, The method includes: Obtain basic parameter information of the cell to be tested in the retired power battery, including cell model, capacity and internal resistance, and obtain the standard impedance spectrum data of the cell model from the preset cell parameter database according to the cell model; For the cells under test in retired power batteries, impedance tests are performed at different frequencies within a preset test frequency range to obtain impedance response data at different frequencies and obtain impedance spectrum data. The acquired impedance spectrum data is preprocessed, including removing outliers and smoothing the data, extracting key feature parameters of the impedance spectrum curve, including the radius of the arc, the coordinates of the center of the arc, and the slope, and constructing the impedance spectrum feature vector. The mean and standard deviation of each feature component are calculated based on the impedance spectrum feature vector, and a standardized feature vector is obtained by using the Gaussian distribution standardization method. Singular value decomposition is performed on the standardized feature vector, and the feature components whose cumulative contribution rate exceeds the contribution rate threshold are retained to obtain a dimensionality-reduced feature vector. This dimensionality-reduced feature vector is then input into a preset deep neural network, and a nonlinear transformation is performed on the output values ​​of each layer node to obtain the predicted remaining life of the battery cell. Fuzzy clustering is used to calculate the membership degree of the predicted remaining life of the battery cell to each level. The grading results are verified by calculating the Mahalanobis distance between the battery cell feature vector and the sample centers of each level, thus obtaining the final battery cell grading data. Obtain the utilization scenario conditions corresponding to different tiers, including temperature range and charge / discharge rate, and match the target tier utilization scenario based on the tier level to which the current test cell belongs. The target tiered utilization scenarios are associated with and stored with retired battery cell model, capacity, internal resistance and impedance spectrum data to form a tiered utilization scheme for retired batteries; In the actual tiered utilization process, according to the established tiered utilization plan for retired batteries, real-time working data of retired power batteries in the corresponding utilization scenarios are collected, including voltage, current and temperature. Through a preset anomaly detection model, it is determined whether the battery has an abnormal state. If an abnormal state of the battery is detected, an early warning mechanism is triggered, and corresponding safety protection measures are taken according to the preset handling strategy, including cutting off the power supply and starting the cooling system.

2. The method according to claim 1, characterized in that, The process involves acquiring basic parameter information of the cells to be tested in retired power batteries, including cell model, capacity, and internal resistance. Based on the cell model, standard impedance spectrum data for that model is retrieved from a pre-set cell parameter database, including: The design specifications are obtained from a pre-set database based on the identification of retired power cells. The design specifications include nominal capacity, rated voltage and standard internal resistance. For retired power battery cells, a constant current and constant voltage method is used to charge them to the rated voltage value, and the cells are then left to stand for a preset time to obtain a stable state. The raw impedance data of the stable-state battery cell was obtained by frequency sweep measurement using an electrochemical impedance spectroscopy instrument, and the real and imaginary impedance numerical sequences were generated by fast Fourier transform. The impedance numerical sequence is validated for data validity, outlier data points are removed, and cubic spline interpolation is used to complete the missing data to generate a continuous impedance curve.

3. The method according to claim 1, characterized in that, The method involves performing impedance tests on the cells under test in retired power batteries at different frequencies within a preset test frequency range, acquiring impedance response data at different frequencies, and obtaining impedance spectrum data, including: The retired power cells are charged with constant current and constant voltage until the charging cutoff voltage value is reached. After charging, the cells are discharged with rated current using a constant current power supply until the discharge cutoff voltage value is reached. After discharge, the open-circuit voltage of the battery cell is monitored. The battery cell is determined to be in a stable state if the open-circuit voltage fluctuation value is less than the preset voltage fluctuation threshold. For a battery cell in a stable state, a sinusoidal AC excitation signal is applied using an electrochemical impedance spectroscopy instrument, and the response of the excitation signal is subjected to Fourier spectrum transformation to obtain an impedance value sequence. The impedance data curve is obtained by performing point supplementation on the impedance numerical sequence using cubic spline interpolation. The position of the semi-circular feature point is identified based on the impedance data curve to obtain the impedance spectrum feature dataset.

4. The method according to claim 1, characterized in that, The preprocessing of the acquired impedance spectrum data includes removing outliers and smoothing data, extracting key feature parameters of the impedance spectrum curve, including the radius of the arc, the coordinates of the center, and the slope, and constructing an impedance spectrum feature vector, including: The local variance of the impedance spectrum data sequence is calculated using a sliding window. If the variance exceeds the threshold of the mean variance of all variances in the frequency range, the marked outlier data points are removed from the impedance spectrum data sequence. The interval between adjacent frequency points is calculated based on the impedance spectrum data sequence after removing outliers. If the interval exceeds the preset frequency step size, a linear interpolation method is used to obtain supplementary frequency point data. For the impedance spectrum sequence after supplementing the frequency point data, high-frequency noise components are removed by Butterworth low-pass filter and four-layer discrete wavelet decomposition, and a continuous and smooth impedance spectrum curve is obtained by cubic spline function. Based on the continuous and smooth impedance spectrum curve, circular and linear fitting are performed on the complex plane. The coordinates of the center and the radius are obtained by iterative least squares method. The real and imaginary parts of the low-frequency slope are obtained by complex number operations. The eigenvectors are then normalized.

5. The method according to claim 1, characterized in that, The process involves obtaining the utilization scenario conditions corresponding to different tiers, including temperature range and charge / discharge rate, and combining these with the tier level to which the current battery cell under test belongs to match the target tier utilization scenario, including: The tiered utilization scenario table is read from the preset scenario parameter database. The tiered utilization scenario table includes temperature range limits, charge / discharge rate limits, voltage range limits, cycle depth limits, and operating time limits. Based on the aforementioned tier levels, a judgment matrix is ​​constructed using the limit parameters in the scenario table, and the weight coefficients of the limit parameters are calculated using the hierarchical analysis method. Operating parameters for energy storage, power, and backup power scenarios are extracted based on the weighting coefficients, and scenario adaptability scores are obtained using data statistical methods. The membership value of the battery cell under test to each scenario is calculated based on the scenario adaptability score, and the performance parameters of the battery cell under test are verified by the Monte Carlo method to see if they meet the scenario requirements.

6. The method according to claim 1, characterized in that, The process of associating and storing target tiered utilization scenarios with retired battery cell model, capacity, internal resistance, and impedance spectrum data to form a tiered utilization scheme for retired batteries includes: The original identification string is generated by concatenating the model identifier, capacity value, internal resistance value and impedance spectrum data, and the scheme identification code is obtained by using a secure hash algorithm; Based on the scheme identification code, a parameter information table, an impedance spectrum data table, a grade information table, and a scene information table are established, with the scheme identification code serving as the primary key to establish the association between the data tables. A null value detection method is used to identify missing data items in the data table, and anomaly data markers are obtained through a data type verification method. A data quality report is then generated based on the anomaly data markers. A structured query statement is used to extract complete data from the data table, and a retired battery tiered utilization scheme document containing a basic information area, a feature data area, and a scenario parameter area is generated based on the complete data.

7. The method according to claim 1, characterized in that, In the actual tiered utilization process, according to the established tiered utilization plan for retired batteries, real-time operating data of retired power batteries under corresponding utilization scenarios are collected, including voltage, current, and temperature. A preset anomaly detection model is used to determine whether the battery is in an abnormal state, including: The operating parameters of the retired battery are obtained by the acquisition unit in the battery management module. The operating parameters include positive and negative electrode voltage values, charging and discharging current values, surface temperature values, and ambient temperature values. The battery state of charge value is obtained by using the current integration method based on the charge and discharge current value. Based on the operating parameters, a Butterworth low-pass filter is used to obtain filtered data, and the filtered data is processed by a Kalman filter to obtain a smoothed data sequence. The sliding window method is used to calculate three indicators: voltage change rate, temperature change rate, and current fluctuation rate, for the smoothed data sequence. Based on the three indicators and the battery state of charge value, the risk of thermal runaway, overcharge and over-discharge, and abnormal temperature are obtained by weighted summation. The risk levels of thermal runaway, overcharge and over-discharge, and abnormal temperature are input into a pre-trained long short-term memory network to obtain risk prediction values. If the risk prediction values ​​exceed a preset threshold, the isolated forest algorithm is used to analyze the abnormal parameter sequence to obtain abnormal alarm data.

8. The method according to claim 1, characterized in that, If an abnormal battery condition is detected, an early warning mechanism is triggered, and corresponding safety protection measures are taken according to a preset processing strategy, including cutting off the power supply and activating the cooling system, including: Based on the abnormality type, abnormal parameter value and abnormality degree recorded in the battery abnormality alarm data, the abnormality level is determined by referring to the preset abnormality level classification table, and the corresponding protective measure instruction sequence is read from the processing strategy database. The main controller generates a standard format control instruction package according to the instruction sequence of the protection measures, and converts the control instruction package into an execution unit control signal; Temperature control and overcharge / over-discharge protection measures are implemented by the execution unit. The status feedback acquisition module is used to obtain the action status data stream of the execution unit. The action status data stream includes relay opening and closing feedback and the working status of the refrigeration system. Based on the action state data stream, a Bayesian network node is constructed to calculate the probability of successful execution of the protection measures, and an anomaly handling report is generated using a lossless compression algorithm and stored in the archive database.

Images