A low-temperature exposure lithium battery aging damage diagnosis method based on electrochemical nondestructive characterization

Abstract

The application discloses a low-temperature exposure lithium battery aging damage diagnosis method based on electrochemical nondestructive characterization. First, the electrochemical response data of the battery to be tested is obtained through electrochemical impedance spectroscopy (EIS) test and incremental capacity analysis (ICA). Second, key characteristic parameters capable of representing the battery ohmic resistance, charge transfer resistance, double-layer capacitance, phase transition peak position, peak area, half-peak width and the like are extracted. Then, a joint prediction model based on machine learning is trained by using the extracted multi-dimensional features, which is used to judge whether the battery to be tested has experienced low-temperature exposure and the current aging damage degree. Through the nondestructive detection means, the application can effectively distinguish the unique aging path introduced by normal aging and low-temperature exposure, realize accurate "traceability" of the battery historical damage and quantitative evaluation of the current state, and provide reliable technical support for intelligent decision of a battery management system (BMS), battery safety warning and echelon utilization.

Description

Technical Field

[0001] This invention relates to the field of lithium-ion battery technology used in electric vehicles and energy storage devices, and in particular to a method for diagnosing aging damage of low-temperature exposed lithium batteries based on electrochemical non-destructive characterization. Background Technology

[0002] With the widespread adoption of new energy vehicles and mobile electronic products, lithium-ion batteries, as key energy storage devices, have received considerable attention for their stability and lifespan. At low temperatures, the electrochemical reaction kinetics within lithium-ion batteries slow down significantly, and the ionic conductivity of the electrolyte drops sharply. If exposed to low temperatures for extended periods, lithium ions will rapidly and unevenly deposit on the negative electrode surface, inducing the formation of lithium dendrites. This not only consumes active lithium, causing capacity loss, but may also puncture the separator, leading to internal short circuits and thermal runaway.

[0003] Existing research on low-temperature damage to lithium batteries focuses on the aging path of batteries during charge and discharge in low-temperature scenarios. However, in real-world scenarios, new energy vehicles have battery thermal management systems that can heat the battery to a suitable operating temperature, but they cannot protect against or identify battery damage caused by low-temperature exposure. Therefore, in-depth research on the performance degradation mechanism and diagnostic methods for aging damage of lithium-ion batteries after low-temperature exposure has significant theoretical and practical value. Summary of the Invention

[0004] The purpose of this invention is to propose a method for diagnosing aging damage of lithium batteries exposed to low temperatures based on electrochemical non-destructive characterization. Based on electrochemical non-destructive characterization technology, this invention designs and develops a method to diagnose the degree of low-temperature damage of batteries, taking into account the multivariate signal differences between lithium-ion batteries that have undergone low-temperature exposure and ordinary batteries. This method identifies the damage of lithium batteries after low-temperature exposure and quantifies their aging degree, providing a reference for subsequent battery management systems (BMS) and battery recycling.

[0005] To address the problems existing in the prior art, this invention adopts the following technical solution: a method for diagnosing aging damage of low-temperature exposed lithium batteries based on electrochemical non-destructive characterization, comprising the following steps:

[0006] Data acquisition: Acquire electrochemical impedance spectroscopy (EIS) data and related test data of the open-circuit voltage (OCV) of the lithium battery under test;

[0007] Feature engineering: The EIS data and OCV-related test data are processed to extract electrochemical impedance spectroscopy characteristic parameters and incremental capacity analysis characteristic parameters;

[0008] Diagnostic assessment: The extracted EIS feature parameters and ICA feature parameters are used as inputs to a pre-trained joint prediction model. The model outputs a diagnostic result to determine whether the lithium battery under test has experienced low-temperature exposure, and an assessment result to quantify its current degree of aging damage.

[0009] Furthermore, the EIS feature parameters in the feature engineering step are obtained by constructing an equivalent circuit model ECM and fitting the EIS data. The EIS feature parameters include at least one of the ohmic internal resistance Rs, charge transfer resistance Rct, and double-layer capacitance Cdl.

[0010] Furthermore, the ICA feature parameters in the feature engineering step are extracted from the incremental capacity curve obtained by processing the OCV-related test data. The ICA feature parameters include the voltage position of the phase transition peak, the peak height, the area under the peak, and the full width at half maximum (FWHM).

[0011] Furthermore, the joint prediction model is a machine learning model, which has at least one classification output for outputting the diagnostic result and at least one regression output for outputting the evaluation result.

[0012] Furthermore, the machine learning model is a multilayer perceptron (MLP) model.

[0013] Furthermore, the diagnostic result is a probability value or classification label used to characterize the likelihood that the lithium battery under test has experienced low-temperature exposure; the evaluation result is a quantitative value used to characterize the degree of aging or capacity decay rate of the lithium battery under test.

[0014] Furthermore, the method also includes a step of training the joint prediction model, the training step comprising:

[0015] i. Construct a training dataset containing EIS and ICA feature parameters of batteries from multiple sets of known low-temperature exposure history and aging state, as well as their corresponding low-temperature exposure labels and aging state labels;

[0016] ii. Using the training dataset, the parameters of the joint prediction model are trained by optimizing a total loss function, which is a weighted combination of the loss function for the classification task and the loss function for the regression task, to determine whether the user has experienced low-temperature exposure.

[0017] Furthermore, the training step also includes a step of expanding the training dataset using data augmentation techniques.

[0018] Furthermore, the voltage and current data for OCV testing are obtained by performing a full charge-discharge cycle on the battery at a low rate.

[0019] Furthermore, the EIS data is obtained by scanning within a preset frequency range by applying an AC excitation signal when the battery is in a preset state of charge (SOC).

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0021] (1) The present invention diagnoses the low-temperature exposure aging damage of lithium batteries by electrochemical non-destructive characterization. Through in-situ analysis methods such as electrochemical impedance spectroscopy (EIS) and incremental capacity curve (ICV), the damage of the battery after low-temperature exposure and the diagnosis of the degree of aging can be identified without damaging the battery.

[0022] (2) After charge-discharge cycle aging, batteries exposed to low temperatures often experience more severe aging and damage than normal batteries, and they have different aging mechanisms and degradation paths than normal batteries. The cracking phenomenon at the cathode particle level after low-temperature exposure cannot be macroscopically displayed, and it is necessary to distinguish whether lithium batteries have been damaged by low-temperature exposure in reality. Therefore, this method has engineering significance.

[0023] (3) This invention focuses on the impact of low temperature exposure on the aging damage of lithium batteries. Through low temperature exposure aging experiments, in-situ battery analysis and testing, and battery disassembly experiments, it reveals the intrinsic mechanism of battery aging under low temperature conditions, providing a theoretical basis for performance optimization and life extension of lithium batteries under low temperature exposure conditions. Attached Figure Description

[0024] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0025] Figure 1 A flowchart of a low-temperature exposed lithium battery aging damage diagnosis method based on electrochemical non-destructive characterization provided by the present invention;

[0026] Figure 2 A flowchart of a low-temperature exposure aging experiment for lithium-ion batteries is provided for this invention.

[0027] Figure 3 A comparison chart of incremental capacity curves provided for embodiments of the present invention;

[0028] Figure 4This is a comparison chart of EIS test curves provided in an embodiment of the present invention;

[0029] Figure 5 This is a diagram showing the diagnostic results of low-temperature exposure damage provided in an embodiment of the present invention;

[0030] Figure 6 This is a scanning electron microscope image from an embodiment of the present invention. Detailed Implementation

[0031] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Those skilled in the art can make similar extensions without departing from the spirit of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0032] Example 1:

[0033] This embodiment uses lithium-ion battery cells with the parameters described in Table 1.

[0034] Table 1

[0035]

[0036] It's crucial to distinguish between reversible effects and potential irreversible damage during low-temperature exposure of lithium-ion batteries. When the battery recovers from low temperature to room temperature, most reversible effects directly caused by temperature (such as increased electrolyte viscosity and transient high resistance due to decreased ion mobility) will subside. However, if improper handling (such as low-temperature charging) or prolonged extreme low temperatures accompany low-temperature exposure, permanent damage may occur. This damage will be reflected in the EIS and ICA characteristics after recovery to room temperature.

[0037] The changes in electrochemical impedance spectroscopy (EIS) are as follows:

[0038] After returning to room temperature, the changes in the EIS spectrum mainly reflect the irreversible damage that may be caused by low-temperature exposure:

[0039] The continuous increase in the semicircle diameter (R and Rctt) in the mid-frequency region is mainly due to the hindered migration of lithium ions on the negative electrode surface at low temperatures. If the battery is charging or self-discharging, lithium plating is highly likely to occur. The deposited lithium metal reacts with the electrolyte in subsequent cycles, forming a thicker, denser, and more resistive SEI (solid electrolyte interface) film. On the EIS, this is reflected in the Nyquist plot as a larger semicircle diameter in the mid-frequency region compared to a normal battery, indicating a permanent increase in the battery's interfacial impedance and charge transfer impedance. Even if the temperature returns to normal, this structural change is irreversible.

[0040] The ohmic impedance (Rohm) in the high-frequency region may increase slightly: this is mainly because if low-temperature exposure causes partial leaching, degradation, or consumption of electrolyte components (e.g., reaction with deposited lithium), the ionic conductivity of the electrolyte may permanently decrease. Additionally, severe lithium deposition leading to the shedding of current collectors or electrode active materials will also increase the ohmic impedance. EIS will show the intersection of the high-frequency region and the real axis may be slightly shifted to the right compared to a normal battery.

[0041] The changes to Incremental Capacity Analysis (ICA) are as follows:

[0042] After returning to room temperature, the changes in the ICA curves primarily reveal battery capacity loss and permanent alterations to the reaction pathway:

[0043] A significant decrease in total capacity (reduced area under the dQ / dV curve): The main reason is the formation of dead lithium. When metallic lithium deposited at low temperatures returns to room temperature, some react with the electrolyte to form an irreversible SEI (Sediment Ion), while some may lose electrical contact with the negative electrode, forming "dead lithium." This portion of lithium no longer participates in reversible charge-discharge reactions, leading to the loss of active lithium ions. The formation of a thickened SEI film consumes active lithium ions within the battery, resulting in a decrease in reversible capacity. Severe lithium deposition may be accompanied by structural damage or stripping of the negative electrode material, causing some active materials to permanently lose their activity. ICA is manifested as a significantly smaller total area under the dQ / dV curve compared to a normal battery, directly reflecting the permanent degradation of battery capacity.

[0044] The decrease in peak height and increase in peak width are mainly due to capacity loss, which reduces the amount of active material and active lithium ions involved in the reaction, making the peaks reflecting specific phase transitions or reactions less sharp. Simultaneously, changes in the SEI film and lithium deposition may cause the electrochemical reaction to become non-uniform. This manifests as a lower and wider peak on the ICA curve, reflecting reduced activity and decreased uniformity of the internal battery reaction.

[0045] Slight shifts in peak position or morphology are possible: The main reasons are the newly formed, thicker, or altered SEI film, and the loss of active lithium ions, which may slightly change the electrochemical reaction potential of the electrode, leading to a slight shift in peak position. In cases of severe structural damage, some reaction pathways may even change. ICA manifests as a slightly larger voltage difference between charge and discharge peaks than in a normal battery, even though the kinetic overpotential recovers at room temperature due to the permanent increase in internal resistance. Minor changes in the relative height or position of certain peaks may also reflect changes in the battery's internal state.

[0046] Based on this, such as Figure 1 As shown, the present invention provides a method for diagnosing aging damage of low-temperature exposed lithium batteries based on electrochemical non-destructive characterization, comprising:

[0047] S1: Collect experimental data and design a systematic experiment for the battery under test, including low-temperature exposure experiments and cycle aging experiments under different operating conditions. At key experimental points, such as before and after low-temperature exposure and specific cycles during cycle aging, perform health status characterization tests on the battery, mainly including open-circuit voltage (OCV) testing and electrochemical impedance spectroscopy (EIS) testing, to obtain multi-dimensional EIS data of the lithium battery under test and test data related to its open-circuit voltage (OCV); such as Figure 2 As shown;

[0048] The overall experimental procedure consists of battery pretreatment, low-temperature exposure experiment, battery aging experiment, OCV test, and EIS test.

[0049] S101: Battery pretreatment experiment screens batteries to address battery consistency issues, ensuring minimal deviation in experimental battery characteristics.

[0050] The OCV calibration cycle method is as follows: first, discharge the battery at 1C constant current to the discharge cutoff voltage of 2.5V, let it stand for 15 minutes, then charge it at 0.04C constant current and constant voltage to the charging cutoff voltage of 4.2V, let it stand for 30 minutes, then discharge it at 0.04C constant current to the cutoff voltage, and record the capacity and voltage change data.

[0051] Electrochemical impedance spectroscopy (EIS) utilizes a small-amplitude AC excitation signal to perturb an electrochemical system in equilibrium, followed by processing and analysis of the response signal. The cell state of charge (SOC) was set to 50%, the frequency range to 0.01-1000 Hz, with 10 points per decibel, an excitation current of 500 mA, and a dual-electrode mode. The horizontal axis represents the real part of the impedance (Z'), and the vertical axis represents the imaginary part of the impedance (-Z''), resulting in a Nyquist plot.

[0052] S102: The battery exposure test uses a high and low temperature chamber. The low temperature environment is set to -20℃. The battery is placed in the battery box and placed in the high and low temperature chamber. The low temperature exposure is set to 0 / 12 / 24 / 36 / 48 hours. After completion, the battery is taken out and placed in a 25℃ constant temperature chamber to restore room temperature for 2 hours. Finally, the battery is taken out and connected to a charge and discharge instrument to complete the low temperature exposure test.

[0053] S102: The battery aging test uses a battery charge / discharge instrument connected to a host computer to control the battery charge / discharge cycles. Additionally, the tested battery is connected externally to a constant temperature chamber at 25°C for aging cycles. Specifically, the aging cycle involves allowing the battery to rest for 15 minutes, then discharging it at a constant current of 2 / 1 / 0.5C to the cutoff voltage, followed by a 15-minute rest, then charging it at a constant current of 2 / 1 / 0.5C to the cutoff voltage, and finally charging it at a constant voltage of 4.2V to the cutoff current of 0.02C. This cycle is repeated 10 times. A capacity calibration cycle is performed after every 10 aging cycles. After three full cycles, an OCV calibration test and an EIS test are performed, for a total of 120 aging cycles. OCV and EIS tests are performed on all batteries before and after low-temperature exposure and after every 30 aging cycles.

[0054] S2: Data Processing: Data processing includes voltage, current, capacity, and impedance data. This involves cleaning the data, removing outliers, normalizing it, extracting impedance parameters from the EIS curve, converting the incremental capacity curve and extracting features, and using data augmentation techniques to expand the data and construct a dataset.

[0055] S201: Processes data from the OCV calibration test, including real-time current. Initial maximum capacity According to the SOH calibration, the current battery capacity meets the following requirements:

[0056]

[0057] like Figure 3 As shown: The Incremental Capacity (ICA) curve is formed by the amount of electricity... For voltage Differential calculations were performed. To avoid excessively high sampling frequencies causing large peaks in the differential curve that would make peak identification impossible, a 50-point smoothing sampling process was used. The ICA curve is the smoothed differential of charge versus voltage from the OCV test data, and the calculation satisfies:

[0058]

[0059] When the capacity released or charged by a lithium battery within a unit voltage range increases (i.e., the plateau region, representing an electrochemical reaction process), it indicates that the capacity released or charged by the lithium battery has increased. The value will increase, and the curve will show the characteristic of a "peak". The peak value corresponds to the phase transition process of the electrochemical reaction inside the battery. The changes in the position, height, area and shape of the peak reflect aging mechanisms such as loss of active material (LAM), lithium inventory loss (LLI), and side reaction formation.

[0060] Compared to normal batteries, the incremental capacity curve of NMC811 batteries exposed to low temperatures shows obvious peak decay, peak position shift, and multiple peaks, which characterize aging mechanisms such as lithium plating, electrode particle cracking, and LAM loss.

[0061] Extracting ICA features: Identifying the voltage positions and peak sizes corresponding to the three phase transition peaks H1, H2, and H3, capturing the shift caused by LAM / LLI and the remaining active material; calculating the area covered by the ICA curve by integration, correlating the reversible capacity, and measuring the half-width at half-maximum (FWHM) of the largest characteristic peak to quantify the kinetic decay.

[0062] S202: Utilize an electrochemical workstation to measure the electrochemical impedance spectroscopy (EIS) information of the battery, and perform data processing on the complex impedance at each single point to extract features.

[0063] like Figure 4 As shown: Based on the EIS data of S101, Fourier analysis yielded the complex impedance. With real axis virtual axis Plot a Nyquist plot, with the ohmic impedance at the first intersection on the real axis and the charge transfer impedance approximated as the diameter of a semicircle. Fit the plot to an equivalent circuit model to obtain polarization resistance (Rp), charge transfer resistance (Rct), and double-layer capacitance (Cdl) data as electrochemical characteristics.

[0064] An equivalent circuit model (ECM) is constructed, taking into account both data processing speed and sensor accuracy. A first-order RC model is selected, and its state-space equation is:

[0065]

[0066]

[0067] In the formula, For SOC, Polarization voltage, For current, Terminal voltage, Where C is the open-circuit voltage and C is the rated capacity. For ohm resistance, These are polarization resistors and capacitors.

[0068] The parameters of the ECM model, namely ohmic resistance Rs, charge transfer impedance Rct, and double layer capacitance Ddl, were calculated using nonlinear least squares fitting (CNLS) EIS curves and entered into the dataset as electrochemical features.

[0069] S203: The method for data set operations is as follows;

[0070] Missing values: If the missing rate of a feature is >5%, the entire row is removed; otherwise, it is filled with the median of the column.

[0071] Outliers: Calculate a robust Z-score for each column: If | If so, then remove that row. Indicates the dataset number Liede Row data, Indicates the first The median of all data in the column. express The absolute deviation of the median of all data points in the column.

[0072] Normalization: All features are scaled to the [0, 1] interval using min-max scaling. .

[0073] Data Augmentation: To address the scarcity of low-temperature exposure aging data, data augmentation techniques are employed to appropriately add noise and scale the acquired data to generate new data. Gaussian noise augmentation is used. Three noisy copies are generated for each original sample. Features after adding noise Features before noise addition Let be a Gaussian random variable. Let be the noise standard deviation, and IRQ be the interquartile range of the j-th column.

[0074] The dataset labels include: low-temperature exposure duration, low-temperature exposure temperature, aging cycle ratio, number of aging cycles, and battery SOH. Features include EIS's Rp, Rct, Cdl, and ICA's Vpeak, Area, and FWHM. 70% of the data is randomly divided into a training set and 30% into a validation set.

[0075] S3: Establish a joint prediction model for battery low-temperature exposure and aging degree based on multilayer perceptron (MLP). The extracted EIS and ICA feature parameters are used as input. The model is trained, fine-tuned, and validated using the dataset from step 2 as the training set to achieve diagnostic identification of the low-temperature exposure degree of lithium batteries and quantification of aging damage. The model outputs a diagnostic result to determine whether the tested lithium battery has undergone low-temperature exposure, and an evaluation result to quantify its current aging damage degree. Specific steps are as follows:

[0076] Considering the characteristics of battery low-temperature exposure identification and aging damage diagnosis tasks, such as weak time dependence, limited training data, high noise, and high risk of overfitting, a multilayer perceptron (MLP) model was chosen. It consists of an input layer, hidden layers, and an output layer. The input layer receives preprocessed EIS and ICA features, constructs multiple hidden layers, and learns a complex representation of the input features through a non-linear activation function (e.g., ReLU). The output layer contains two output units, used to predict the battery's low-temperature exposure level (classification y_cls) and battery aging level (regression y_reg), respectively.

[0077] The dataset was divided into a 70% training set and a 30% test set. When training the model, a binary cross-entropy loss function was used. Used for low temperature exposure prediction and mean squared error loss function It is used for predicting age-related damage degradation, and the weighted sum of the two loss functions is used as the total loss function. .in: , , These represent the classification, regression tasks, and total loss values, respectively. , For predictive labels of low temperature and degree of aging, , These are the weighting coefficients.

[0078] Training strategy: Backpropagation is used to optimize model parameters. All weights are initialized with a uniform Xavier distribution and a bias of 0. The Adam optimizer is used to adaptively adjust the learning rate, and early stopping is used to prevent overfitting. Training is terminated and the model is backtracked to the optimal weights on the training set when L_total shows no improvement for 15 consecutive epochs. The number of input nodes is the number of features, the number of hidden layers is set to 3, the number of hidden units is 64 / layer, the Dropout ratio is 0.3, and the initial learning rate is... Each batch is 64 in size.

[0079] With the above design, the model achieved a classification AUC of 0.93 ± 0.02 and a regression RMSE of ( ) on the training and validation sets. =2.1%.

[0080] S4: Based on the model, determine whether the lithium battery has ever experienced low-temperature exposure at 100% SOC, the extent of damage, and its current aging level.

[0081] For unknown samples x* collected on-site (batteries whose low-temperature exposure and aging damage levels are unknown), OCV and EIS data are collected and processed and feature extracted as described in S2 to form a test set. This set is then input into the model trained in S3 and forward propagated to obtain the low-temperature exposure probability of the battery. Degree of damage According to the threshold values ​​in the example: P_exp < 0.35 indicates "virtually no low-temperature exposure", 0.35 ≤ P_exp < 0.65 indicates "suspected exposure, requiring retesting", and P_exp ≥ 0.65 indicates "confirmed low-temperature exposure damage". If exposure is determined to have occurred, further steps are taken according to... Damage severity classification: ≥95% is considered "mild", 75% ≤ <95% is considered "moderate". <75% is classified as "severe"; specific diagnostic results are as follows: Figure 5 As shown.

[0082] To verify the accuracy of this method, a disassembly experiment was conducted on the aforementioned battery. The microstructure of the lithium battery cathode, anode, and separator was observed using a scanning electron microscope (SEM), directly verifying the accuracy of the low-temperature exposure battery identification and aging damage diagnosis method proposed in this invention. Figure 6 It was found that the cathode surface of the brand-new battery consisted of neat NMC811 particles, which remained as rough, intact spheres after 120 aging cycles. However, after 24 / 48 hours of low-temperature exposure, the cathode particles developed numerous small cracks and fragmentation. With increasing low-temperature exposure time, the dead lithium deposits on the anode surface became thicker and rougher, and a certain amount of lithium dendrites were generated. Furthermore, the porosity of the separator surface decreased with increasing low-temperature exposure time, and deposits appeared on the surface. Therefore, the accuracy of the model was verified, enabling the identification of low-temperature exposure damage and the diagnosis of aging levels in lithium batteries.

[0083] The above embodiments are used to explain and illustrate the present invention, but not to limit the present invention. Any modifications and changes made to the present invention within the spirit and scope of the claims shall fall within the protection scope of the present invention.

Claims

1. A method for diagnosing aging damage in low-temperature exposed lithium batteries based on electrochemical non-destructive characterization, characterized in that, Includes the following steps: Data acquisition: Acquire electrochemical impedance spectroscopy (EIS) data and related test data of the open-circuit voltage (OCV) of the lithium battery under test; Feature engineering: The EIS data and OCV-related test data are processed to extract EIS characteristic parameters and ICA incremental capacity analysis characteristic parameters; The specific process is as follows: EIS characteristic parameters are obtained by constructing an equivalent circuit model ECM and fitting the EIS data. The EIS characteristic parameters include at least one of ohmic internal resistance Rs, charge transfer resistance Rct, and double-layer capacitance Cdl. The ICA characteristic parameters are extracted from the incremental capacity curve obtained by processing the OCV-related test data. The ICA characteristic parameters include the voltage position of the phase transition peak, the peak height, the area under the peak, and the full width at half maximum (FWHM). Diagnostic assessment: The extracted EIS feature parameters and ICA feature parameters are used as inputs to a pre-trained joint prediction model. The model outputs a diagnostic result to determine whether the lithium battery under test has experienced low-temperature exposure, and an assessment result to quantify its current degree of aging damage.

2. The method according to claim 1, characterized in that, The joint prediction model is a machine learning model that has at least one classification output for outputting the diagnostic results and at least one regression output for outputting the evaluation results.

3. The method according to claim 2, characterized in that, The machine learning model is a multilayer perceptron (MLP) model.

4. The method according to claim 1, characterized in that, The diagnostic result is a probability value or classification label used to characterize the likelihood that the lithium battery under test has experienced low-temperature exposure; the evaluation result is a quantitative value used to characterize the degree of aging or capacity decay rate of the lithium battery under test.

5. The method according to claim 1, characterized in that, The method further includes a step of training the joint prediction model, the training step including: i. Construct a training dataset containing EIS and ICA feature parameters of batteries from multiple sets of known low-temperature exposure history and aging state, as well as their corresponding low-temperature exposure labels and aging state labels; ii. Using the training dataset, the parameters of the joint prediction model are trained by optimizing a total loss function, which is a weighted combination of the loss function for the classification task and the loss function for the regression task, to determine whether the user has experienced low-temperature exposure.

6. The method according to claim 5, characterized in that, The training steps also include a step of expanding the training dataset using data augmentation techniques.

7. The method according to claim 1, characterized in that, The voltage and current data from the OCV test are obtained by performing a full charge-discharge cycle on the battery at a low rate.

8. The method according to claim 1, characterized in that, EIS data is obtained by scanning within a preset frequency range by applying an AC excitation signal when the battery is in a preset state of charge (SOC).

Images