A multi-dimensional health state monitoring and early warning method for vehicle-mounted battery pack

Abstract

This invention provides a multi-dimensional health status monitoring and early warning method for vehicle battery packs. The method includes collecting benchmark data from the target battery pack under three monitoring modes: standard charge / discharge, real-vehicle operating conditions, and AC impedance, forming a benchmark monitoring dataset. During real-time operation, dynamic monitoring data is acquired and an iterative evaluation process is initiated. The two datasets are then matched using multi-dimensional features to generate a health matching degree. If the health threshold is not met, a step size factor is selected based on the cell type to update the data and the iteration continues. If the threshold is met, a comprehensive health degradation index and multi-dimensional health status are calculated, including individual cell health values ​​and overall battery pack inconsistency indicators. Finally, a balancing management strategy is generated and output. This invention can accurately quantify the inconsistency between the health status of individual cells and the overall battery pack, and generate differentiated active / passive balancing management strategies accordingly, improving the accuracy of battery pack health monitoring and the timeliness of early warnings.

Description

Technical Field

[0001] This invention relates to the field of battery management technology, and more specifically, to a multi-dimensional health status monitoring and early warning method for vehicle battery packs. Background Technology

[0002] With the rapid development of the new energy vehicle industry, the health status of the power battery pack, as a core component of electric vehicles, directly affects the vehicle's range, safety performance, and lifespan. Battery pack state of health (SOH) monitoring is one of the key functions of the battery management system (BMS). Accurately assessing the battery pack's health status is crucial for ensuring driving safety, optimizing charging and discharging strategies, and extending battery life.

[0003] In existing technologies, battery health monitoring mainly relies on a single monitoring mode, such as obtaining capacity decay and internal resistance growth data through standard charge-discharge tests, estimating voltage, current, and temperature characteristics based on real-vehicle operating data, or using electrochemical impedance spectroscopy (EIS) to analyze the internal electrochemical reaction mechanism of the battery. However, a single monitoring mode cannot comprehensively reflect the multi-dimensional health characteristics of the battery pack: standard charge-discharge tests have long cycles and cannot be monitored in real time; real-vehicle operating data are greatly affected by driving behavior and environmental factors, and feature extraction is complex; EIS testing requires specialized equipment and is difficult to implement online. In addition, existing methods mostly focus on the overall health status of the battery pack, lacking a refined assessment of individual cell differences and battery pack inconsistencies, resulting in insufficient targeting of equalization management strategies and an inability to effectively delay battery pack performance degradation.

[0004] In implementing the embodiments of the present invention, the prior art has at least the following problems or defects: the existing monitoring methods do not fully extract single-dimensional features and are difficult to take into account both static parameters and dynamic response characteristics; the health status assessment lacks an iterative optimization mechanism, does not adequately consider differences in cell types, and has limited assessment accuracy; no correlation model has been established between individual health values ​​and overall inconsistency indicators, making it difficult for the balanced management strategy to achieve intelligent switching between active and passive modes; the early warning mechanism is lagging behind and cannot identify abnormal cells and generate maintenance suggestions in time before the battery pack's health status deteriorates severely. Summary of the Invention

[0005] This invention provides a multi-dimensional health status monitoring and early warning method for vehicle battery packs, including: Collect benchmark monitoring data of the target battery pack under multiple preset monitoring modes to form a benchmark monitoring dataset, wherein the multiple monitoring modes include at least a first monitoring mode, a second monitoring mode and a third monitoring mode; Set the initial evaluation rounds and acquire dynamic monitoring data of the target battery pack during real-time operation to form a dynamic monitoring dataset; Enter the iterative evaluation process: perform multi-dimensional feature matching between the benchmark monitoring dataset and the dynamic monitoring dataset to generate a health matching degree; If the health matching degree does not reach the preset health threshold, then according to the type information of the cells in the target battery pack, the corresponding step size factor is selected from the preset step size mapping table, the step size factor is used to update the dynamic monitoring dataset, and the evaluation rounds are accumulated and returned to the iterative evaluation process. If the health matching degree reaches the health threshold, the iteration is terminated, and the comprehensive health degradation index of the target battery pack is calculated based on the current evaluation round and the step size factor used in the iteration process. Based on the comprehensive health degradation index and the evaluation round, the multi-dimensional health status of the target battery pack is calculated. The multi-dimensional health status includes the individual health value of each cell and the overall inconsistency index of the battery pack. Based on the individual health values ​​and the overall inconsistency index, a balanced management strategy for the target battery pack is generated and output.

[0006] Furthermore, in the iterative evaluation process, if the current evaluation round exceeds a preset maximum round threshold, a fault warning is triggered, and the following operations are performed: Based on the cell type information and current dynamic monitoring data, the location and type of abnormal cell are identified. Generate a fault report containing the abnormal cell number, abnormal parameters, and warning level; The fault report is sent to the vehicle display terminal for display.

[0007] Furthermore, the benchmark monitoring dataset is formed by collecting benchmark monitoring data of the target battery pack under multiple preset monitoring modes, including: When the monitoring mode is the first monitoring mode, the target battery pack is controlled to perform constant current charging and discharging at a preset rate under standard ambient temperature, and the terminal voltage, charging and discharging capacity and DC internal resistance of each cell are recorded to form the first reference subset; When the monitoring mode is the second monitoring mode, the voltage, current and temperature time series data of the target battery pack are continuously collected under the actual vehicle driving conditions, and the dynamic response characteristics of each cell are extracted to form the second benchmark subset. When the monitoring mode is the third monitoring mode, AC excitation signals at multiple frequency points are applied to the target battery pack, the response signals are collected and the electrochemical impedance spectral characteristic parameters of each cell are obtained by analysis, forming the third reference subset; The first benchmark subset, the second benchmark subset, and the third benchmark subset are combined into a benchmark monitoring dataset.

[0008] Further, the step of performing multi-dimensional feature matching between the benchmark monitoring dataset and the dynamic monitoring dataset to generate a health matching score includes: The baseline feature vectors of each cell under the corresponding monitoring mode are extracted from the baseline monitoring dataset. The baseline feature vectors include capacity retention rate, internal resistance growth rate, voltage plateau offset, and impedance arc radius. Extract dynamic feature vectors of the same dimension from the dynamic monitoring dataset; Calculate the Euclidean distance or cosine similarity between the baseline feature vector and the dynamic feature vector of each cell to obtain the cell-level matching degree; The overall health matching degree of the target battery pack is obtained by weighted averaging of the matching degree of all cells.

[0009] Furthermore, the calculation of the comprehensive health degradation index of the target battery pack based on the current evaluation round and the step size factor used in the iteration process includes: The first attenuation component is obtained by multiplying the preset baseline step size factor by the baseline round portion in the evaluation round; When the cell type is the first type, the fine-tuning step size factor is multiplied by the fine-tuning round part in the evaluation round to obtain the second attenuation component, and the sum of the first attenuation component and the second attenuation component is used as the comprehensive health attenuation index. When the cell type is type 2, the first preset step size factor is multiplied by the small step size part of the evaluation round to obtain the third attenuation component, and the second preset step size factor is multiplied by the large step size part of the evaluation round to obtain the fourth attenuation component. The sum of the first attenuation component, the third attenuation component and the fourth attenuation component is used as the comprehensive health attenuation index.

[0010] Furthermore, the calculation of the multi-dimensional health status of the target battery pack based on the comprehensive health degradation index and the evaluation round includes: When the preset monitoring mode is any one of the first monitoring mode, the second monitoring mode or the third monitoring mode, if the cell type is the first type, the individual health value of each cell is calculated according to the comprehensive health degradation index, the evaluation round, the baseline step size factor and the fine-tuning step size factor. If the cell type is the second type, the overall inconsistency index of the battery pack is calculated based on the comprehensive health degradation index, the evaluation round, the benchmark step size factor, the first preset step size factor, and the second preset step size factor.

[0011] Further, the calculation of the individual health value of each cell based on the comprehensive health degradation index, the evaluation round, the baseline step size factor, and the fine-tuning step size factor includes: Calculate the individual health value of the i-th cell using the following formula. : ; in, The comprehensive health decline index is N, where N is the assessment round. As the baseline step size factor, To fine-tune the step size factor, Let be the voltage deviation coefficient of the i-th cell. Let i be the temperature coefficient of the i-th cell. Let i be the base value for capacity decay of the i-th cell. and These are the preset weighting factors.

[0012] Further, the step of calculating the overall inconsistency index of the battery pack based on the comprehensive health degradation index, the evaluation round, the baseline step size factor, the first preset step size factor, and the second preset step size factor includes: Calculate the overall inconsistency index using the following formula : ; in, The comprehensive health decline index is N, where N is the assessment round. L1 is the baseline step size factor, L2 is the first preset step size factor, and L3 is the second preset step size factor. To evaluate the small step size portion of the rounds, To evaluate the large-step portion of the evaluation cycle, M represents the total number of cells. Let be the voltage of the i-th cell. As the reference voltage, Let be the internal resistance of the i-th cell. As the reference internal resistance, and These are the preset weighting coefficients.

[0013] Further, the step of generating a balanced management strategy for the target battery pack based on the individual health value and the overall inconsistency index includes: When the calculated multi-dimensional health status is an individual health value, the individual health value of each cell is compared with a preset individual health threshold. Cells with health values ​​lower than the individual health threshold are selected, and active balancing instructions are generated for these cells. The active balancing instructions control the battery management unit to supplement the charging or discharging of the cells through energy transfer. When the calculated multi-dimensional health status is an overall inconsistency indicator, the overall inconsistency indicator is compared with a preset overall inconsistency threshold. If it exceeds the overall inconsistency threshold, a passive balancing command is generated. The passive balancing command controls the battery management unit to balance the cells with abnormal voltage or internal resistance through energy consumption.

[0014] Furthermore, the method also includes: Based on the individual health values ​​and the overall inconsistency index, combined with a preset lifespan prediction model, the remaining effective lifespan of the target battery pack is estimated. Based on the remaining effective service life, generate maintenance prompts that include maintenance time and maintenance content suggestions; The maintenance prompts and the balance management strategy are sent together to the in-vehicle infotainment system for display.

[0015] The embodiments of the present invention have at least the following beneficial effects: 1. This invention integrates three monitoring modes: standard charge and discharge testing, real-vehicle operating condition monitoring, and AC impedance analysis. It constructs a multi-dimensional feature vector system covering capacity retention rate, internal resistance growth rate, voltage plateau offset, and impedance arc radius, achieving a comprehensive characterization of the static parameters and dynamic response characteristics of the battery pack. This solves the technical problems of insufficient information dimensions and one-sided feature extraction in a single monitoring mode, and improves the accuracy and reliability of health status assessment.

[0016] 2. This invention innovatively introduces an iterative evaluation mechanism based on cell type information. By dynamically selecting step size factors through a step size mapping table to update and optimize monitoring data, and establishing a correlation calculation model between the comprehensive health degradation index and the evaluation round and step size factor, it achieves differentiated and accurate evaluation of cells with different chemical systems and different aging degrees. This overcomes the shortcomings of existing technologies that ignore individual cell differences and have limited evaluation accuracy. At the same time, by triggering fault warnings through the maximum round threshold, it achieves early identification and location of abnormal cells.

[0017] 3. This invention constructs a two-dimensional evaluation system based on individual health values ​​and overall inconsistency indicators. Based on individual health values, it generates active balancing commands to supplement charging or discharging of weaker cells through energy transfer. Based on overall inconsistency indicators, it generates passive balancing commands to suppress deviations in cells with abnormal voltage or internal resistance through energy consumption, achieving intelligent coordination between active and passive balancing. Furthermore, it combines a lifespan prediction model to output maintenance time and content suggestions, forming a closed-loop management system encompassing condition monitoring, health assessment, balancing management, and lifespan prediction. This delays battery pack performance degradation and improves the safety and economy of the battery system. Attached Figure Description

[0018] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated in the drawings by way of example and not limitation, wherein: Figure 1 This is a flowchart illustrating a multi-dimensional health status monitoring and early warning method for vehicle battery packs provided in an embodiment of the present invention. Detailed Implementation

[0019] The principles and spirit of the invention will now be described with reference to several exemplary embodiments. It should be understood that these embodiments are provided merely to enable those skilled in the art to better understand and implement the invention, and are not intended to limit the scope of the invention in any way. Rather, these embodiments are provided to make the invention more thorough and complete, and to fully convey the scope of the invention to those skilled in the art.

[0020] Those skilled in the art will recognize that embodiments of the present invention can be implemented as a system, apparatus, device, method, or computer program product. Therefore, the present invention can be specifically implemented in the following forms: entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software.

[0021] It should be noted that the number of any elements in the accompanying drawings is for illustrative purposes only and not as a limitation, and any naming is for distinction only and has no limiting meaning.

[0022] The following is for reference. Figure 1 , Figure 1 This is a flowchart illustrating a multi-dimensional health status monitoring and early warning method for vehicle battery packs provided in an embodiment of the present invention. Figure 1 As shown, a multi-dimensional health status monitoring and early warning method for vehicle battery packs includes: Step 1: Collect benchmark monitoring data of the target battery pack under various preset monitoring modes to form a benchmark monitoring dataset.

[0023] First, a dataset representing the health baseline of the target battery pack needs to be established. The target battery pack refers to the power battery pack installed in an electric vehicle that requires health status monitoring, typically composed of multiple individual cells connected in series and parallel. Several preset monitoring modes are included, such as the first monitoring mode, the second monitoring mode, and the third monitoring mode, which comprehensively assess the battery health status from different dimensions. The first monitoring mode corresponds to offline standard testing, the second monitoring mode corresponds to real-vehicle driving conditions, and the third monitoring mode corresponds to electrochemical impedance spectroscopy analysis. The collected data is called baseline monitoring data, and the collection of all baseline monitoring data constitutes the baseline monitoring dataset. This dataset serves as a reference standard for subsequent iterative evaluations and is stored in on-board storage or a cloud server.

[0024] Step 2: Set the initial evaluation rounds and acquire dynamic monitoring data of the target battery pack during real-time operation to form a dynamic monitoring dataset.

[0025] An initial counter with a value of zero is set, representing the initial evaluation round, to record the number of subsequent iterative evaluations. A sensor network consisting of onboard voltage, current, and temperature sensors collects real-time data on the battery pack's voltage, current, and temperature during driving or charging. This data, known as dynamic monitoring data, reflects the battery's current real-time state. All collected dynamic data constitutes a dynamic monitoring dataset, which is continuously updated as the vehicle operates and stored in a cache for real-time analysis.

[0026] Step 3: Enter the iterative evaluation process: perform multi-dimensional feature matching between the baseline monitoring dataset and the dynamic monitoring dataset to generate a health matching degree.

[0027] The iterative evaluation process is initiated, which is a cyclical process. Its core operation is multi-dimensional feature matching, which compares the similarity between baseline data and dynamic data across multiple physicochemical dimensions such as voltage, current, temperature, and impedance. Using pre-written feature extraction and similarity calculation algorithms, the degree of similarity between the two is calculated, resulting in a quantitative indicator called the health matching degree. A higher health matching degree indicates that the battery's current state is closer to its health baseline; conversely, a lower degree indicates significant battery degradation or abnormality. This matching degree is used to determine whether further adjustments and analysis are needed.

[0028] Step 4: If the health matching degree does not reach the preset health threshold, select the corresponding step size factor from the preset step size mapping table according to the cell type information in the target battery pack, update the dynamic monitoring dataset using the step size factor, and return to step 3 after accumulating the evaluation rounds.

[0029] The preset health threshold is a pre-defined value, such as 0.95, used as a boundary to determine whether the battery state significantly deviates from the baseline. If the currently calculated health matching degree is lower than this threshold, further in-depth analysis is required. At this point, based on the chemical material system or structure of the cells in the battery pack, i.e., the cell type information, such as lithium iron phosphate as type I and ternary lithium as type II, a table pre-stored in memory, i.e., the preset step size mapping table, is consulted to select the corresponding adjustment amplitude parameter, i.e., the step size factor. The step size factor can be a numerical value or a function, used to correct or supplement the current dynamic monitoring dataset, such as adjusting the data time window width, filtering coefficient, or weighting coefficient, thereby generating a new dynamic monitoring dataset. At the same time, the evaluation round counter is incremented by 1. After the update is completed, the program flow returns to step 3, using the updated dynamic data to match the baseline data again.

[0030] Step 5: If the health matching degree reaches the health threshold, terminate the iteration and calculate the comprehensive health degradation index of the target battery pack based on the current evaluation round and the step size factor used in the iteration process. If the health matching degree reaches the preset health threshold, it indicates that the current analysis results are accurate enough, and the iterative calculation can be stopped. At this point, based on the final determined evaluation round (i.e., the total number of iterations) and all step size factors used in the iteration process, a value that can characterize the overall degree of battery degradation is obtained through comprehensive calculation, namely the comprehensive health degradation index. This index comprehensively reflects the overall change of the battery from the baseline state to the current state. Its calculation method varies depending on the cell type, which will be detailed in subsequent steps.

[0031] Step 6: Based on the comprehensive health degradation index and the evaluation round, calculate the multi-dimensional health status of the target battery pack. The multi-dimensional health status includes the individual health values ​​of each cell and the overall inconsistency index of the battery pack. Based on the comprehensive health degradation index and assessment rounds, further refined calculations yield more specific health indicators, namely, multi-dimensional health status. This method's multi-dimensional health status comprises two levels: a micro-level individual cell health value, describing the health status of each single cell and expressed as a percentage or numerical value; and a macro-level overall battery pack inconsistency index, describing the degree of performance difference between all cells within the battery pack, such as voltage dispersion or internal resistance dispersion. This difference is a key factor leading to overall battery pack performance degradation and safety hazards. Specific calculation formulas are provided in subsequent steps.

[0032] Step 7: Based on individual health values ​​and overall inconsistency indicators, generate a balanced management strategy for the target battery pack and output the balanced management strategy. Based on individual battery health values ​​and overall inconsistency indicators, specific control commands, or balancing management strategies, are generated for the battery management system. For example, cells with low health values ​​may require energy replenishment, activating the active balancing circuit; battery packs with excessive inconsistency may require activating the energy consumption balancing circuit. The generated strategies are ultimately output to the execution unit of the battery management system to guide it in precise maintenance and management of the battery pack, thereby extending battery life and ensuring driving safety. Balancing management strategies may include parameters such as balancing type, balancing current, and balancing time.

[0033] Step 101: Based on the cell type information and current dynamic monitoring data, identify the location and type of abnormal cell.

[0034] A fault warning is triggered when the current evaluation round exceeds a preset maximum round threshold. Based on cell type information, such as the chemical system, and current dynamic monitoring data, such as voltage drops or temperature spikes, a built-in diagnostic algorithm precisely locates the problematic cell (abnormal cell) and determines the nature of the problem, i.e., the type of abnormality, such as voltage abnormality, internal resistance abnormality, temperature abnormality, or capacity abnormality. The diagnostic algorithm can be implemented based on threshold comparison, machine learning classification, or rule-based reasoning.

[0035] Step 102: Generate a fault report containing the abnormal cell number, abnormal parameters, and warning level.

[0036] After identifying the abnormal battery cell, a structured report, or fault report, is generated. This report contains at least three core pieces of information: the specific battery cells involved (the abnormal cell numbers), the key data causing the abnormality (the abnormal parameters, such as cell number 5 having a voltage of 2.5 volts), and the severity of the problem (the warning level, such as Level 1 or Level 2 warning). Warning levels are categorized based on the degree to which the abnormal parameters deviate from the normal range; for example, a slight deviation is a Level 3 warning, and a severe deviation is a Level 1 warning. This report provides an accurate basis for subsequent repairs and handling.

[0037] Step 103: Send the fault report to the vehicle display terminal for display.

[0038] The fault report generated in step 102 is sent to the in-vehicle display terminal, i.e., the car's dashboard or central control screen, via the vehicle's internal communication network, such as the controller area network bus. The displayed content may include the abnormal cell number, abnormal parameter value, warning level, and suggested maintenance measures, so that the driver or maintenance personnel can obtain detailed fault information of the battery pack as soon as possible and take timely measures.

[0039] Step 201: When the monitoring mode is the first monitoring mode, control the target battery pack to perform constant current charging and discharging at a preset rate under standard ambient temperature, record the terminal voltage, charging and discharging capacity and DC internal resistance of each cell to form the first reference subset.

[0040] When the monitoring mode is set to the first monitoring mode, it corresponds to the offline testing scenario in a laboratory or repair shop. The test is conducted in a standard ambient temperature environment, such as a constant temperature chamber at 25 degrees Celsius. The control device commands the battery pack to perform a complete charge and discharge at a fixed current rate, i.e., a preset rate, such as 1C. During this process, the high-precision acquisition system records the terminal voltage change curve of each cell, the total amount of electricity charged or discharged (i.e., the charge / discharge capacity), and calculates the DC internal resistance of each cell using an algorithm that divides the instantaneous voltage change by the current change. These precise data collectively constitute the first benchmark subset, mainly reflecting the battery's capacity and basic resistance characteristics, and are stored in the benchmark monitoring dataset.

[0041] Step 202: When the monitoring mode is the second monitoring mode, the voltage, current and temperature time series data of the target battery pack are continuously collected under actual vehicle driving conditions, and the dynamic response characteristics of each cell are extracted to form the second reference subset.

[0042] When the monitoring mode is set to the second monitoring mode, data is collected under actual road driving conditions. During normal vehicle operation, the battery management system continuously collects the voltage, total current, and temperature of each cell of the entire battery pack at a high frequency, e.g., 10 times per second, forming a time-varying sequence of real-time data. Subsequently, signal processing algorithms such as wavelet transform or Kalman filtering are used to extract characteristic parameters that characterize the dynamic behavior of the cells, i.e., dynamic response features, such as voltage fluctuation amplitude and response time constant. These features, along with the raw data, form the second benchmark subset, reflecting the battery's actual performance under real-world operating conditions, and are stored in the benchmark monitoring dataset.

[0043] Step 203: When the monitoring mode is the third monitoring mode, apply AC excitation signals at multiple frequency points to the target battery pack, collect the response signals and analyze the electrochemical impedance spectral characteristic parameters of each cell to form the third reference subset.

[0044] When the monitoring mode is set to the third monitoring mode, this is a sophisticated electrochemical analysis method. Dedicated equipment, such as an electrochemical workstation, applies a series of small sinusoidal currents of varying frequencies, ranging from high-frequency millihertz to low-frequency hertz, as AC excitation signals to the battery pack. The voltage response signal of the battery at each frequency is then acquired. By analyzing the amplitude and phase changes of the response signal, the complex impedance characteristics inside the battery are revealed, and an electrochemical impedance spectroscopy (EIS) is plotted. Key characteristic parameters extracted from this EIS, such as ohmic impedance, charge transfer impedance, and Weber impedance coefficient, are called electrochemical impedance spectroscopy characteristic parameters. These parameters constitute the third benchmark subset, providing a deep understanding of the chemical reactions and aging state inside the battery, and are stored in the benchmark monitoring dataset.

[0045] Step 204: Combine the first benchmark subset, the second benchmark subset, and the third benchmark subset into a benchmark monitoring dataset.

[0046] The data subsets obtained from the three different monitoring modes in steps 201, 202, and 203—namely, the first benchmark subset, the second benchmark subset, and the third benchmark subset—are uniformly formatted and stored, and merged into a complete, multi-dimensional benchmark monitoring dataset. Data consistency must be maintained during the merging process, such as unifying sampling frequency, units, and timestamps. This dataset provides a comprehensive and multi-dimensional comparative benchmark for subsequent health status assessments and is stored in non-volatile memory.

[0047] Step 301: Extract the benchmark feature vector of each cell under the corresponding monitoring mode from the benchmark monitoring dataset. The benchmark feature vector includes capacity retention rate, internal resistance growth rate, voltage plateau offset, and impedance arc radius.

[0048] A set of key values ​​representing the health status of each cell is extracted from the benchmark monitoring dataset; this set is known as the benchmark feature vector. Specifically, the capacity retention rate is the percentage of the current benchmark capacity relative to the nominal initial capacity; the internal resistance growth rate is the percentage increase in the benchmark internal resistance relative to the initial internal resistance; the voltage plateau offset describes the degree of voltage plateau drift in the charge-discharge curve, which can be obtained by comparing voltage curves under different cycles; and the impedance arc radius is the radius of the semicircle in the Nyquist plot extracted from electrochemical impedance spectroscopy data, which is related to the difficulty of the electrochemical reaction. This set of vectors constitutes a standard template for evaluating cell health and is stored in the feature library.

[0049] Step 302: Extract dynamic feature vectors of the same dimension from the dynamic monitoring dataset.

[0050] From the dynamic monitoring dataset, using the exact same algorithm and parameters as in step 301, a set of feature values, i.e., dynamic feature vectors, is extracted for each cell. The dimension of this set of vectors, i.e., the types of parameters it contains, such as capacity retention rate and internal resistance growth rate, must be completely consistent with the baseline feature vector in step 301 for subsequent comparisons to be made. The extraction of dynamic feature vectors requires real-time computation; therefore, an efficient algorithm must be employed to ensure real-time performance.

[0051] Step 303: Calculate the Euclidean distance or cosine similarity between the baseline feature vector and the dynamic feature vector of each cell to obtain the cell-level matching degree.

[0052] For each battery cell, its baseline feature vector and dynamic feature vector are input into a similarity calculation function. This method can employ two similarity algorithms: Euclidean distance, which measures the difference by calculating the straight-line distance between two vectors in multidimensional space (the smaller the distance, the more similar); or cosine similarity, which measures consistency by calculating the cosine of the angle between the directions of two vectors (the smaller the angle, the more similar). The calculated result is the cell-level matching degree, quantifying the degree of deviation of the current state of a single battery cell from its own health baseline. The cell-level matching degree is a value between 0 and 1, where 1 represents a perfect match and 0 represents a complete mismatch.

[0053] Step 304: Calculate the weighted average of the matching degree of all cells to obtain the overall health matching degree of the target battery pack.

[0054] The matching degrees of all individual cells calculated in step 303 are summarized to obtain a comprehensive index representing the entire battery pack. Considering that different cells may have different positions within the battery pack (e.g., cells in the middle have higher temperatures), a weighted average method is used to assign different weights to cells at different positions. These weights can be set based on historical temperature data or factory parameters for each cell's location. Finally, the overall health matching degree of the target battery pack is calculated. This value is the health matching degree used in step 3 to determine whether the iteration should continue.

[0055] Step 501: Multiply the preset baseline step size factor by the baseline round portion in the evaluation round to obtain the first attenuation component.

[0056] First, a fixed coefficient related to the basic aging rate of the battery cell is retrieved from a pre-defined step size mapping table; this is the pre-defined baseline step size factor. Simultaneously, the total evaluation rounds are divided into several parts, with the most fundamental and inevitable part called the baseline round part, which is a fixed number of rounds included in all iterations. Multiplying the two results in the first degradation component, representing the unavoidable basic degradation of the battery under ideal conditions. The baseline step size factor is pre-calibrated according to the cell material type.

[0057] Step 502: When the cell type is the first type, multiply the fine-tuning step size factor by the fine-tuning round part in the evaluation round to obtain the second attenuation component, and use the sum of the first attenuation component and the second attenuation component as the comprehensive health attenuation index.

[0058] When the cell type is Type I, such as lithium iron phosphate battery, in addition to the basic degradation, some minor, fine-tuning adjustments need to be considered. The fine-tuning step size factor corresponding to this type is retrieved from the step size mapping table, and the portion of the evaluation round used to handle these fine-tuning adjustments is called the fine-tuning round portion. Multiplying these two components yields the second degradation component. Finally, the first degradation component from step 501 is added to the second degradation component here, and the sum is the comprehensive health degradation index for this type of cell. The fine-tuning step size factor reflects the subtle aging characteristics of Type I cells.

[0059] Step 503: When the cell type is the second type, multiply the first preset step size factor by the small step size part of the evaluation round to obtain the third attenuation component, multiply the second preset step size factor by the large step size part of the evaluation round to obtain the fourth attenuation component, and use the sum of the first attenuation component, the third attenuation component and the fourth attenuation component as the comprehensive health attenuation index.

[0060] When the cell type is Type II, such as ternary lithium batteries, the degradation mode of these batteries is more complex, exhibiting multi-scale degradation characteristics. Two different step size factors are used: a first preset step size factor is used to match the relatively gradual changes in the evaluation rounds, i.e., the small step size rounds, to obtain the third degradation component; the second preset step size factor is used to match the rapidly changing parts in the evaluation rounds, i.e., the large step size rounds, to obtain the fourth degradation component. Finally, the first, third, and fourth degradation components are added together, and the sum is the comprehensive health degradation index for this type of cell. The first and second preset step size factors are predetermined based on accelerated aging test data of Type II cells.

[0061] Step 601: When the preset monitoring mode is any one of the first monitoring mode, the second monitoring mode, or the third monitoring mode, if the cell type is the first type, calculate the individual health value of each cell based on the comprehensive health degradation index, the evaluation round, the baseline step size factor, and the fine-tuning step size factor. Once the cell type is determined to be Type I, such as lithium iron phosphate, the logic for calculating the individual health value begins. Using the comprehensive health degradation index calculated in step 502, the evaluation round, and the corresponding baseline and fine-tuning step size factors, a specific formula is used to calculate the independent health value for each cell. This formula comprehensively considers the total degradation and the step size factor during the iteration process to eliminate the influence of different step sizes on degradation quantification.

[0062] Step 602: If the cell type is the second type, calculate the overall inconsistency index of the battery pack based on the comprehensive health degradation index, evaluation round, benchmark step size factor, first preset step size factor and second preset step size factor.

[0063] When the cell type is determined to be Type II, such as ternary lithium, the focus shifts from the health of individual cells to the consistency of all cells within the entire battery pack. Using the comprehensive health degradation index calculated in step 503, the evaluation round, and the corresponding baseline step size factor, first preset step size factor, and second preset step size factor, a specific formula is used to calculate an overall inconsistency index representing the dispersion of the battery pack. This index reflects overall inconsistency by statistically analyzing the deviations of the voltage and internal resistance of all cells relative to the baseline values.

[0064] Individual health scores are calculated using the following formula: ; in, The comprehensive health decline index is N, where N is the assessment round. As the baseline step size factor, To fine-tune the step size factor, Let be the voltage deviation coefficient of the i-th cell. Let i be the temperature coefficient of the i-th cell. Let i be the base value for capacity decay of the i-th cell. and These are the preset weighting factors.

[0065] The voltage deviation coefficient is calculated by the difference between the real-time voltage and the average voltage, reflecting the voltage consistency of the battery cell; the temperature coefficient is determined by the degree of deviation between the current temperature and the optimal operating temperature, reflecting the thermal management effect; the capacity decay baseline is estimated by the ampere-hour integration method using historical charge and discharge data. Weighting factors. and Through experimental calibration, for example, using values ​​of 0.6 and 0.4 respectively, this formula can be used to calculate a precise individual health value for each cell, typically ranging from 0 to 1, where 1 represents a brand new state.

[0066] The overall inconsistency index is calculated using the following formula: ; in, The comprehensive health decline index is N, where N is the assessment round. L1 is the baseline step size factor, L2 is the first preset step size factor, and L3 is the second preset step size factor. To evaluate the small step size portion of the rounds, To evaluate the large-step portion of the evaluation cycle, M represents the total number of cells. Let be the voltage of the i-th cell. As the reference voltage, Let be the internal resistance of the i-th cell. As the reference internal resistance, and These are the preset weighting coefficients.

[0067] The first part of the formula is the degradation intensity normalization factor, and the second part calculates the average of the absolute values ​​of the relative deviations between the voltages of all cells and the reference voltage, as well as the average of the absolute values ​​of the relative deviations between the internal resistances of all cells and the reference internal resistance. The reference voltage and reference internal resistance can be taken as the average values ​​of all cells or the factory nominal values. The weighting coefficients β1 and β2 are, for example, 0.5 and 0.5 respectively, to balance the contribution of voltage and internal resistance inconsistencies. This formula can provide a quantified overall battery pack inconsistency index; the larger the value, the more severe the inconsistency.

[0068] Step 701: When the calculated multi-dimensional health status is an individual health value, the individual health value of each cell is compared with the preset individual health threshold, cells with health values ​​lower than the individual health threshold are selected, and active balancing instructions are generated for these cells.

[0069] When step 6 outputs an individual health value, the balancing management strategy is executed on an individual cell basis. A preset individual health threshold, such as 80%, is set. The individual health value of each cell is compared one by one, and all cells with a health value below 80% are selected. An active balancing command is generated for these lagging cells, instructing the battery management unit to activate the energy transfer balancing circuit. This circuit transfers energy from the high-energy cells to these low-energy cells through energy storage components such as capacitors or inductors, achieving energy redistribution and improving the consistency of the entire battery pack. The active balancing command includes parameters such as the cell number to be balanced, the balancing current magnitude, and the balancing duration.

[0070] Step 702: When the calculated multidimensional health status is an overall inconsistency indicator, compare the overall inconsistency indicator with the preset overall inconsistency threshold. If it exceeds the overall inconsistency threshold, generate a passive balancing instruction.

[0071] When step 6 outputs an overall inconsistency index, the balancing management strategy focuses on overall execution. A preset overall inconsistency threshold is set, for example, 0.15. If the calculated overall inconsistency index value is greater than 0.15, it indicates that the overall dispersion of the battery pack is too large. At this time, a passive balancing command is generated, instructing the battery management unit to activate the energy-consuming balancing circuit. This circuit discharges cells with abnormally high voltage or internal resistance through parallel resistors or other methods, consuming their excess energy and bringing them closer to the same level as other cells, thereby reducing overall inconsistency. The passive balancing command includes parameters such as the cell number to be balanced, the discharge current magnitude, and the balancing time.

[0072] Step 801: Based on individual health values ​​and overall inconsistency indicators, combined with a preset lifespan prediction model, estimate the remaining effective lifespan of the target battery pack.

[0073] After obtaining individual health values ​​and overall inconsistency indicators, this data is input into a pre-trained machine learning model or empirical formula, i.e., a pre-defined lifespan prediction model. This model can be a neural network trained on a large amount of battery aging data, such as a multilayer perceptron or long short-term memory network, or a support vector machine regression model. The model input consists of the individual health value sequence and the overall inconsistency indicator, and the output is the remaining effective lifespan, expressed as the remaining number of cycles or the remaining calendar time. Historical aging data, including complete cycles from battery newness to failure, is used during model training.

[0074] Step 802: Based on the remaining effective service life, generate maintenance prompt information that includes maintenance time suggestions and maintenance content suggestions.

[0075] Based on the remaining lifespan estimated in step 801, this is transformed into valuable recommendations for the user. Maintenance reminders are generated according to the remaining lifespan; for example, if the predicted remaining lifespan is less than one year, the reminder suggests a battery health check in six months. Simultaneously, combining individual health values ​​and overall inconsistency indicators, more specific maintenance recommendations are given, such as recommending replacement of cells 3 and 7, or suggesting battery pack balancing maintenance. Maintenance reminders are stored in text format.

[0076] Step 803: Send the maintenance prompt information and the balance management strategy to the in-vehicle infotainment system for display. The maintenance reminder information generated in step 802 and the battery balancing management strategy generated in step 7 are packaged and sent to the in-vehicle infotainment system (the central control screen) via the vehicle network. This information is clearly displayed to the driver on the screen in the form of text, charts, or warning lights, reminding them to perform timely vehicle maintenance or understand the current battery management measures, thereby improving user experience and vehicle safety. Information may include remaining battery life percentage, recommended maintenance date, battery balancing status, etc.

[0077] Furthermore, the storage medium in the embodiments of this application stores program instructions capable of implementing all the above methods. These program instructions can be stored in the storage medium in the form of a software product, including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) or processor to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks, or terminal devices such as computers, servers, mobile phones, and tablets.

[0078] The above description is merely an explanation of some preferred embodiments of the present invention and the technical principles employed. Those skilled in the art should understand that the scope of the invention as described in the embodiments of the present invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of the present invention.

Claims

1. A multi-dimensional health status monitoring and early warning method for vehicle battery packs, characterized in that, Includes the following steps: Collect benchmark monitoring data of the target battery pack under multiple preset monitoring modes to form a benchmark monitoring dataset, wherein the multiple monitoring modes include at least a first monitoring mode, a second monitoring mode and a third monitoring mode; Set the initial evaluation rounds and acquire dynamic monitoring data of the target battery pack during real-time operation to form a dynamic monitoring dataset; Enter the iterative evaluation process, perform multi-dimensional feature matching between the benchmark monitoring dataset and the dynamic monitoring dataset, and generate a health matching degree. If the health matching degree does not reach the preset health threshold, then according to the type information of the cells in the target battery pack, the corresponding step size factor is selected from the preset step size mapping table, the step size factor is used to update the dynamic monitoring dataset, and the evaluation rounds are accumulated and returned to the iterative evaluation process. If the health matching degree reaches the health threshold, the iteration is terminated, and the comprehensive health degradation index of the target battery pack is calculated based on the current evaluation round and the step size factor used in the iteration process. Based on the comprehensive health degradation index and the evaluation round, the multi-dimensional health status of the target battery pack is calculated. The multi-dimensional health status includes the individual health value of each cell and the overall inconsistency index of the battery pack. Based on the individual health values ​​and the overall inconsistency index, a balanced management strategy for the target battery pack is generated and output.

2. The method according to claim 1, characterized in that, In the iterative evaluation process, if the current evaluation round exceeds a preset maximum round threshold, a fault warning is triggered, and the following operations are performed: Based on the cell type information and current dynamic monitoring data, the location and type of abnormal cell are identified. Generate a fault report containing the abnormal cell number, abnormal parameters, and warning level; The fault report is sent to the vehicle display terminal for display.

3. The method according to claim 1, characterized in that, The benchmark monitoring dataset is formed by collecting benchmark monitoring data of the target battery pack under multiple preset monitoring modes, including: When the monitoring mode is the first monitoring mode, the target battery pack is controlled to perform constant current charging and discharging at a preset rate under standard ambient temperature, and the terminal voltage, charging and discharging capacity and DC internal resistance of each cell are recorded to form the first reference subset; When the monitoring mode is the second monitoring mode, the voltage, current and temperature time series data of the target battery pack are continuously collected under the actual vehicle driving conditions, and the dynamic response characteristics of each cell are extracted to form the second benchmark subset. When the monitoring mode is the third monitoring mode, AC excitation signals at multiple frequency points are applied to the target battery pack, the response signals are collected and the electrochemical impedance spectral characteristic parameters of each cell are obtained by analysis, forming the third reference subset; The first benchmark subset, the second benchmark subset, and the third benchmark subset are combined into a benchmark monitoring dataset.

4. The method according to claim 1, characterized in that, The step of performing multi-dimensional feature matching between the benchmark monitoring dataset and the dynamic monitoring dataset to generate a health matching score includes: The baseline feature vectors of each cell under the corresponding monitoring mode are extracted from the baseline monitoring dataset. The baseline feature vectors include capacity retention rate, internal resistance growth rate, voltage plateau offset, and impedance arc radius. Extract dynamic feature vectors of the same dimension from the dynamic monitoring dataset; Calculate the Euclidean distance or cosine similarity between the baseline feature vector and the dynamic feature vector of each cell to obtain the cell-level matching degree; The overall health matching degree of the target battery pack is obtained by weighted averaging of the matching degree of all cells.

5. The method according to claim 1, characterized in that, The calculation of the comprehensive health degradation index of the target battery pack based on the current evaluation round and the step size factor used in the iteration process includes: The first attenuation component is obtained by multiplying the preset baseline step size factor by the baseline round portion in the evaluation round; When the cell type is the first type, the fine-tuning step size factor is multiplied by the fine-tuning round part in the evaluation round to obtain the second attenuation component, and the sum of the first attenuation component and the second attenuation component is used as the comprehensive health attenuation index. When the cell type is type 2, the first preset step size factor is multiplied by the small step size part of the evaluation round to obtain the third attenuation component, and the second preset step size factor is multiplied by the large step size part of the evaluation round to obtain the fourth attenuation component. The sum of the first attenuation component, the third attenuation component and the fourth attenuation component is used as the comprehensive health attenuation index.

6. The method according to claim 5, characterized in that, The calculation of the multi-dimensional health status of the target battery pack based on the comprehensive health degradation index and the evaluation round includes: When the preset monitoring mode is any one of the first monitoring mode, the second monitoring mode or the third monitoring mode, if the cell type is the first type, the individual health value of each cell is calculated according to the comprehensive health degradation index, the evaluation round, the baseline step size factor and the fine-tuning step size factor. If the cell type is the second type, the overall inconsistency index of the battery pack is calculated based on the comprehensive health degradation index, the evaluation round, the benchmark step size factor, the first preset step size factor, and the second preset step size factor.

7. The method according to claim 6, characterized in that, The calculation of the individual health value of each cell based on the comprehensive health degradation index, the evaluation round, the baseline step size factor, and the fine-tuning step size factor includes: Calculate the individual health value of the i-th cell using the following formula. : ; in, The comprehensive health decline index is N, where N is the assessment round. As the baseline step size factor, To fine-tune the step size factor, Let be the voltage deviation coefficient of the i-th cell. Let i be the temperature coefficient of the i-th cell. Let i be the base value for capacity decay of the i-th cell. and These are preset weighting factors.

8. The method according to claim 6, characterized in that, The step of calculating the overall inconsistency index of the battery pack based on the comprehensive health degradation index, the evaluation round, the baseline step size factor, the first preset step size factor, and the second preset step size factor includes: Calculate the overall inconsistency index using the following formula : ; in, The comprehensive health decline index is N, where N is the assessment round. L1 is the baseline step size factor, L2 is the first preset step size factor, and L3 is the second preset step size factor. To evaluate the small step size portion of the rounds, To evaluate the large-step portion of the evaluation cycle, M represents the total number of cells. Let be the voltage of the i-th cell. The reference voltage, Let be the internal resistance of the i-th cell. As the reference internal resistance, and These are the preset weighting coefficients.

9. The method according to claim 7 or 8, characterized in that, The step of generating a balanced management strategy for the target battery pack based on the individual health values ​​and the overall inconsistency index includes: When the calculated multi-dimensional health status is an individual health value, the individual health value of each cell is compared with a preset individual health threshold. Cells with health values ​​lower than the individual health threshold are selected, and active balancing instructions are generated for these cells. The active balancing instructions control the battery management unit to supplement the charging or discharging of the cells through energy transfer. When the calculated multi-dimensional health status is an overall inconsistency indicator, the overall inconsistency indicator is compared with a preset overall inconsistency threshold. If it exceeds the overall inconsistency threshold, a passive balancing command is generated. The passive balancing command controls the battery management unit to balance the cells with abnormal voltage or internal resistance through energy consumption.

10. The method according to claim 9, characterized in that, The method further includes: Based on the individual health values ​​and the overall inconsistency index, combined with a preset lifespan prediction model, the remaining effective lifespan of the target battery pack is estimated. Based on the remaining effective service life, generate maintenance prompts that include maintenance time and maintenance content suggestions; The maintenance prompts and the balance management strategy are sent together to the in-vehicle infotainment system for display.

Images