Method for evaluating state of health and life warning of power battery

Abstract

The present application relates to the field of power battery evaluation technology, and discloses a power battery health state evaluation and life warning method. The method comprises the following steps: dividing a target power battery module into several layer evaluation areas, and randomly selecting several measurement position points in each layer evaluation area; then, collecting battery state original data of each measurement position point in each layer evaluation area in real time, and comprehensively analyzing the data to obtain a health state evaluation value of the target power battery module; then, judging and analyzing the health state evaluation value and a preset health state threshold value; if the health state evaluation value is lower than the preset threshold value, triggering a preset life warning measure; if the health state evaluation value is higher than the preset threshold value, maintaining the current evaluation period. The method can improve the accuracy of battery health state evaluation, optimize the life warning opportunity, and is suitable for health management of various power battery modules.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power battery evaluation, in particular to a power battery health state evaluation and life warning method. BACKGROUND

[0002] With the rapid development of new energy industry, as the core components of new energy vehicles, energy storage systems, etc., the operation safety and stability of power batteries directly affect the reliable operation of the entire system. In the long-term use of power batteries, affected by factors such as charge-discharge cycle, temperature change, vibration impact, etc., phenomena such as electrode material aging, electrolyte decomposition, and separator performance degradation occur inside the battery, causing the battery health state to decline continuously. If the battery health state cannot be grasped in time and effective warning is not carried out, it may cause battery performance to drop sharply, the cruising range to be shortened, and even safety hazards such as thermal runaway, fire, and explosion.

[0003] The current industry method for evaluating the health state of power batteries mostly adopts a whole measurement method, that is, a few fixed measurement points are selected for data collection on the entire power battery module. This method ignores the temperature differences and uneven current distribution problems existing in different areas of the battery module during use, resulting in insufficient representativeness of the collected raw data, making it difficult to accurately reflect the true health state of each area of the battery module, and further causing a large deviation between the final health state evaluation value and the actual situation.

[0004] In the data comprehensive analysis link of the existing evaluation method, only the changes of a single parameter (such as voltage, temperature) are often focused on, lacking system integration and correlation analysis of multi-dimensional battery state data, and unable to comprehensively capture the change rule of the battery health state. In terms of life warning, the warning threshold set by most methods is relatively single, without considering the differences in battery health state under different use scenarios and different aging degrees, resulting in inaccurate timing of triggering the warning measures, either triggering too early to cause unnecessary increase in maintenance cost, or triggering too late to fail to avoid safety risks in time. In addition, the evaluation period of the existing evaluation method is mostly fixed, which cannot be dynamically adjusted according to the actual changes of the battery health state, causing resource waste when the battery health state is good, and missing the best intervention opportunity when the battery health state changes abnormally due to the too long evaluation period. SUMMARY

[0005] The purpose of the present application is to provide a power battery health state evaluation and life warning method to solve the problems raised in the background.

[0006] To achieve the above purpose, the present application provides a power battery health state evaluation and life warning method, which comprises:

[0007] The target power battery module is divided into several layers of evaluation areas, and several measurement position points are randomly selected in each layer of evaluation area;

[0008] Real-time collection of battery state raw data at each measurement position point in each layer of evaluation area in the target power battery module is performed, and comprehensive analysis is performed to obtain a health state evaluation value of the target power battery module;

[0009] The health state evaluation value of the target power battery module is compared with a preset health state threshold value;

[0010] If the health state evaluation value of the target power battery module is lower than the preset health state threshold value, a preset life warning measure is triggered;

[0011] If the health state evaluation value of the target power battery module is higher than the preset health state threshold value, the current evaluation period is maintained.

[0012] Preferably, the battery state raw data includes battery voltage values, battery temperature values and battery internal resistance values at several time points.

[0013] Preferably, the step of obtaining the health state evaluation value of the target power battery module comprises:

[0014] The battery voltage values, battery temperature values and battery internal resistance values at each measurement position point in each layer of evaluation area in the target power battery module are read, and comprehensive analysis is performed to obtain battery voltage measurement mean value, battery temperature measurement mean value and battery internal resistance measurement mean value of each layer of evaluation area;

[0015] The battery voltage measurement mean value, battery temperature measurement mean value and battery internal resistance measurement mean value of each layer of evaluation area are comprehensively analyzed to generate voltage distribution feature set, temperature distribution feature set and internal resistance distribution feature set of the target power battery module;

[0016] The voltage distribution feature set, temperature distribution feature set and internal resistance distribution feature set are jointly analyzed to obtain the health state evaluation value.

[0017] Preferably, the step of triggering the preset life warning measure comprises:

[0018] The health state evaluation value of the target power battery module is compared with a preset health state threshold value to obtain a health state deviation value;

[0019] The current running state data and historical running reference data of the target power battery module are obtained, wherein the current running state data includes current charge-discharge rate value and current environment temperature value, and the historical running reference data includes historical average charge-discharge rate value and historical average environment temperature value;

[0020] inputting the health state deviation value, the current operation state data and the historical operation benchmark data into a preset early warning grading model for comprehensive analysis to obtain a life early warning grade index.

[0021] Preferably, the preset early warning grading model selects different calculation logics according to a numerical interval in which the health state deviation value is located:

[0022] When the health state deviation value is lower than a first deviation threshold value, a first weight coefficient is used to perform weighted correction on the health state deviation value;

[0023] When the health state deviation value is between the first deviation threshold value and a second deviation threshold value, an exponential correction is performed in combination with a rate difference value of the current charge-discharge rate value and a historical average charge-discharge rate value;

[0024] When the health state deviation value is higher than the second deviation threshold value, an interactive correction is performed by introducing an absolute value of a temperature difference between the current environmental temperature value and a historical average environmental temperature value, and the life early warning grade index is output.

[0025] Preferably, the step of triggering a preset life early warning measure further comprises:

[0026] matching and analyzing the life early warning grade index with a plurality of preset early warning grade intervals, wherein each early warning grade interval corresponds to a preset life early warning measure;

[0027] executing the life early warning measure corresponding to the matched life early warning grade interval.

[0028] Preferably, the step of maintaining the current evaluation period comprises:

[0029] performing difference calculation on the health state evaluation value of the target power battery module and a preset health state threshold value to obtain a health state redundancy value;

[0030] obtaining a current cycle number of the target power battery module and a cycle number design threshold value;

[0031] inputting the health state redundancy value, the current cycle number and the cycle number design threshold value into a preset period adjustment model for comprehensive analysis to obtain an evaluation period extension coefficient.

[0032] Preferably, the preset period adjustment model performs differential calculation according to a numerical range in which the health state redundancy value is located:

[0033] When the health state redundancy value is lower than a first redundancy threshold value, a linear scaling is performed based on a proportion value of the current cycle number in the cycle number design threshold value;

[0034] When the health state redundancy value is higher than a first redundancy threshold value, a product of a logarithmic transformation result of the health state redundancy value and a cycle number margin value is adopted for dynamic adjustment, and an evaluation period extension coefficient is output.

[0035] Preferably, the method further comprises a dynamic threshold updating step:

[0036] Based on a historical health state evaluation value sequence, a health state decay rate value is calculated;

[0037] In combination with the health state decay rate value and a preset life termination boundary value, the health state threshold value is recalculated;

[0038] The updated health state threshold value is taken as a reference for next time judgment and analysis.

[0039] Preferably, the dynamic threshold updating step further comprises:

[0040] Degradation characteristic data of the same type of power battery module is collected, and a common degradation mode is extracted;

[0041] Based on the common degradation mode, the health state decay rate value is compensated and corrected to generate a corrected health state decay rate value;

[0042] The health state threshold value is updated using the corrected health state decay rate value.

[0043] Compared with the prior art, the present application has the following beneficial effects:

[0044] By dividing the target power battery module into several layers of evaluation areas and randomly selecting several measurement location points in each layer of evaluation areas, the limitation of traditional overall measurement of only selecting a few fixed points is changed. The layered evaluation areas can be targeted to cover different positions inside the battery module, and considering the possible distribution differences of temperature, current and the like in different areas of the battery module during use, the random selection of multiple measurement points in each layer of areas can effectively avoid the local data deviation that may be caused by fixed measurement points, so that the collected battery state raw data can more comprehensively and uniformly reflect the actual state of each layer of areas of the battery module, and provide a more representative data basis for subsequent health state evaluation.

[0045] In the data processing link, the method comprehensively analyzes the battery state raw data of each measurement position point of each layer of the evaluation area collected in real time, rather than limiting to the analysis of a single parameter. This multi-dimensional and comprehensive data analysis method can integrate various battery state parameters such as voltage, temperature and current, mine the correlation between different parameters, more accurately capture the subtle changes of the battery health state, and thus obtain a health state evaluation value that is more in line with the actual situation of the battery module, reduce the evaluation deviation caused by single parameter analysis or local data collection, and improve the accuracy and reliability of the evaluation result.

[0046] In the health state judgment and early warning link, the method compares and analyzes the health state evaluation value with the preset health state threshold value to determine whether to trigger the life warning measure. Compared with the traditional single threshold setting method, the method can flexibly set a health state threshold value that is more in line with the actual demand according to the design parameters, use scenarios and aging characteristics of the battery module, so that the early warning judgment is more targeted. When the health state evaluation value is lower than the preset threshold value, the preset warning measure is triggered, which can timely remind the relevant personnel to take intervention measures such as maintenance and replacement, effectively avoiding the safety risks caused by the deterioration of the battery health state; and when the health state evaluation value is higher than the preset threshold value, the current evaluation period is maintained, which can avoid unnecessary frequent evaluation when the battery state is good, reduce the resource consumption in the data collection and analysis process, and realize the reasonable allocation of evaluation resources.

[0047] The setting of "if the health state evaluation value is higher than the preset threshold value, the current evaluation period is maintained" in the method implies the possibility of dynamically adjusting the evaluation period according to the battery health state (when the evaluation value approaches the threshold value or shows a downward trend, the evaluation period can be subsequently optimized to be shortened), which can better adapt to the changes of the battery health state compared with the traditional fixed evaluation period. When the battery health state is stable and good, maintaining the current period can reduce the evaluation frequency and save manpower and equipment resources; when the battery health state shows a downward trend and approaches the threshold value, the evaluation period can be adjusted subsequently to increase the evaluation frequency, timely track the changes of the battery health state, ensure that the response can be made quickly when needed, and avoid the problems of resource waste or untimely warning caused by the fixed period. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 A working principle diagram of the power battery health state evaluation and life warning method described in the application;

[0049] Figure 2 A flowchart for calculating the health state evaluation value;

[0050] Figure 3 A flowchart for triggering the life warning measure;

[0051] Figure 4 Flowchart for maintaining the current evaluation cycle

[0052] Figure 5 Flowchart for dynamically updating the threshold compensation correction. DETAILED DESCRIPTION

[0053] The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative work fall within the protection scope of the present application.

[0054] Please refer to Figure 1 The present application provides a power battery health state evaluation and life warning method, which comprises the following steps:

[0055] By hierarchical partitioning random sampling and multi-source data fusion analysis, fine evaluation and forward-looking warning of the health state of the power battery module are realized. In specific implementation, the target power battery module is divided into several layers of evaluation areas in the physical space. This division can be based on the physical structure, thermal management layout or electrical connection characteristics of the battery module. In each layer of evaluation area, a certain number of measurement location points are randomly selected. This is aimed at reducing the systematic bias that may be caused by fixed point monitoring through random sampling, so as to more comprehensively capture the state differences inside the module. Then, through the sensor units deployed at each measurement location point, the battery state raw data at each measurement point in each layer are collected in real time. The collected raw data are transmitted to the central processing unit for comprehensive analysis. This analysis process integrates multi-dimensional information, and finally calculates a comprehensive health state evaluation value (SOH). This evaluation value is compared with the preset health state threshold value. If the evaluation value is lower than the threshold value, it indicates that the battery health state has degraded to a level that needs attention, and the preset life warning measures are triggered immediately. If the evaluation value is higher than the threshold value, it indicates that the battery state is in an acceptable range, and the current evaluation cycle remains unchanged, and the monitoring continues.

[0056] Embodiment 1: Please refer to Figure 2, assuming the target object is a lithium-ion power battery module applied to an electric vehicle, which is composed of multiple battery cells in series and parallel, and is physically divided into three evaluation zones. This division is based on the mechanical structure and thermal management layout of the module, for example, the upper layer is close to the heat dissipation outlet, the middle layer is the core area, and the lower layer is close to the cooling liquid inlet. Within each evaluation zone, the system randomly selects three measurement location points. For example, the upper layer may randomly select points near the edge, center, and a certain connecting sheet of the module; the middle and lower layers also perform similar random selection to ensure that the sampling points cover different positions in the layer and reduce the risk of missing local abnormalities due to fixed point monitoring.

[0057] At each selected measurement location point, a high-precision sensor unit is deployed. These sensors collect three key parameters in real time and synchronously: battery voltage value, battery temperature value, and battery internal resistance value. The voltage value is obtained by directly measuring the potential difference between the battery terminals through a high-precision analog-to-digital converter; the temperature value is measured by a patch-type temperature sensor closely attached to the surface of the battery shell; the internal resistance value is calculated by injecting a small AC signal of a specific frequency into the battery and measuring its response voltage. Data collection is performed at fixed time intervals (e.g., once every second) with precise time stamps, ensuring the synchronization of different location points and different parameters in the time dimension, providing a time-consistent data basis for subsequent comprehensive analysis.

[0058] The central processing unit receives the raw data stream from all measurement location points in real time through the data bus. The first step in calculating the health status evaluation value is to perform layered mean calculation. The processing unit first reads the battery voltage value sequence uploaded by all three measurement location points in a certain evaluation zone (e.g., the upper layer). Assuming that at a certain analysis time, the voltage values reported by the three points in the upper layer are 3.32V, 3.30V, and 3.31V. The system performs a simple arithmetic mean calculation on these values: (3.32+3.30+3.31) / 3=3.31V, which is the battery voltage measurement mean value of the upper evaluation zone. Using the same algorithm, the temperature values reported by the three measurement points in this layer (e.g., 28.5°C, 29.0°C, 28.7°C) are averaged to obtain the battery temperature measurement mean value of the layer (28.73°C). Similarly, the internal resistance values of the three points in this layer are averaged to obtain the battery internal resistance measurement mean value of the layer (2.15mΩ). This process is sequentially performed on the data of all measurement location points in the middle and lower layers, respectively calculating the battery voltage measurement mean value, battery temperature measurement mean value, and battery internal resistance measurement mean value of the middle and lower layers. Layered mean calculation effectively integrates the information of multiple points in the layer, smooths the instantaneous fluctuations or small measurement errors of individual points, and reflects the overall level status of the layer in terms of voltage, temperature, and internal resistance.

[0059] After the layer-wise mean calculation, the system enters the comprehensive analysis phase, aiming to generate a feature set describing the voltage distribution of the entire module. The system collects the mean voltage measurement data of all layers (upper, middle, and lower). Statistical analysis is performed on these layer mean data to calculate their maximum, minimum, range (maximum minus minimum), variance, or standard deviation. For example, if the three layer voltage means are 3.31 V (upper), 3.29 V (middle), and 3.30 V (lower), the maximum is 3.31 V, the minimum is 3.29 V, and the range is 0.02 V. At the same time, the average of these three layer means (total average voltage) and the standard deviation are calculated. In addition, the system analyzes the distribution pattern of these layer means in the module space, such as observing whether there is a voltage gradient change from the upper layer to the lower layer, or whether the middle layer voltage is significantly lower. All these calculations and analysis results together form the voltage distribution feature set, which quantifies the uniformity or inconsistency of the voltage distribution within the module.

[0060] The same analysis logic is used to process the mean battery temperature measurement data of all layers. The maximum, minimum, range, and standard deviation of the three layer temperature means are calculated, and their spatial distribution patterns are analyzed (e.g., whether the upper layer has the lowest temperature due to its proximity to the heat dissipation port, whether the lower layer also has a lower temperature due to its proximity to the cooling liquid inlet, and whether the middle layer has the highest temperature). These statistical quantities and distribution pattern information form the temperature distribution feature set, reflecting the distribution characteristics and temperature difference of the temperature field within the module.

[0061] The last step is to jointly analyze the generated voltage distribution feature set, temperature distribution feature set, and internal resistance distribution feature set to obtain a single comprehensive health status evaluation value. This process uses a multi-factor fusion algorithm. This algorithm assigns a weight coefficient to each feature set, and the weight coefficient is set based on historical data and battery failure mechanism analysis. For example, the increase in internal resistance is usually more closely related to battery aging, so the internal resistance distribution feature set (especially its range or standard deviation) may be given a higher weight; the uniformity of voltage distribution may affect battery performance and life, with a lower weight; the temperature distribution is important, but its short-term fluctuations may be larger, with a relatively low weight. The weight coefficient can be obtained by analyzing and learning a large amount of historical battery degradation data.

[0062] In the specific calculation, the system scores or quantifies the key indicators (such as range, standard deviation) within each feature set, and then performs a weighted sum according to their weights. For example, the score of the voltage feature set may be based on which preset interval its range (such as 0.02V) falls into to give a score; the temperature feature set is scored based on its maximum temperature difference; and the internal resistance feature set is scored based on its coefficient of dispersion. Then these scores are multiplied by the respective weight coefficients and added together to obtain a preliminary comprehensive score. In order to normalize the result to a range that is easy to understand and compare (such as between 0 and 1, 1 representing brand new), the preliminary comprehensive score will undergo a linear or nonlinear conversion process. The final normalized value is the health state evaluation value of the target power battery module at the current time. This value comprehensively reflects the overall health level of the module in terms of voltage consistency, temperature distribution uniformity, and internal resistance consistency, providing a core basis for subsequent threshold judgment and early warning decision-making.

[0063] Example 2: see Figure 3 When the health state evaluation value is lower than the preset threshold value, the specific process of triggering the life warning measure is triggered. The core of this process is to calculate the health state deviation value, and combined with real-time and historical operation data, through the analysis of the early warning grading model, a quantitative life warning level index is finally generated. Assuming that the lithium-ion power battery module of a certain electric vehicle, after the comprehensive analysis described in Example 1, calculates that its current health state evaluation value is 0.78. The system has a preset health state threshold value of 0.80. When making judgment and analysis, the system confirms that 0.78 is lower than 0.80, and immediately starts the life warning trigger process.

[0064] The first step of this process is to calculate the difference. The system performs a subtraction operation between the health state evaluation value 0.78 and the health state threshold value 0.80: 0.78-0.80=-0.02. This result -0.02 is the health state deviation value. This value is negative, which intuitively indicates the degree to which the current health state is lower than the expected threshold value, and the absolute value size reflects the severity of the deviation. In this case, the deviation value is -0.02, meaning that the health state evaluation value is 2 percentage points lower than the threshold value. The system needs to obtain more extensive context information to deeply understand the meaning and potential risks of this deviation. This includes obtaining the current operating state data and historical operating benchmark data of the target power battery module. The current operating state data reflects the stress level currently experienced by the battery module. The system reads the current charge-discharge rate value from the real-time data stream. For example, at this time the vehicle may be in an accelerating and climbing state, and the actual discharge rate reported by the battery management system is 1.5C. At the same time, the system reads the current ambient temperature value, which is usually measured by an environmental sensor installed in the battery pack, and assumes that the current reported value is 35°C. These two values together depict the severity of the current working condition of the battery.

[0065] The current snapshot data alone is insufficient for a comprehensive assessment, the system also needs to retrieve the long-term operational baseline data of this battery module from the historical database to establish a comparison baseline. The historical operational baseline data is a representative value statistically derived from the vast amount of operational data recorded since the inception of this battery module. The system calculates the average charge-discharge rate of all charge-discharge cycles in the history of this module, which is the historical average charge-discharge rate value. Assuming this vehicle is primarily used for urban commuting, the average discharge intensity is low, and the calculated historical average charge-discharge rate value is 0.7C. Similarly, the system calculates the average value of the ambient temperature experienced during its long-term operation, which is the historical average ambient temperature value. Assuming this vehicle is primarily driven in a temperate climate region, its historical average ambient temperature value is 25°C. These historical baseline values represent the "typical" or "normal" operating conditions experienced by this battery module.

[0066] The system has gathered all the necessary input parameters: the state-of-health deviation value quantifying the extent of state-of-health degradation (-0.02), the current operational state data reflecting the current instantaneous stress (current charge-discharge rate value 1.5C, current ambient temperature value 35°C), and the historical operational baseline data providing long-term reference baselines (historical average charge-discharge rate value 0.7C, historical average ambient temperature value 25°C).

[0067] These data are collectively input into a pre-defined early warning classification model for comprehensive analysis. The early warning classification model is a computational logic module encapsulating expert knowledge and data analysis algorithms. Its design goal is to make a more accurate judgment of the urgency and severity of potential risks by considering the absolute level of state-of-health degradation, the current operational load, and the change in operating conditions relative to historical baselines, rather than relying solely on a simple state-of-health threshold.

[0068] Upon receiving these inputs, the model initiates its internal analysis procedure, which first focuses on the state-of-health deviation value -0.02, which is a clear signal of degradation. Then, the model compares the current charge-discharge rate value 1.5C with the historical average charge-discharge rate value 0.7C. It can be found that the current discharge intensity is much higher than the historical average, indicating that the battery is in a high-load operating state, which can accelerate the further degradation of the already aged battery. At the same time, the model compares the current ambient temperature value 35°C with the historical average ambient temperature value 25°C, and the positive temperature difference of 10°C indicates that the battery is in a relatively hot environment, and high temperature can also intensify the chemical side reactions inside the battery, adversely affecting the state-of-health.

[0069] The core algorithm of the early warning grading model integrates these factors, which may give certain weights to the current high rate and high ambient temperature, as they constitute additional stress on the battery that is already in a sub-healthy state. The logic inside the model can be similar to this: the state of health itself has a small deviation, but it is superimposed on an exceptionally severe operating condition, which makes the actual risk higher than what is shown by simply looking at the state of health deviation value. The model processes and calculates all input parameters through its built-in algorithm and possible weighting.

[0070] After a series of analysis and calculations, the early warning grading model finally outputs a quantitative result, i.e., a life warning level index. This index is a numerical value, for example, it may output a value of 2.8. The index reflects the risk level assessed based on the current state of health deviation and operating conditions. The higher the index value, the more urgent the risk and the higher the level of warning measures that need to be taken. In this example, the generation of the index 2.8 is not only based on the state of health deviation of -0.02, but more importantly, it takes into account the superimposed effect of the current high rate of 1.5C and the high temperature environment of 35°C on the battery that has already appeared to be in decline. The model judges that under this combined stress, the possibility of further accelerated degradation of battery performance significantly increases, so it outputs a risk index of medium to high.

[0071] This life warning level index provides accurate and quantitative decision-making basis for subsequent decisions on what specific warning measures to take. It makes the warning behavior no longer a simple "yes / no" trigger, but can be responded to in stages according to the subtle differences in risk, thus achieving more refined and more reasonable battery health management. The entire process from discovering the state of health deviation threshold to collecting multi-dimensional operating data and finally generating a warning index through an intelligent model completes a complete pre-warning analysis.

[0072] Example 3: It describes the segmented calculation logic inside the preset early warning grading model according to the different intervals of the state of health deviation value, and how to execute the corresponding warning measures according to the final output of the life warning level index. The preset early warning grading model is an algorithm module that contains multiple judgment branches and calculation paths. Its input parameters include the state of health deviation value , the current charge and discharge rate value , the historical average charge and discharge rate value , the current ambient temperature value , and the historical average ambient temperature value . The primary judgment condition of the model is the numerical interval of the state of health deviation value , which is divided into two preset critical values, the first deviation threshold and the second deviation threshold (where ) are divided into three different ranges, and different calculation logics are adopted for each range.

[0073] When the health state deviation value is below the first deviation threshold , it indicates that the current health state degradation is relatively mild and has not entered the interval that requires high vigilance. In this case, the model adopts a conservative calculation strategy, the core of which is to make a weighted correction to the health state deviation value . Specifically, the model is built with a first weight coefficient , which is a constant between 0 and 1, and its value is determined by analyzing the impact of mild deviation on long-term life in historical data. The calculation logic of the model is to scale the original deviation value using the weight coefficient, i.e., to calculate the corrected deviation value . Since , this operation essentially reduces the contribution of mild deviation in subsequent exponential calculation, preventing the system from overreacting to initial, minor performance degradation, thereby avoiding unnecessary early warning interference. This corrected value will be used to generate the final life warning level index .

[0074] When the health state deviation value is between the first deviation threshold and the second deviation threshold , it indicates that the health state of the battery has experienced moderate degradation and requires more attention. At this time, the calculation logic of the model considers the absolute deviation value while introducing the difference between the current operating condition and the historical average operating condition as a correction factor. The model first calculates the difference between the current charge / discharge rate value and the historical average charge / discharge rate value, i.e., the rate difference value . This value quantifies the degree to which the current discharge or charge intensity deviates from its normal level. Subsequently, the model uses this rate difference value to make an exponential correction to the health state deviation value . One specific implementation is embodied in the following formula:

[0075]

[0076] Where: represents the calculated life warning level index, is the input health state deviation value, is a natural constant, is a positive coefficient fitted according to the chemical properties of the battery, used to adjust the amplification effect of the rate difference on the warning level, is the calculated rate difference value. The mathematical property of this formula is that when the rate difference value The value of the exponential term increases rapidly as the state of health deviation becomes larger, thus amplifying the effect of the early warning index . This reflects the fact that when a battery is already somewhat aged, forcing it to work under a much higher than normal load will significantly accelerate its degradation process, thus posing a higher potential risk and requiring a higher early warning level.

[0077] When the state of health deviation is higher than a second deviation threshold , it indicates that the state of health of the battery has already deteriorated significantly and entered a high-risk interval. In this case, the model adopts the most sensitive calculation strategy. In addition to considering the severe state of health deviation itself, the model also introduces the abnormality of the ambient temperature as another key correction factor. The model calculates the absolute difference between the current ambient temperature value and the historical average ambient temperature value, i.e., the temperature difference absolute value . This value reflects the degree of deviation of the current environment of the battery from the normal environment it has experienced historically. Subsequently, the model interacts to correct the state of health deviation value and the temperature difference absolute value . A feasible calculation method is to multiply the two, and then multiply by a proportional coefficient , i.e., . The proportional coefficient is used to adjust the calculation result to the appropriate order of magnitude. This interaction emphasizes a core logic: for a battery with a very poor state of health, an extreme temperature environment (whether too high or too low) will pose a very serious threat to its safety and remaining life, and this "double whammy" effect must be reflected by significantly increasing the early warning level.

[0078] Through the above three differentiated calculation paths, the early warning grading model finally outputs a determined life early warning index . This index is a continuous or graded numerical value, whose size directly corresponds to the risk level determined by the system.

[0079] After generating the life early warning index , the system enters the early warning measure matching and execution phase. The system internally presets several early warning level intervals, which are predefined numerical ranges, and each interval uniquely corresponds to a preset, specific life early warning measure. For example, the system may set: when , it corresponds to "Level 1 Early Warning", and the measure is "record this state of health abnormal event in the maintenance log"; when , it corresponds to "Level 2 Early Warning", and the measure is "display a yellow warning icon and brief prompt information on the vehicle instrument panel or battery management system human-machine interaction interface"; when When this occurs, corresponding to a "Level 3 Warning," the measure is to "trigger the audible and visual alarm and send a voice prompt to the driver"; when In this case, corresponding to a "Level 4 warning", the measure is to "send an emergency alarm message to the remote monitoring center or fleet management personnel through the vehicle-mounted wireless communication module, and suggest arranging maintenance immediately".

[0080] The system will calculate the lifespan warning level index. The system performs a matching analysis against these preset intervals to determine which interval it falls into. Once the matching process is complete, the system automatically triggers and executes the lifespan warning measures corresponding to that interval. The entire process, from model calculation to measure implementation, achieves the classification, quantification, and automated response to battery health risks, ensuring the timeliness of warnings and the targeted nature of measures.

[0081] Example 4: See Figure 4 This paper describes how the system maintains and potentially extends the current assessment cycle when the health status assessment value is determined to be higher than a preset threshold. This process aims to optimize the use of monitoring resources and reduce unnecessary frequent checks when the battery is in good condition. Its core lies in calculating an assessment cycle extension coefficient. Assume that a lithium iron phosphate power battery module in an energy storage power station has completed a periodic health status assessment. Based on the comprehensive analysis in Example 1, its health status assessment value is 0.92. The system's preset health status threshold is 0.85. During the judgment process, the system confirms that 0.92 is higher than 0.85 and immediately initiates the "maintain current assessment cycle" process. This process does not simply keep the original cycle unchanged but includes an intelligent cycle adjustment decision-making mechanism.

[0082] The first step in this process is to calculate the health status redundancy value. The system subtracts the health status assessment value of 0.92 from the health status threshold of 0.85: 0.92 - 0.85 = 0.07. This result, 0.07, is the health status redundancy value. This positive value directly indicates how much the current health status is better than the safety threshold. This value will become an important input parameter for assessing whether the system can "let its guard down."

[0083] The system needs to acquire the life consumption information of this battery module to assist decision making. The system reads two key data from the historical record of the battery management system: the current cycle count and the cycle count design threshold. The current cycle count refers to the total number of complete charge and discharge cycles completed by this module since it was put into operation, assuming that the current record is 1200 times. The cycle count design threshold is the rated cycle life of this type of battery provided by the manufacturer under standard conditions, assuming that the value is 4000 times. These two data together depict the position of the battery module in its theoretical life cycle. The system has already gathered three core parameters required for periodic adjustment decision making: the state of health redundancy value representing the state of health margin (0.07), the current cycle count reflecting the consumed life (1200 times), and the cycle count design threshold representing the total life target (4000 times). These data are input into a preset periodic adjustment model for comprehensive analysis. The model is a calculation module with built-in expert logic and algorithms, and its design goal is to: under the premise of safety, when the battery state of health is good, intelligently extend the interval time of the next health assessment, thereby saving system calculation and communication resources, and reducing operation and maintenance costs.

[0084] The preset periodic adjustment model internally performs differentiated calculation strategies according to the value range of the input state of health redundancy value. This range is divided into two main intervals by a preset critical value - the first redundancy threshold. When the state of health redundancy value is lower than the first redundancy threshold, it indicates that although the current state of health is better than the safety benchmark, the margin is not very significant. In this case, the model adopts a relatively cautious strategy. Its calculation logic is mainly based on the life consumption progress of the battery, i.e. the proportion of the current cycle count to the cycle count design threshold. This proportion value reflects the used degree of the theoretical life of the battery. The built-in algorithm of the model will perform a linear scaling operation according to this proportion value. For example, the higher the life consumption proportion (i.e. the used cycle count is closer to the design threshold), the smaller the allowed assessment period extension range even if there is a certain health redundancy; on the contrary, for a new battery, even if the health redundancy is the same, the allowed extension range can be larger. This calculation method embodies the conservative principle that as the age of the battery increases, even if the state is good, the monitoring frequency should be appropriately increased.

[0085] When the state of health redundancy value is higher than the first redundancy threshold, it indicates that the battery is in very good condition, with significant performance margin. In this case, the model adopts a more aggressive adjustment strategy. Its calculation logic takes into account both the state of health margin and the remaining life potential. The model first performs a logarithmic transformation on the state of health redundancy value. Logarithmic transformation is a mathematical method that can convert a larger numerical value into a new value with gradually slowing growth, which can prevent excessive health redundancy values from causing the calculated extension coefficient to be too large, thereby avoiding unreasonable lengthening of the monitoring interval. At the same time, the model calculates the cycle number margin value, i.e., the cycle number design threshold minus the current cycle number (4000-1200=2800 times), which represents the theoretical remaining cycle life of the battery. Finally, the model multiplies the logarithmic transformation result of the state of health redundancy value with the cycle number margin value to dynamically adjust an evaluation period extension coefficient. This coefficient comprehensively reflects the information of "how good is the current state" and "how long can it be used in theory" in two dimensions.

[0086] To more intuitively show the output of the cycle adjustment model under different inputs, the following table simulates several scenarios. It is emphasized that the data in the table are only examples, and the actual coefficient is calculated accurately by the internal algorithm of the model.

[0087] Table 1: Evaluation period extension coefficient example.

[0088]

[0089] After the comprehensive analysis of the cycle adjustment model, the system finally outputs a specific evaluation period extension coefficient. Assuming that in the current example (redundancy value 0.07, cycle 1200 / 4000 times), the model outputs a coefficient of 1.45. The original basic evaluation period of the system is 7 days. The new evaluation period will be calculated by multiplying the basic period by this extension coefficient: 7 days * 1.45 = 10.15 days. The system may round this result, for example, setting the next evaluation at 10 days later.

[0090] This newly calculated, extended evaluation period will be updated to the system's timing task. This means that the system will not actively initiate a comprehensive state of health evaluation of the module within the next 10 days (but will continue to perform basic safety monitoring), thereby reducing the consumption of computing resources and the burden of data transmission. The entire process realizes a dynamic, state-based predictive maintenance scheduling, seeking the best balance between reliability and economy.

[0091] Example 5: see Figure 5, a step of dynamically updating the health state threshold is described. The core of this step is to make the judgment reference no longer a fixed value, but can be adjusted adaptively according to the actual degradation history of the battery itself and the group experience of similar devices, so as to improve the accuracy and foresight of the early warning system. The starting of the dynamic updating threshold process is derived from the systematic analysis of historical evaluation data. The system maintains a historical health state evaluation value sequence arranged in chronological order, which records the health state evaluation value calculated at the end of each evaluation period since the target power battery module was put into operation. Based on this time series data, the system performs a key calculation: the health state attenuation rate value. The calculation of this value is not simply dependent on the difference between two adjacent time points, but uses a statistical method based on a longer historical window, such as linear fitting of the sequence to calculate its slope, or calculating the average change of the evaluation value in a period of time. This process aims to smooth short-term fluctuations and capture the overall degradation trend of the battery health state on a medium and long-term time scale. The calculated health state attenuation rate value is a signed value, usually negative, and its magnitude reflects the speed of battery performance degradation over time.

[0092] After obtaining the attenuation rate, the system combines it with a preset, unchangeable parameter, the life termination boundary value. The life termination boundary value is usually defined by the battery manufacturer according to product specifications or relevant industry standards, for example, it is often set to 0.7 or 0.8, representing the battery capacity degradation to 70% or 80% of the initial capacity, which is considered to be the end of life. The system uses the health state attenuation rate value and the life termination boundary value to recalculate the health state threshold through a predictive algorithm. The logic of this algorithm is: based on the current degradation rate, predict the remaining time or cycle number required to drop from the latest health state evaluation value to the life termination boundary value. Then, according to a preset early warning advance (for example, it is hoped that the warning will be given 100 cycles or 3 months before the end of life), a new health state threshold corresponding to it is back calculated. This updated threshold is more in line with the actual aging dynamics of the individual battery module, and compared with the fixed global threshold, it can provide more accurate early warning opportunities.

[0093] The dynamic updating threshold step further comprises a data-driven enhancement link, aiming to optimize individual prediction by leveraging the wisdom of the crowd. The system collects degradation characteristic data of similar battery modules from cloud platforms or locally integrated databases through data interfaces. "Similar" is usually defined as battery modules with the same cell model, similar capacity specifications, and similar application scenarios. The system performs data mining and pattern recognition analysis on these massive crowd data to extract common degradation patterns. These patterns may reveal universal laws of specific battery models during the aging process, such as a platform period when the state of health evaluation value drops to around 0.9, or an inflection point feature of accelerated degradation when it falls below 0.75. These extracted knowledge from crowd data is predictive information that individual batteries do not have in their early life.

[0094] The system uses the extracted common degradation patterns to compensate and correct the state of health decay rate value calculated based on its own historical data. For example, if the current decay rate of the battery is calculated based on linear degradation from 0.95 to 0.92, but the common pattern shows that the battery will significantly accelerate after the state of health falls below 0.90, the system will adjust the linear decay rate value to a higher value in a certain interval to reflect the upcoming nonlinear accelerated degradation trend. Thus, a corrected state of health decay rate value is generated, which contains both the historical decay information of the battery and the predictive knowledge of the group performance.

[0095] The system uses this corrected, more accurate state of health decay rate value to run the above predictive algorithm again, combining with the end-of-life boundary value, to recalculate and update the state of health threshold. This makes the threshold updating process not only a simple extrapolation of past behavior, but also an intelligent learning process that integrates both the battery's historical performance and the group's degradation characteristics. The updated threshold is stored in the system configuration, replacing the old threshold, as the basis for the next state of health evaluation value judgment analysis. In this way, the early warning system can continuously evolve, with its judgment basis becoming more and more in line with the actual characteristics of the monitored object, thus achieving more accurate and forward-looking management of battery life.

[0096] It is to be understood that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting; it is not intended to exclude myriad other embodiments of the present application that other inventors can develop based on the same general inventive concepts embodied by the described embodiments. That is, although the present application is described in terms of particular embodiments and implementations, it is to be understood that the terminology used is for the purpose of descriptive clarity and that it is intended to be limited only by the words recited in the appended claims. It is to be understood that the terms such as first and second, etc., merely are used to differentiate one from another without necessarily implying or requiring any actual relationship or order between them. Moreover, the terms "comprises", "comprising", or any other variations thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but can include other elements not expressly listed or inherent to such process, method, article, or apparatus.

[0097] While the embodiments of the application have been shown and described herein, it is to be understood that the application is not limited to these embodiments. Rather, many modifications, changes, substitutions, and alterations can be made thereto without departing from the spirit and scope of the present application as defined by the appended claims and their equivalents.

Claims

1. A method for assessing the health status and predicting the lifespan of a power battery, characterized in that, Includes the following steps: The target power battery module is divided into several evaluation areas, and several measurement locations are randomly selected for each evaluation area. The raw battery status data at each measurement point in each evaluation area within the target power battery module is collected in real time and comprehensively analyzed to obtain the health status evaluation value of the target power battery module. The health status assessment value of the target power battery module is compared with the preset health status threshold for judgment and analysis. If the health status assessment value of the target power battery module is lower than the preset health status threshold, a preset lifespan warning measure will be triggered. If the target power battery evaluation value is higher than the preset health status threshold, the current evaluation cycle will be maintained. The steps for triggering the preset lifespan warning measures include: The difference between the health status assessment value of the target power battery module and the preset health status threshold is calculated to obtain the health status deviation value. The current operating status data and historical operating benchmark data of the target power battery module are obtained, wherein the current operating status data includes the current charge / discharge rate value and the current ambient temperature value, and the historical operating benchmark data includes the historical average charge / discharge rate value and the historical average ambient temperature value. The health status deviation value, current operating status data, and historical operating baseline data are input into a preset early warning classification model for comprehensive analysis to obtain the lifespan early warning level index. The preset early warning classification model selects different calculation logics based on the numerical range in which the health status deviation value falls: When the health status deviation value is lower than the first deviation threshold, the health status deviation value is weighted and corrected using the first weighting coefficient; When the health status deviation value is between the first deviation threshold and the second deviation threshold, an exponential correction is performed based on the difference between the current charge / discharge rate value and the historical average charge / discharge rate value. When the health status deviation value is higher than the second deviation threshold, the absolute value of the temperature difference between the current ambient temperature value and the historical average ambient temperature value is introduced for interactive correction, and the lifespan warning level index is output.

2. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 1, characterized in that, The raw battery status data includes battery voltage, battery temperature, and battery internal resistance values ​​at several time points.

3. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 2, characterized in that, The step of obtaining the health status assessment value of the target power battery module includes: The battery voltage, battery temperature, and battery internal resistance values ​​at each measurement point in each evaluation area within the target power battery module are read and comprehensively analyzed to obtain the average battery voltage, average battery temperature, and average battery internal resistance measurements for each evaluation area. The average battery voltage measurement, average battery temperature measurement, and average battery internal resistance measurement of each evaluation area are comprehensively analyzed to generate the voltage distribution feature set, temperature distribution feature set, and internal resistance distribution feature set of the target power battery module. The health status assessment value is obtained by jointly analyzing the voltage distribution feature set, temperature distribution feature set, and internal resistance distribution feature set.

4. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 1, characterized in that, The step of triggering the preset lifespan warning measure also includes: The lifespan warning level index is matched and analyzed with several preset warning level intervals, wherein each warning level interval corresponds to a preset lifespan warning measure. Execute the lifespan warning measures corresponding to the matched warning level range.

5. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 1, characterized in that, The steps for maintaining the current evaluation cycle include: The health status assessment value of the target power battery module is compared with the preset health status threshold to calculate the difference and obtain the health status redundancy value. Obtain the current cycle count and the cycle count design threshold of the target power battery module; The health status redundancy value, the current cycle count, and the cycle count design threshold are input into a preset cycle adjustment model for comprehensive analysis to obtain the evaluation cycle extension coefficient.

6. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 5, characterized in that, The preset periodic adjustment model performs differentiated calculations based on the numerical range of the health status redundancy value: When the health status redundancy value is lower than the first redundancy threshold, linear scaling is performed based on the ratio of the current loop count to the loop count design threshold. When the health status redundancy value is higher than the first redundancy threshold, the product of the logarithmic transformation result of the health status redundancy value and the remaining value of the number of cycles is used for dynamic adjustment, and the evaluation cycle extension coefficient is output.

7. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 1, characterized in that, It also includes a step for dynamically updating the threshold: Calculate the rate of decline of health status based on the historical health status assessment value sequence; The health status threshold is recalculated by combining the health status decay rate value and the preset lifespan termination boundary value. The updated health status threshold will be used as the benchmark for the next judgment analysis.

8. The method for assessing the health status and providing early warning of the lifespan of a power battery according to claim 7, characterized in that, The dynamic threshold update step further includes: Collect degradation characteristic data of similar power battery modules and extract common degradation patterns; Based on the common degradation mode, the health state decay rate value is compensated and corrected to generate a corrected health state decay rate value. The health status threshold is updated using the corrected health status decay rate value.

Images