Battery abnormality determination method and energy storage device

Abstract

The application provides a battery abnormality determination method and an energy storage device, and relates to the technical field of energy storage devices, aiming to at least solve the problem that only relying on a single pressure difference value or voltage temperature parameter leads to lagging abnormality judgment and poor foresight in the related art. The determination method comprises the following steps: acquiring pressure information collected by a plurality of sensors arranged on the surface of a battery; determining a representation parameter representing the dispersion degree of the pressure of the battery based on the pressure information; determining a local pressure gradient between adjacent sensors based on the pressure values corresponding to the adjacent sensors; and determining whether the battery has a local swelling abnormality based on the representation parameter and the local pressure gradient. Through the dual determination of the overall dispersion degree and the local mutation degree, the application can capture early signs of local swelling earlier, improve the foresight and accuracy of the early warning.

Description

Technical Field

[0001] This invention relates to the field of energy storage device technology, and more specifically, to a method for determining battery anomalies and an energy storage device. Background Technology

[0002] Currently, determining whether a battery exhibits localized expansion or aging anomalies typically relies solely on a single pressure difference or voltage / temperature parameter. This fails to consider the overall dispersion of the pressure distribution (e.g., standard deviation, variance) and characteristic parameters such as local pressure gradients. This results in a lag in the assessment of localized expansion anomalies and poor predictive power. Furthermore, traditional methods struggle to integrate multi-dimensional features (such as global expansion trends and temperature-pressure coupling relationships) for a comprehensive evaluation of battery aging anomalies, hindering early and accurate prediction of health status and remaining lifespan. Therefore, a battery anomaly determination method based on pressure distribution characteristic parameters is urgently needed to improve the predictive power and accuracy of identifying localized expansion and aging anomalies. Summary of the Invention

[0003] The present invention aims to at least solve the problems in related technologies that rely solely on a single pressure difference or voltage and temperature parameter, resulting in delayed anomaly detection, poor foresight, and the inability to integrate multi-dimensional features for aging assessment.

[0004] The first aspect of the present invention provides a method for determining battery anomalies, comprising: acquiring pressure information collected by multiple sensors deployed on the surface of the battery; determining a characterization parameter representing the degree of pressure dispersion of the battery based on the pressure information; determining a local pressure gradient between adjacent sensors based on the pressure values ​​corresponding to adjacent sensors; and determining whether the battery has a local expansion anomaly based on the characterization parameter and the local pressure gradient.

[0005] This invention, through a dual determination of overall dispersion and local mutation, can capture early signs of local expansion earlier, thus improving the foresight and accuracy of early warning.

[0006] In the above technical solution, optionally, the step of determining the local pressure gradient between adjacent sensors based on the pressure values ​​corresponding to adjacent sensors includes: determining the pressure difference between adjacent sensors; and calculating the ratio of the pressure difference to the distance between adjacent sensors to obtain the local pressure gradient.

[0007] In this technical solution, the influence of sensor spacing normalization is eliminated on gradient calculation by introducing sensor spacing normalization, thereby improving the accuracy of anomaly location.

[0008] In the above technical solution, optionally, the step of determining the characterization parameter that characterizes the dispersion of battery pressure based on pressure information includes: performing discrete calculations on the pressure values ​​corresponding to multiple sensors to obtain the characterization parameter; wherein the characterization parameter includes at least one of the following: the standard deviation of all pressure values, the variance of all pressure values, and the coefficient of variation.

[0009] In this technical solution, the uniformity of pressure distribution is characterized by a variety of dispersion indicators, which can be adapted to different battery types and operating conditions, thereby enhancing versatility.

[0010] In the above technical solution, optionally, the step of determining whether the battery has a local expansion anomaly based on the characterization parameters and the local pressure gradient includes: within a preset number of detection cycles, when the growth rate of the characterization parameters exceeds a first threshold, or the maximum local pressure gradient reaches a second threshold, determining that the battery has a local expansion anomaly.

[0011] This technical solution uses both trend change rate and instantaneous mutation warnings to balance foresight and immediacy, reducing the risk of missed reports.

[0012] Optionally, in the above technical solution, before determining the characterization parameter representing the degree of pressure dispersion of the battery based on pressure information, the method further includes: determining whether the battery is in a preset state of charge; and determining the characterization parameter if the battery is in the preset state of charge.

[0013] In this technical solution, pressure analysis is performed under full charge to improve the observability of pressure distribution differences, thereby enhancing the accuracy of anomaly detection.

[0014] In the above technical solution, optionally, the step of determining whether the battery is in a preset state of charge includes: acquiring the battery voltage signal; calculating the pressure change rate corresponding to each sensor; calculating the mean and standard deviation of the pressure change rate; determining the spatial synchronization index based on the ratio of the standard deviation and the mean; and determining whether the battery is in a preset state of charge based on the voltage signal, the mean, and the spatial synchronization index.

[0015] In this technical solution, the robustness and accuracy of fully charged state identification are improved by jointly determining the spatial consistency of voltage and pressure change rates.

[0016] In the above technical solution, optionally, the determination method further includes: determining the average value of the peak pressure of all sensors in each detection cycle based on pressure information, and determining the global expansion coefficient according to the rate of change of the average value with the detection cycle; acquiring temperature information collected by multiple sensors; determining the temperature-pressure coupling coefficient to characterize the degree of correlation between temperature change and pressure change based on pressure information and temperature information; and determining whether the battery has aging abnormalities based on the global expansion coefficient, temperature-pressure coupling coefficient and characterization parameters.

[0017] In this technical solution, by integrating temperature and pressure information, the internal state of the battery can be characterized more comprehensively, thereby improving the accuracy of identifying aging and abnormalities.

[0018] In the above technical solution, optionally, the step of determining whether the battery has aging abnormalities based on the global expansion coefficient, temperature-pressure coupling coefficient, and characterization parameters includes: inputting the global expansion coefficient, temperature-pressure coupling coefficient, and characterization parameters into a pre-trained prediction model to obtain a health status index and a predicted remaining life value; determining whether the battery has aging abnormalities based on the health status index and the predicted remaining life value; wherein, the prediction model is trained based on historical sample data, and the historical sample data includes the global expansion coefficient, temperature-pressure coupling coefficient, characterization parameters, health status index, and predicted remaining life value of multiple historical batteries at different stages.

[0019] In this technical solution, a prediction model trained based on historical data is introduced to achieve nonlinear fusion of multiple feature parameters, thereby improving the accuracy of battery health status assessment.

[0020] In the above technical solution, optionally, the step of determining whether the battery is aging abnormally based on the health status index and the remaining life prediction value includes: when either the health status index or the remaining life prediction value is less than its corresponding standard value, the battery is determined to be aging abnormally.

[0021] This technical solution improves the sensitivity and reliability of identifying aging anomalies by jointly judging the dual indicators of health status and remaining lifespan.

[0022] Optionally, the above technical solution further includes: when it is determined that there is a local swelling abnormality in the battery, performing at least one of the following: generating a charging current limiting command, which is used to limit the charging current of the battery to below a preset safe current; displaying the location of the swelling area; and reporting to a remote monitoring platform.

[0023] This technical solution improves the safety and maintainability of the battery system through multi-level response mechanisms including current limiting, location tracking, and remote reporting.

[0024] In the above technical solution, optionally, multiple sensors are distributed in multiple areas on the surface of the battery, including the tab area, the center of the cell and the edge area of ​​the cell.

[0025] In this technical solution, by deploying sensors in different areas of the battery, the spatial coverage of pressure information collection is improved, thereby enhancing the accuracy of anomaly detection.

[0026] A second aspect of the present invention provides an energy storage device, comprising: a battery and a plurality of sensors disposed on the surface of the battery; a battery anomaly determination device, the battery anomaly determination device being used to execute the battery anomaly determination method of any of the above-described technical solutions of the present invention. Attached Figure Description

[0027] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:

[0028] Figure 1 One of the flowcharts for determining a battery malfunction according to an embodiment of the present invention is shown;

[0029] Figure 2 A second flowchart of a method for determining battery abnormalities according to an embodiment of the present invention is shown;

[0030] Figure 3 A flowchart of a method for determining battery abnormalities according to an embodiment of the present invention is shown in part three.

[0031] Figure 4 This diagram illustrates the connection structure of a battery, a data aggregation and synchronization unit, and a main controller according to an embodiment of the present invention.

[0032] Figure 5 A structural block diagram of an energy storage device according to an embodiment of the present invention is shown;

[0033] Figure 6 A flowchart illustrating the determination of a local pressure gradient according to an embodiment of the present invention is shown.

[0034] The system includes 100 energy storage devices, 1 battery, 2 data aggregation and synchronization units, 3 main controllers, 4 sensors, and 5 battery anomaly detection devices. Detailed Implementation

[0035] To better understand the above-mentioned objectives, features, and advantages of this application, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0036] Many specific details are set forth in the following description in order to provide a full understanding of this application. However, this application may also be implemented in other ways different from those described herein. Therefore, the scope of protection of this application is not limited to the specific embodiments disclosed below.

[0037] like Figure 1 As shown, the first aspect of the present invention provides a method for determining battery abnormalities, comprising:

[0038] S102: Acquire pressure information collected by multiple sensors deployed on the surface of the battery;

[0039] S104: Based on pressure information, determine the characterization parameters that characterize the degree of dispersion of battery pressure;

[0040] S106: Determine the local pressure gradient between adjacent sensors based on the pressure values ​​corresponding to adjacent sensors;

[0041] S108: Based on characterization parameters and local pressure gradient, determine whether there is local expansion anomaly in the battery.

[0042] In this technical solution, pressure sensors deployed in multiple areas on the battery surface (e.g., the tab area, the cell center, and the cell edge area) first collect pressure values ​​at each measuring point in real time. Then, the collected pressure values ​​are discretely calculated to obtain characterization parameters reflecting the overall uniformity of the pressure distribution, such as standard deviation, variance, or coefficient of variation. These parameters reflect the overall dispersion of the pressure distribution on the battery surface. Simultaneously, the pressure difference between adjacent sensors is calculated, and this difference is divided by the distance between the sensors to obtain the local pressure gradient. This gradient reflects the degree of pressure change in a local area, effectively identifying local bulges or indentations. Finally, the characterization parameters and the local pressure gradient are combined to determine whether the battery exhibits abnormal local expansion. This invention, through the dual determination of overall dispersion and local abrupt changes, can capture early signs of local expansion more quickly, improving the foresight and accuracy of early warning.

[0043] In the above technical solutions, optionally, such as Figure 6 As shown, the steps for determining the local pressure gradient between adjacent sensors based on the pressure values ​​corresponding to adjacent sensors include:

[0044] S602: Determine the pressure difference between adjacent sensors;

[0045] S604: Calculate the ratio of the pressure difference to the distance between adjacent sensors to obtain the local pressure gradient.

[0046] In this technical solution, to improve the calculation accuracy of the local pressure gradient and eliminate the influence of sensor density, the difference between the pressure values ​​measured by two adjacent sensors is first calculated to obtain the pressure difference. Then, the actual physical distance between these two sensors is obtained (e.g., calculated based on preset sensor layout coordinates), and the pressure difference is divided by the distance to obtain the rate of pressure change per unit distance, i.e., the local pressure gradient. This gradient value can objectively reflect the drastic change in pressure in space, and is independent of sensor density, thus making batteries with different layout schemes comparable and enabling precise location of pressure abrupt changes.

[0047] In this technical solution, the influence of sensor spacing normalization is eliminated on gradient calculation by introducing sensor spacing normalization, thereby improving the accuracy of anomaly location.

[0048] In the above technical solution, optionally, the step of determining the characterization parameter that characterizes the dispersion of battery pressure based on pressure information includes: performing discrete calculations on the pressure values ​​corresponding to multiple sensors to obtain the characterization parameter; wherein the characterization parameter includes at least one of the following: the standard deviation of all pressure values, the variance of all pressure values, and the coefficient of variation.

[0049] In this technical solution, to comprehensively characterize the non-uniformity of the pressure distribution on the battery surface, various statistical indicators can be used as characterization parameters. The standard deviation is the square root of the average of the squares of the differences between each pressure value and the mean, reflecting the overall fluctuation of the pressure values; the variance is the square of the standard deviation, also reflecting the degree of dispersion; the coefficient of variation is the ratio of the standard deviation to the mean, suitable for comparing the degree of dispersion under different pressure levels. In practical applications, one or more indicators can be selected based on the battery type, the number of sensors, and computing resources. By calculating the dispersion of pressure values ​​from all sensors, the overall uniformity of the pressure distribution on the battery surface can be quantitatively described, providing a quantitative basis for anomaly detection.

[0050] In this technical solution, the uniformity of pressure distribution is characterized by a variety of dispersion indicators, which can be adapted to different battery types and operating conditions, thereby enhancing versatility.

[0051] In the above technical solution, optionally, the step of determining whether the battery has a local expansion anomaly based on the characterization parameters and the local pressure gradient includes: within a preset number of detection cycles, when the growth rate of the characterization parameters exceeds a first threshold, or the maximum local pressure gradient reaches a second threshold, determining that the battery has a local expansion anomaly.

[0052] In this technical solution, two independent triggering conditions are employed to balance proactive early warning and immediate alarm. The first condition focuses on the changing trend of a characterizing parameter (e.g., standard deviation) over time: within several consecutive detection cycles (e.g., 5 cycles), the growth rate of the characterizing parameter (i.e., the percentage increase in the current cycle's characterizing parameter value relative to the previous cycle or baseline cycle) is calculated. If the growth rate exceeds a preset first threshold (e.g., 20%), it indicates that the overall non-uniformity of the pressure distribution is accelerating, a signal of early local expansion, and the system issues an early warning. The second condition focuses on the maximum value of the local pressure gradient: the local pressure gradient between all adjacent sensors is calculated in real time, and the maximum value is identified. If this maximum value exceeds a preset second threshold (e.g., 5 kPa / mm), it indicates a significant local pressure abrupt change, and the battery may have already experienced local bulging, prompting the system to immediately issue an alarm. The two conditions use an "OR" logic; satisfying either condition determines an abnormal local expansion, thus enabling both the detection of slowly deteriorating trends and a rapid response to sudden bulging.

[0053] It is important to understand that when there are multiple characterization parameters, if the growth rate of any one of the characterization parameters exceeds a preset first threshold, the battery is considered to have localized expansion anomalies.

[0054] This technical solution uses both trend change rate and instantaneous mutation warnings to balance foresight and immediacy, reducing the risk of missed reports.

[0055] Optionally, in the above technical solution, before determining the characterization parameter representing the degree of pressure dispersion of the battery based on pressure information, the method further includes: determining whether the battery is in a preset state of charge; and determining the characterization parameter if the battery is in the preset state of charge.

[0056] In this technical solution, since the pressure distribution of the battery is closely related to its state of charge (SOC), and the pressure values ​​differ significantly under different SOCs, the feature extraction time is uniformly limited to a preset SOC to ensure the comparability of characterization parameters and the accuracy of anomaly detection. The preset SOC is preferably the fully charged state (SOC=100%), because the battery expansion is most significant and the pressure distribution difference is greatest under full charge, making it easiest to observe abnormal features. Specifically, the SOC value is monitored in real time by the battery management system (BMS). Characterization parameter calculation and subsequent judgment are only triggered when the SOC reaches the preset value (e.g., 100%). Optionally, the preset SOC is SOC greater than or equal to 90%. If the battery has not reached the preset SOC, monitoring continues without anomaly detection. This limitation eliminates the interference of SOC fluctuations on feature values, making the characterization parameters comparable across different cycle periods.

[0057] In this technical solution, pressure analysis is performed under full charge to improve the observability of pressure distribution differences, thereby enhancing the accuracy of anomaly detection.

[0058] In the above technical solution, optionally, the step of determining whether the battery is in a preset state of charge includes: acquiring the battery voltage signal; calculating the pressure change rate corresponding to each sensor; calculating the mean and standard deviation of the pressure change rate; determining the spatial synchronization index based on the ratio of the standard deviation and the mean; and determining whether the battery is in a preset state of charge based on the voltage signal, the mean, and the spatial synchronization index.

[0059] In this technical solution, to more robustly determine whether the battery has reached full charge, a spatial consistency criterion for pressure change rate is introduced in addition to the traditional voltage threshold judgment. First, the battery voltage signal collected by the BMS is acquired. Then, the pressure change rate of each sensor at the current moment (i.e., the derivative of pressure with respect to time) is calculated, and the arithmetic mean and standard deviation of the pressure change rates of all sensors are calculated. Next, the spatial synchronization index is calculated, defined as the ratio of the standard deviation to the mean, used to characterize the consistency of the pressure change rate at each measuring point. When the battery voltage reaches the voltage threshold (e.g., a single cell voltage of 4.2V), the average pressure change rate of all sensors is not higher than the change rate threshold (e.g., 0.8), and the spatial synchronization index is not lower than the synchronization index threshold (e.g., 0.8), and all three conditions are simultaneously met and maintained for a preset duration (e.g., 30 seconds), the battery is determined to be in a fully charged state. This joint judgment method is more reliable than simply relying on voltage and can avoid misjudgments caused by voltage fluctuations or single-point sensor failures. Of course, depending on different standards, if any one of the following three conditions is met—battery voltage, the average rate of change of pressure from all sensors, or the spatial synchronization index—the battery can also be considered to be fully charged.

[0060] In this technical solution, the robustness and accuracy of fully charged state identification are improved by jointly determining the spatial consistency of voltage and pressure change rates.

[0061] In the above technical solution, optionally, the determination method further includes: determining the average value of the peak pressure of all sensors in each detection cycle based on pressure information, and determining the global expansion coefficient according to the rate of change of the average value with the detection cycle; acquiring temperature information collected by multiple sensors; determining the temperature-pressure coupling coefficient to characterize the degree of correlation between temperature change and pressure change based on pressure information and temperature information; and determining whether the battery has aging abnormalities based on the global expansion coefficient, temperature-pressure coupling coefficient and characterization parameters.

[0062] In this technical solution, to assess the overall aging state of the battery, a multi-dimensional feature fusion-based method for judging aging anomalies is further extended. First, in each detection cycle (e.g., each charge-discharge cycle), the peak pressure of each sensor within that cycle is recorded, and the arithmetic mean of the peak pressures from all sensors is calculated to obtain the global expansion trajectory. Then, the rate of change of this average value with the detection cycle is calculated to obtain the global expansion coefficient, which reflects the cumulative rate of irreversible expansion of the battery and is positively correlated with capacity decay. Simultaneously, battery surface temperature information is collected by sensors, and the Pearson correlation coefficient between the pressure field and temperature field at the same moment is calculated to obtain the temperature-pressure coupling coefficient. This coefficient reflects the change in the thermo-mechanical coupling mode caused by the increase in internal resistance during aging. Finally, the global expansion coefficient, the temperature-pressure coupling coefficient, and the aforementioned characterization parameters are used as multi-dimensional features to comprehensively determine whether the battery exhibits aging anomalies. For example, when the global expansion coefficient continuously increases, the temperature-pressure coupling coefficient deviates from the normal range, and the characterization parameters rise, it can be determined that the battery has entered an accelerated aging stage.

[0063] In this technical solution, by integrating temperature and pressure information, the internal state of the battery can be characterized more comprehensively, thereby improving the accuracy of identifying aging and abnormalities.

[0064] In the above technical solution, optionally, the step of determining whether the battery has aging abnormalities based on the global expansion coefficient, temperature-pressure coupling coefficient, and characterization parameters includes: inputting the global expansion coefficient, temperature-pressure coupling coefficient, and characterization parameters into a pre-trained prediction model to obtain a health status index and a predicted remaining life value; determining whether the battery has aging abnormalities based on the health status index and the predicted remaining life value; wherein, the prediction model is trained based on historical sample data, and the historical sample data includes the global expansion coefficient, temperature-pressure coupling coefficient, characterization parameters, health status index, and predicted remaining life value of multiple historical batteries at different stages.

[0065] In this technical solution, a data-driven prediction model is employed to achieve quantitative assessment of aging status. First, under laboratory conditions, multiple identical batteries undergo cyclic aging tests. The global expansion coefficient, temperature-pressure coupling coefficient, and characterization parameters of each battery are recorded at different cycle counts. Simultaneously, a capacity test is conducted to obtain the State of Health Index (SOH, the ratio of current capacity to rated capacity) and Remaining Life (RUL, the estimated number of usable cycles remaining). These data are compiled into a training dataset. Then, a regression model (such as support vector regression, random forest, or neural network) is selected for training. The global expansion coefficient, temperature-pressure coupling coefficient, and characterization parameters are used as inputs, while the State of Health Index and Remaining Life Predicted Value are used as outputs. The model parameters are optimized by minimizing the prediction error. After training, the multi-dimensional features calculated in real-time are input into the model, which then outputs the current battery's State of Health Index and Remaining Life Predicted Value. When the State of Health Index falls below a preset threshold (e.g., 80%) or the Remaining Life Predicted Value falls below a safety limit, the battery is deemed to have aging abnormalities.

[0066] In this technical solution, a prediction model trained based on historical data is introduced to achieve nonlinear fusion of multiple feature parameters, thereby improving the accuracy of battery health status assessment.

[0067] In the above technical solution, optionally, the step of determining whether the battery is aging abnormally based on the health status index and the remaining life prediction value includes: when either the health status index or the remaining life prediction value is less than its corresponding standard value, the battery is determined to be aging abnormally.

[0068] In this technical solution, to ensure the sensitivity and reliability of aging anomaly detection, an "OR" logic is used: if the State of Health (SOH) is lower than a preset health threshold (e.g., 85%), or the Remaining Life Predicted Value (RUL) is lower than a preset lifespan threshold (e.g., 50 cycles), the battery is determined to have an aging anomaly. This is because battery aging can manifest as rapid capacity decay (low SOH) but with many remaining cycles, or it can manifest as a short remaining lifespan but a still high SOH (e.g., a sharp increase in internal resistance but no significant capacity decrease). Using dual indicators for joint judgment can cover different types of aging modes and avoid missed detections. Once an aging anomaly is detected, the system can issue a lifespan reminder, prompting the user to arrange maintenance or battery replacement.

[0069] This technical solution improves the sensitivity and reliability of identifying aging anomalies by jointly judging the dual indicators of health status and remaining lifespan.

[0070] Optionally, the above technical solution further includes: when it is determined that there is a local swelling abnormality in the battery, performing at least one of the following: generating a charging current limiting command, which is used to limit the charging current of the battery to below a preset safe current; displaying the location of the swelling area; and reporting to a remote monitoring platform.

[0071] In this technical solution, a multi-level response mechanism is designed to take timely safety measures after detecting localized expansion anomalies. Specifically, after determining that there is a localized expansion anomaly, the Battery Management System (BMS) can perform one or more of the following operations: First, it generates a charging current limiting command to reduce the battery's maximum allowable charging current to a preset safe current value, thereby mitigating gas production and expansion deterioration and preventing thermal runaway; Second, by analyzing the location of the sensor pairs corresponding to the maximum local pressure gradient, it determines the specific area where the bulging occurs and displays the location of the bulging area on a local display screen or mobile terminal, facilitating rapid location by maintenance personnel; Third, it reports the anomaly information (including anomaly type, occurrence time, location information, etc.) to a remote monitoring platform (such as a cloud server or operations and maintenance center) to achieve remote alarm and data recording. Through the above multi-level response, the spread of risk can be effectively controlled, improving the overall safety of the battery system.

[0072] This technical solution improves the safety and maintainability of the battery system through multi-level response mechanisms including current limiting, location tracking, and remote reporting.

[0073] In the above technical solution, optionally, multiple sensors are distributed in multiple areas on the surface of the battery, including the tab area, the center of the cell and the edge area of ​​the cell.

[0074] In this technical solution, to improve the spatial coverage and representativeness of pressure information acquisition, the sensor placement is optimized based on the spatial heterogeneity of electrochemical reactions. Specifically, sensors are placed in the battery's tab region (where current density is highest, Joule heat and internal stress are concentrated, making it a key area for monitoring local overheating and mechanical stress concentration caused by overcharging / over-discharging), the cell center (where expansion is usually most significant), and the cell edge (where different constraints make local bulging more likely). By placing sensing units in these key areas, the spatial characteristics of the battery surface pressure distribution can be captured more comprehensively, improving the detection sensitivity for local expansion anomalies and aging anomalies. Simultaneously, this layout strategy also reduces the number of sensors, lowering system costs while maintaining monitoring effectiveness.

[0075] In this technical solution, by deploying sensors in different areas of the battery, the spatial coverage of pressure information collection is improved, thereby enhancing the accuracy of anomaly detection.

[0076] Another embodiment of the present invention provides a method for determining battery malfunctions.

[0077] It is important to understand that existing battery management systems primarily rely on voltage, current, and a limited number of temperature points for battery state estimation. They cannot detect changes in the internal mechanical state of the battery due to aging, nor can they capture its spatial heterogeneity. This results in delayed and inaccurate predictions of battery health and remaining lifespan, and an inability to provide early warnings of potential safety hazards such as localized swelling. Therefore, there is an urgent need for a method capable of sensing battery degradation characteristics at a two-dimensional spatiotemporal level.

[0078] The method provided in this embodiment is based on a distributed thin-film temperature and pressure integrated sensing network, enabling the detection of battery anomalies, including localized expansion anomalies, and the prediction of battery health status and lifespan. Its core idea is to synchronously and in-situ acquire two-dimensional pressure and temperature field signals of the battery during operation through a flexible distributed sensor array tightly attached to the battery surface; extract a set of "field characteristic" physical quantities from this spatiotemporal signal that characterize overall battery aging and localized failures; finally, using a prediction model that integrates spatial and temporal information, these field characteristics are mapped into accurate SOH estimates, RUL predictions, and safety risk warning signals. This solution achieves a fundamental paradigm shift in battery management from "black-box indirect estimation" to "transparent direct sensing," and from "single-point time series" to "spatiotemporal field analysis."

[0079] like Figure 2 As shown, the method for determining battery malfunctions provided in this embodiment includes the following steps:

[0080] S202: Construct a distributed sensor network and collect data.

[0081] Specifically, for prismatic pouch cells, the sensor distribution needs to balance the spatial heterogeneity of electrochemical reactions with the feasibility of engineering implementation. A typical layout is based on matrix-like coverage of key areas, such as... Figure 4 As shown, battery 1 has an electrode tab region, a cell center region, and an edge region. Sensors are placed in the electrode tab region (first and second sensing nodes), where the current density is highest and Joule heating and internal stress are concentrated. This region is key to monitoring local overheating and mechanical stress concentration caused by overcharging / over-discharging. Sensors are placed in the cell center and edge regions (third to seventh sensing nodes). The expansion is usually most significant in the center region (e.g., the fifth sensing node), while the constraints on the surrounding edges (third, fourth, sixth, and seventh sensing nodes) are different. Their stress-temperature data are used to analyze the uniformity of overall expansion and identify local bulges.

[0082] After each sensor node completes data acquisition, the data is uniformly aggregated to the data aggregation and synchronization unit 2. This unit serves as the core connecting sensor node between the distributed sensor network and the main control layer. It calibrates the timestamps of data from multiple sensor nodes, unifies the format, and filters out outliers, thereby achieving time synchronization and standardized integration of sensor data from different locations. The multi-dimensional monitoring data processed by the data aggregation and synchronization unit 2 will be transmitted to the main controller 3 (BMS). As the core control center of the battery system, the main controller 3 receives the synchronization data from this unit and analyzes the SOH (State of Health) and RUL (Relative Lifetime) based on the built-in CNN-LSTM (CNN stands for Convolutional Neural Network and LSTM stands for Network-Long Short-Term Memory) model and other algorithms. At the same time, it triggers first-level / second-level early warnings and executes active safety measures based on data feedback, and sends control commands back to each sensor node or battery actuator through the data aggregation and synchronization unit, forming a closed-loop architecture of "sensor acquisition - data synchronization - intelligent control - feedback optimization", realizing efficient fusion of data from key monitoring areas on the battery surface and full life cycle control.

[0083] To process multidimensional data, a spatial coordinate system for the sensor network is first established. Let the battery surface be a two-dimensional plane, and the coordinates of the i-th sensor node be (x, y). i y i At each sampling time t, the system collects a multi-channel data vector M(t):

[0084] M(t)=[P1(t), T1(t), P2(t), T2(t),…,P N (t), T N (t)] T ;

[0085] Where N is the total number of sensor nodes, P i (t) and T i (t) represents the compensated pressure and temperature at node i, respectively.

[0086] S204: Determine if the battery is fully charged.

[0087] Specifically, since the pressure distribution of a battery is closely related to its state of charge (SOC), the pressure values ​​vary significantly under different SOCs. To ensure the comparability and consistency of subsequently extracted feature parameters (such as the aforementioned characterization parameters (i.e., pressure distribution non-uniformity), global expansion trajectory, local pressure gradient, and temperature-pressure coupling coefficient), feature extraction needs to be performed under a uniform preset state of charge. The fully charged state (SOC=100%) is the optimal observation window, as the battery expansion is most significant and the pressure distribution difference is greatest at this time, best reflecting the irreversible expansion characteristics caused by aging. Therefore, before extracting feature parameters, it is necessary to accurately determine whether the battery has reached a fully charged state.

[0088] During the final stage of constant current charging, the battery expands due to the insertion of lithium ions. As charging nears completion, the expansion of all regions should synchronously cease. Based on this physical law, a "spatial synchronization index" S(t) is defined to quantify the consistency of the pressure change rate at each measuring point.

[0089] ;

[0090] in, and These are the pressure change rates at all nodes. The mean and standard deviation.

[0091] Physical meaning: When all points stop expanding synchronously, As S(t) approaches 0, S(t) approaches 1. The joint condition for determining full charge is:

[0092] Condition A: Battery voltage V(t) ≥ voltage threshold;

[0093] Condition B: Mean pressure change rate ≤γ (γ is 0.8);

[0094] Condition C: Spatial synchronization index S(t) ≥ δ (δ is 0.8).

[0095] If conditions A, B, and C are simultaneously satisfied and last for a time ΔT, then the cell is considered fully charged. However, depending on the specific circumstances, if any one of A, B, or C is satisfied, the cell is considered fully charged.

[0096] S206: Prediction of Health Status (SOH) and Life Length (RUL).

[0097] Specifically, the core feature parameters are extracted, and the calculation methods and physical association information of each feature parameter are shown in Table 1 below:

[0098] Table 1

[0099]

[0100] In Table 1, n is the cycle number and N is the total number of sensor nodes. This is the global expansion trajectory. This represents the peak pressure recorded by the i-th sensor in the nth iteration; This refers to the non-uniformity of pressure distribution. This is the maximum value of the spatial pressure gradient. These are the pressure values ​​measured by the i-th and j-th sensors at the moment of full charge, respectively. Let be the spatial distance between the i-th sensor and the j-th sensor. This represents the set of pressure values ​​from all n sensors at the moment of full charge. Represents the set of temperature values ​​from all n sensors at full charge, where Corr is the Pearson correlation coefficient calculation function. This is the temperature-pressure coupling coefficient.

[0101] After calculating the above feature parameters, a deep learning prediction framework with the feature parameters as input is constructed.

[0102] The input layer consists of a three-dimensional tensor for m historical loops within a time window. Where m is the time step (number of cycles), N is the number of spatial nodes, and f is the number of features per node (including global expansion trajectory, pressure distribution non-uniformity, and temperature-pressure coupling coefficient). The model architecture adopts a CNN-LSTM hybrid model:

[0103] Spatial feature extraction: One-dimensional convolutional layers (in the sensor node dimension) are used to automatically learn spatial correlation patterns and capture features such as bulge propagation.

[0104] Time series modeling: The extracted spatial feature sequence is input into the LSTM layer to learn the evolution of aging over time.

[0105] Fusion and Prediction: Finally, the current SOH estimate and the probability distribution of the future RUL are output through the fully connected layer.

[0106] S208: Online Deployment and Early Warning.

[0107] Specifically, the trained model parameters are stored in the BMS's Flash memory. During real-time system operation, after each loop, the feature parameters from the most recent 30 loops are automatically input into the model to obtain real-time SOH and RUL. A two-level early warning mechanism is implemented.

[0108] Level 1 Warning (Battery Life Reminder): When the predicted SOH is less than 85%, or the RUL is less than the RUL threshold, it will be marked in the system log and the user will be notified that the battery has entered the aging period.

[0109] Level 2 Warning (Safety Alarm): When the rate of increase of pressure distribution non-uniformity exceeds 50% for three consecutive detection cycles, or when the maximum local pressure gradient suddenly exceeds the threshold (e.g., 10 kPa / mm), the system determines that there is a high-risk local bulging. The BMS immediately issues an alarm and limits the charging current to 0.2C via the CAN bus. At the same time, it reports the bulging risk location information (by analyzing the sensor pair (i, j) corresponding to the maximum local pressure gradient to locate the bulging area) and reports it to maintenance personnel.

[0110] Through the above steps, the method provided in this embodiment can identify local bulging and aging abnormalities earlier and more accurately, significantly improving the safety and predictive capabilities of the battery management system.

[0111] like Figure 3 As shown, another embodiment of the present invention discloses a method for determining battery malfunctions, comprising the following steps:

[0112] S301, the sensor collects pressure and temperature data in real time;

[0113] Specifically, flexible thin-film pressure-temperature integrated sensors are deployed in multiple areas on the battery surface (such as the tab area, the center of the cell, and the edge of the cell) to collect pressure and temperature values ​​at each measuring point in real time at a set sampling frequency (e.g., 10Hz). The collected data is preprocessed by signal conditioning circuits (such as amplifiers and filters), then converted into digital signals by an analog-to-digital converter, and transmitted to the microcontroller or dedicated processor of the battery management system (BMS). Each sensor node has unique spatial coordinates (x, y, y). i y i The collected data forms a multi-channel data vector, containing the pressure P of each detection node. i (t) and temperature T i (t).

[0114] S302, extract the feature vector F(n) at the end of each loop;

[0115] Specifically, after a charge-discharge cycle (e.g., from full charge to full discharge and back to full charge), the system extracts a set of feature parameters characterizing the battery state based on the pressure and temperature data collected throughout the cycle, forming a feature vector F(n). The feature vector includes, but is not limited to:

[0116] F1 (Global Expansion Trajectory): Calculates the average of the peak pressures from all sensors during this cycle, reflecting the cumulative irreversible expansion of the battery as a whole.

[0117] F2 (Pressure Distribution Non-uniformity): Calculates the standard deviation of the pressure values ​​of each sensor at full charge, characterizing the overall dispersion of the pressure distribution.

[0118] F3 (Pressure Distribution Non-uniformity Growth Rate): The rate of change of the current cycle F2 relative to the previous few cycles F2, used to identify the accelerating trend of non-uniformity.

[0119] F4 (Maximum Local Pressure Gradient): Calculates the ratio of the pressure difference between all adjacent sensors to the distance between them, and takes the maximum value to detect sudden changes in local bulges.

[0120] F5 (Temperature-Pressure Coupling Coefficient): Calculates the temperature-pressure coupling coefficient at the same moment, reflecting the change in the thermo-mechanical coupling mode.

[0121] S303, input the feature vector into the CNN-LSTM model to obtain SOH (Health Status Index) and RUL (Life Deficit Value).

[0122] Specifically, the feature vectors from the most recent 30 cycles are input into a pre-trained CNN-LSTM hybrid prediction model. The model first extracts local spatial correlation patterns (such as bulge propagation features) in the sensor space dimension through a one-dimensional convolutional layer. Then, it inputs the spatial feature sequence into an LSTM layer to learn temporal dependencies (such as aging evolution patterns). Finally, it outputs the current cycle's health status index (SOH) (percentage, e.g., 85%) and remaining lifespan prediction (RUL) (estimated number of remaining cycles, e.g., 200) through a fully connected layer. This model has been trained based on historical aging data, and its parameters are stored in the BMS's Flash memory.

[0123] S304, determine if SOH is less than 85%; if yes, proceed to S305; if no, proceed to S306.

[0124] Specifically, the SOH value output by the model is compared with a preset health threshold (85%). If SOH < 85%, it indicates that the battery has entered a significant lifespan degradation period, and the process proceeds to step S305 to trigger a level one warning; if SOH ≥ 85%, it indicates that the battery health status is still acceptable, and the process proceeds to step S306 to further check other characteristic parameters.

[0125] Understandably, in this step, aging can be determined based on SOH, RUL, or both.

[0126] S305 triggers a Level 1 warning, logs are recorded and the user is notified, then proceeds to S313;

[0127] Specifically, the first-level warning is the "lifespan reminder" level. The BMS records the current cycle number, SOH value and trigger time in the internal log, and at the same time prompts the user through the human-machine interface (such as LCD (Liquid-Crystal-Display) screen, mobile APP (Application) or buzzer) that "the battery has entered the aging period and it is recommended to arrange maintenance or replacement".

[0128] S306, determine whether one of the following conditions is met: F3 increases by more than 50% for three consecutive times or F4 increases by more than 30 kPa / mm; if yes, proceed to S307; if no, proceed to S312.

[0129] Specifically, check two conditions:

[0130] Condition 1: The F3 values ​​of the current cycle and the two preceding consecutive cycles (a total of three cycles) are all greater than 50%, indicating that the unevenness of pressure distribution is deteriorating rapidly, which is an early trend signal of local bulging.

[0131] Condition 2: The current cycle's F4 exceeds the preset second threshold of 30 kPa / mm, indicating that there is a significant local pressure change point, and the battery is likely to have bulged.

[0132] If any one of the conditions is met, it is determined that there is a high-risk local bulge, and the process proceeds to step S307 to trigger a level 2 warning; if neither condition is met, the process proceeds to step S312 to maintain the normal state.

[0133] S307, triggering a Level II warning;

[0134] S308 implements proactive safety measures, including: limiting the current to 0.2C via the CAN (Controller Area Network) bus, locating the bulging area and displaying it on the screen, and reporting it to the cloud platform;

[0135] Specifically, the BMS simultaneously implements the following three measures:

[0136] Current limiting: Sending instructions to the charging controller via the CAN bus to limit the maximum allowable charging current of the battery to 0.2C (e.g., 3A for a 15Ah battery) to slow down gas production and swelling, and prevent thermal runaway.

[0137] Location display: Based on the bulging area located in step S307 (such as "center to the left of the cell"), the abnormal location is highlighted on the local display screen or mobile terminal interface, and the pressure gradient value is marked to facilitate quick location by maintenance personnel.

[0138] Reporting to the cloud: Abnormal information (including battery ID, time, SOH, F3, F4, and location coordinates) is reported to the remote monitoring platform via 4G / 5G or WiFi module to achieve alarm recording and remote notification.

[0139] S309, determine whether the warning has been lifted within 30 minutes; if yes, proceed to S310; if no, proceed to S311.

[0140] S310, the warning is lifted, the system returns to normal, and then proceeds to S313;

[0141] S311, the system enters the safety lock state, cuts off charging and discharging, and then proceeds to S313;

[0142] Specifically, if the risk is not eliminated within 30 minutes of a Level 2 warning being triggered, indicating that the swelling may continue to worsen, the BMS (Battery Management System) will switch the system to a safety lockout state: disconnecting the main relay or contactor to completely cut off the battery's charging and discharging circuit, prohibiting any energy input or output. Simultaneously, the screen will display "Battery fault, please send for repair immediately," and the system will report to the cloud again. This lockout state requires manual intervention (such as on-site inspection and reset by a professional repair technician) to be lifted.

[0143] S312, normal state, update display;

[0144] S313, update the display and prompt the user.

[0145] The method provided in this embodiment, through feature parameters extracted by a distributed sensor network, particularly the pressure field non-uniformity index and the maximum local pressure gradient, can capture early, localized aging signals that are undetectable by traditional single-point sensors and voltage methods. Combined with the powerful learning capabilities of the CNN-LSTM hybrid model for spatiotemporal patterns, this invention achieves the following beneficial effects:

[0146] Earlier and more accurate SOH / RUL prediction: The prediction point can be predicted at least 50 cycles earlier than methods based solely on capacity decay backtracking.

[0147] Unique safety warning function: It can detect and accurately locate battery swelling at an early stage (before it is visible to the naked eye), which greatly improves the safety of the system.

[0148] Strong anti-interference capability: Multi-sensor data is fused through "field characteristics" to avoid misjudgment caused by the failure of a single sensor.

[0149] like Figure 5As shown, a second aspect of the present invention provides an energy storage device 100, including: a battery 1 and a plurality of sensors 4 disposed on the surface of the battery 1, and further including a battery abnormality determination device 5, which is used to execute the battery abnormality determination method of any of the above embodiments of the present invention.

[0150] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "illustrative embodiment," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of this application. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one embodiment or example.

[0151] Although embodiments of this application have been shown and described, those skilled in the art will understand that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of this application, the scope of which is defined by the claims and their equivalents.

Claims

1. A method for determining battery malfunctions, characterized in that, include: Acquire pressure information collected by multiple sensors deployed on the surface of the battery; Based on the pressure information, characterization parameters that characterize the degree of dispersion of battery pressure are determined; Based on the pressure values ​​corresponding to adjacent sensors, determine the local pressure gradient between adjacent sensors; Based on the characterization parameters and the local pressure gradient, it is determined whether the battery has local expansion anomalies.

2. The determination method according to claim 1, characterized in that, The step of determining the characterization parameters representing the degree of dispersion of battery pressure based on the pressure information includes: Discrete calculations are performed on the pressure values ​​corresponding to multiple sensors to obtain the characterization parameters; The characterization parameters include at least one of the following: the standard deviation of all pressure values, the variance of all pressure values, and the coefficient of variation.

3. The determination method according to claim 1, characterized in that, The step of determining whether the battery has local expansion anomalies based on the characterization parameters and the local pressure gradient includes: Within a preset number of detection cycles, when the growth rate of the characterization parameter exceeds a first threshold, or the maximum local pressure gradient reaches a second threshold, it is determined that the battery has a local expansion anomaly.

4. The determination method according to claim 1, characterized in that, Before the step of determining the characterization parameters representing the degree of dispersion of battery pressure based on the pressure information, the method further includes: Determine whether the battery is in a preset state of charge; The characterization parameters are determined when the battery is in the preset state of charge.

5. The determination method according to claim 4, characterized in that, The step of determining whether the battery is in a preset state of charge includes: Obtain the voltage signal of the battery; Calculate the pressure change rate corresponding to each sensor; Calculate the mean and standard deviation of the pressure change rate; The spatial synchronization index is determined based on the ratio of the standard deviation to the mean. Based on the voltage signal, the mean value, and the spatial synchronization index, it is determined whether the battery is in a preset state of charge.

6. The determination method according to claim 1, characterized in that, The determination method further includes: Based on the pressure information, the average value of the peak pressure of all sensors in each detection cycle is determined, and the global expansion coefficient is determined according to the rate of change of the average value with the detection cycle. Acquire temperature information collected by multiple sensors; Based on the pressure information and the temperature information, a temperature-pressure coupling coefficient is determined to characterize the degree of correlation between temperature changes and pressure changes. The presence of aging abnormalities in the battery is determined based on the global expansion coefficient, the temperature-pressure coupling coefficient, and the characterization parameters.

7. The determination method according to claim 6, characterized in that, The step of determining whether the battery has aging abnormalities based on the global expansion coefficient, the temperature-pressure coupling coefficient, and the characterization parameters includes: The global expansion coefficient, the temperature-pressure coupling coefficient, and the characterization parameters are input into a pre-trained prediction model to obtain the health status index and the predicted remaining lifespan. Based on the health status index and the remaining life prediction value, it is determined whether the battery is aging abnormally; The prediction model is trained based on historical sample data, which includes global expansion coefficient, temperature-pressure coupling coefficient, characterization parameters, health status index and remaining lifetime prediction values ​​of multiple historical batteries at different stages.

8. The determining method according to any one of claims 1 to 7, characterized in that, Also includes: If it is determined that the battery has a localized swelling abnormality, perform at least one of the following: Generate a charging current limit instruction, which is used to limit the charging current of the battery to below a preset safe current; Displays the location of the expansion region; Report to the remote monitoring platform.

9. The determining method according to any one of claims 1 to 7, characterized in that, The sensors are distributed in multiple areas on the surface of the battery, including the tab area, the center of the cell, and the edge area of ​​the cell.

10. An energy storage device, characterized in that, include: A battery and multiple sensors disposed on the surface of the battery; A battery malfunction determination apparatus, the battery malfunction determination apparatus being used to perform the battery malfunction determination method as described in any one of claims 1 to 9.

Images