A lithium battery fault early warning method, system, device and storage medium

Abstract

The present disclosure relates to the technical field of battery management, and particularly relates to a lithium battery fault early warning method, system, device and storage medium, wherein the method comprises the following steps: acquiring a time sequence relaxation curve of an impedance real part in a first preset frequency range of a lithium battery, and performing power function fitting to obtain a fitting goodness; acquiring time sequence data of an impedance imaginary part in a second preset frequency range of the lithium battery and the impedance real part in the first preset frequency range and a pole temperature, and determining a lithium battery internal temperature and a falling slope; determining a first joint fusion entropy increase value based on the lithium battery internal temperature, the falling slope and the pole temperature; collecting time sequence data of a voltage, a current, a load state and the impedance imaginary part in the first preset frequency range of the lithium battery, determining a change slope of the impedance imaginary part, and determining a second joint fusion entropy increase value; comparing the first joint fusion entropy increase value and the second joint fusion entropy increase value with a preset threshold to obtain a comparison result, and performing fault early warning according to the comparison result, effectively distinguishing fault types and implementing targeted processing.

Description

Technical Field

[0001] This disclosure relates to the field of battery management technology, and in particular to a lithium battery fault early warning method, system, device and storage medium. Background Technology

[0002] Lithium-ion batteries, with their advantages of high energy density, long cycle life, and high charge / discharge efficiency, are widely used in new energy vehicles, energy storage power stations, and portable electronic devices. However, during long-term operation, lithium-ion batteries are prone to faults such as micro-short circuits, overheating, and overcharging due to factors such as internal aging, external operating conditions, and improper use. If these faults are not addressed in time, they may lead to battery capacity decay, shortened lifespan, and in severe cases, thermal runaway, fire, explosion, and other safety accidents, threatening the safety of personnel and equipment.

[0003] Existing lithium battery fault early warning methods have obvious limitations. They are designed for only a certain type of fault and cannot cover the core safety risks of lithium batteries, which increases the complexity of the system. Alternatively, they use general indicators, which are not specific enough to capture the specific characteristics of different faults, making it easy to false alarms and missed alarms, and the early warning is delayed.

[0004] Therefore, existing technologies for lithium battery fault warnings cannot distinguish fault types, affecting handling efficiency. Summary of the Invention

[0005] To address the aforementioned technical problems, this disclosure provides a lithium battery fault early warning method, system, device, and storage medium.

[0006] This disclosure provides a lithium battery fault early warning method, including: Obtain the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and perform power function fitting on the time-series relaxation curve to obtain the goodness of fit; The timing data of the imaginary part of the impedance and the electrode temperature in the second preset frequency band of the lithium battery, as well as the timing data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery, are obtained. The internal temperature of the lithium battery is determined by the timing data of the imaginary part of the impedance, and the falling slope of the real part of the impedance is determined by the timing data of the real part of the impedance. The frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. The first joint fusion entropy increment is determined based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature. The voltage, current, load status, and timing data of the imaginary part of impedance in the first preset frequency band of the lithium battery are collected. Based on the timing data of the imaginary part of impedance, the slope of the change of the imaginary part of impedance is determined, and the second joint fusion entropy value is determined by the voltage, the current, the load status, and the slope of change. The goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment are compared with preset thresholds to obtain comparison results, and fault warnings are issued based on the comparison results.

[0007] Further, the step of obtaining the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and performing power function fitting on the time-series relaxation curve to obtain the goodness of fit, includes: The measured value of the real part of the impedance is obtained from the time-series relaxation curve of the real part of the impedance. The mean value of the measured values ​​is determined based on the measured values ​​of the real part of the impedance. The goodness of fit is determined based on the measured value of the real part of the impedance and the mean of the measured values.

[0008] Further, the step of acquiring the timing data of the imaginary part of the impedance and the terminal temperature in the second preset frequency band of the lithium battery, and the timing data of the real part of the impedance and the terminal temperature in the first preset frequency band of the lithium battery, and determining the internal temperature of the lithium battery through the timing data of the imaginary part of the impedance, and determining the decreasing slope of the real part of the impedance through the timing data of the real part of the impedance, wherein the frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery, includes: The internal temperature of the lithium battery is determined based on the timing data of the imaginary part of the impedance and the preset quadratic polynomial curve. The measured values ​​of the real part of the impedance at the first time point and the second time point are obtained respectively; Based on the measured value of the real part of the impedance at the first time point and the measured value of the real part of the impedance at the second time point, the difference between the measured values ​​at the first time point and the second time point is determined, and the time interval is determined based on the first time point and the second time point; The slope of the decrease in the real part of the impedance is determined based on the difference between the measured values ​​and the time interval.

[0009] Further, determining the first joint fusion entropy increment based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature includes: Based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature, the single-parameter permutation entropy of the internal temperature of the lithium battery, the single-parameter permutation entropy of the descent slope, and the single-parameter permutation entropy of the electrode temperature are determined respectively. The single-parameter arrangement entropy of the internal temperature of the lithium battery, the single-parameter arrangement entropy of the decreasing slope, and the single-parameter arrangement entropy of the electrode temperature are weighted and fused with a preset ratio to obtain the first joint fusion entropy value.

[0010] Further, determining the slope of the change of the imaginary part of the impedance based on the time-series data of the imaginary part of the impedance includes: The measured values ​​of the imaginary part of the impedance at the first and second time points are obtained respectively. Based on the measured value of the imaginary part of the impedance at the first time point and the measured value of the imaginary part of the impedance at the second time point, the difference between the measured values ​​at the first time point and the second time point is determined, and the time interval is determined based on the first time point and the second time point; The slope of the change in the imaginary part of the impedance is determined based on the difference between the measured values ​​and the time interval.

[0011] Further, determining the second joint fusion entropy increment using the voltage, the current, the load state, and the change slope includes: Based on the voltage, the current, the load state, and the slope of change, determine the single-parameter permutation entropy of the voltage, the current, the load state, and the slope of change, respectively; The single-parameter arrangement entropy of the voltage, the single-parameter arrangement entropy of the current, the single-parameter arrangement entropy of the load state, and the single-parameter arrangement entropy of the change slope are weighted and fused with a preset ratio to obtain the second joint fusion entropy value.

[0012] Further, the step of comparing the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment with preset thresholds respectively to obtain comparison results, and performing fault warning based on the comparison results, includes: When the goodness of fit is less than a first preset threshold, a first-level warning is triggered; When the increase in the first joint fusion entropy is greater than the second preset threshold range under the corresponding working condition, a second-level warning is triggered. When the increase in the second joint fusion entropy exceeds the third preset threshold range under the corresponding operating condition, a third-level warning is triggered.

[0013] This disclosure also provides a lithium battery fault early warning method, including: The fitting module is used to obtain the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and to perform power function fitting on the time-series relaxation curve to obtain the goodness of fit. The first determining module is used to acquire the timing data of the imaginary part of the impedance and the electrode temperature in the second preset frequency band of the lithium battery, as well as the timing data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery, and to determine the internal temperature of the lithium battery through the timing data of the imaginary part of the impedance and to determine the falling slope of the real part of the impedance through the timing data of the real part of the impedance, wherein the frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. The second determining module is used to determine the first joint fusion entropy increment based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature; The third determining module is used to collect the voltage, current, load status and time-series data of the imaginary part of the impedance in the first preset frequency band of the lithium battery, determine the slope of the change of the imaginary part of the impedance based on the time-series data of the imaginary part of the impedance, and determine the second joint fusion entropy value through the voltage, the current, the load status and the slope of change; The early warning module is used to compare the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment with preset thresholds respectively to obtain comparison results, and to issue fault warnings based on the comparison results.

[0014] This disclosure also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the lithium battery fault warning method.

[0015] This disclosure also provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of the lithium battery fault warning method.

[0016] The technical solution provided in this disclosure has the following advantages compared with the prior art: The goodness of fit is obtained by acquiring the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery and fitting the time-series relaxation curve with a power function. The time-series data of the imaginary part of the impedance in the second preset frequency band and the real part of the impedance in the first preset frequency band, along with the electrode temperature, are acquired. The internal temperature of the lithium battery is determined using the time-series data of the imaginary part of the impedance, and the descent slope of the real part of the impedance is determined using the time-series data of the real part of the impedance. Based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature, the first joint fusion entropy increment is determined. The time-series data of the voltage, current, load state, and the imaginary part of the impedance in the first preset frequency band of the lithium battery are collected. The descent slope of the imaginary part of the impedance is determined based on the time-series data of the imaginary part of the impedance, and the second joint fusion entropy increment is determined using the voltage, current, load state, and descent slope. The goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment are compared with preset thresholds to obtain comparison results. Fault warnings are then issued based on the comparison results to effectively distinguish fault types and implement targeted treatments. Attached Figure Description

[0017] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0018] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 A schematic diagram of a lithium battery fault early warning method provided in an embodiment of this disclosure; Figure 2 A schematic diagram illustrating a method for obtaining goodness of fit provided in an embodiment of this disclosure; Figure 3 This is a schematic diagram of a lithium battery fault early warning system provided in an embodiment of this disclosure. Detailed Implementation

[0020] To better understand the above-mentioned objectives, features, and advantages of this disclosure, the solutions disclosed herein will be further described below. It should be noted that, unless otherwise specified, the embodiments and features described herein can be combined with each other.

[0021] Numerous specific details are set forth in the following description in order to provide a full understanding of this disclosure, but this disclosure may also be implemented in other ways different from those described herein; obviously, the embodiments in the specification are only some, and not all, of the embodiments of this disclosure.

[0022] Figure 1 This is a schematic diagram of a lithium battery fault early warning method provided in an embodiment of this disclosure; as shown Figure 1 As shown, a lithium battery fault early warning method includes: Step S1: Obtain the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and perform power function fitting on the time-series relaxation curve to obtain the goodness of fit. In this embodiment, the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery is obtained. This time-series relaxation curve reflects the dynamic trajectory of the electrochemical process gradually releasing and recovering over time under specific low-frequency excitation. Since micro-short circuits typically do not immediately cause significant exceedances in voltage, current, or temperature rise, but rather alter the ion transport, polarization recovery, and interface reaction characteristics within the battery, the relaxation behavior of the real part of the impedance at low frequencies will first exhibit a trend deviating from the normal pattern. Fitting the time-series relaxation curve with a power function essentially matches the original curve to a decay or evolution model with a clear physical meaning. The goodness of fit is used to measure whether the curve still conforms to the typical relaxation law of a healthy battery. A high goodness of fit indicates that the curve is similar to the response mode of a healthy state; a significant decrease in goodness of fit signifies a structural distortion in the evolution of the real part of the impedance, often corresponding to problems such as early micro-short circuits, enhanced local side reactions, membrane damage, or the formation of micro-conductive channels within the battery. Instead of waiting for the fault to develop into obvious overheating, abnormal voltage difference, or sudden capacity drop before alarming, it uses the characteristic that low-frequency impedance is very sensitive to internal micro-deterioration to detect abnormalities in the early stage of faults, thus identifying them earlier, significantly improving the ability to move the early warning forward, and buying time for subsequent measures such as load reduction, re-inspection, or isolation.

[0023] For example, Figure 2 This is a schematic diagram of a method for obtaining goodness of fit provided in an embodiment of this disclosure; as shown below. Figure 2 As shown, step S1 involves obtaining the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and performing power function fitting on the time-series relaxation curve to obtain the goodness of fit. This includes: step S11, obtaining the measured value of the real part of the impedance based on the time-series relaxation curve of the real part of the impedance; step S12, determining the mean of the measured values ​​based on the measured values ​​of the real part of the impedance; and step S13, determining the goodness of fit based on the measured values ​​of the real part of the impedance and the mean of the measured values.

[0024] Specifically, the time-series relaxation curves of the real part of the impedance collected within the first preset frequency band of the lithium battery are transformed into a set of statistically and fittingly meaningful numerical sequences. The goodness-of-fit is then used to determine whether this sequence still conforms to the typical relaxation law under healthy conditions. The measured values ​​of the real part of the impedance originate from continuous sampling of the battery impedance response under excitation within a certain frequency band. These measured values ​​are not isolated but form curves with a physical evolution process over time. First, obtaining the measured values ​​of the real part of the impedance from the time-series relaxation curves is to discretize the continuous curves into calculable data points, providing a basis for subsequent fitting analysis. Then, determining the mean of all measured values ​​is to construct a reference center reflecting the overall level, used to measure the deviation between each data point and the overall mean. Finally, calculating the goodness-of-fit based on the measured values ​​and the mean of the measured values ​​evaluates the explanatory power of the power function model for the curve, i.e., whether the curve still retains the morphological characteristics that a normal relaxation process should have. When there are no micro-short circuits, localized damage to the separator, or abnormal side reactions inside the battery, the relaxation trend of the real part of its low-frequency impedance usually exhibits good continuity and regularity. Therefore, it matches the power function fitting well and has a high goodness of fit. However, when early micro-short circuits, localized leakage, or abnormal electrode contact begin to appear inside the battery, the evolution of the curve is disrupted, the measured values ​​fluctuate more around the mean, the deviation from the fitting model increases, and the goodness of fit decreases significantly. This method does not rely on whether the absolute impedance value exceeds the limit at a single moment, but rather starts from the structural changes in the entire relaxation process, making it more suitable for capturing very early anomalies that are highly concealed and develop slowly. On the other hand, the goodness of fit is a normalized evaluation index, which can reduce the impact of capacity differences, initial impedance differences, and environmental disturbances between different individual batteries, making the early warning judgment more stable and comparable. This significantly improves the early identification capability of micro-short circuit faults and provides a reliable first screening basis for subsequent graded early warning.

[0025] Step S2: Obtain the timing data of the imaginary part of the impedance and the electrode temperature in the second preset frequency band of the lithium battery, as well as the timing data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery. Determine the internal temperature of the lithium battery through the timing data of the imaginary part of the impedance and determine the decreasing slope of the real part of the impedance through the timing data of the real part of the impedance. The frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. In this embodiment, the time-series data of the imaginary part of impedance in the second preset frequency band, the real part of impedance in the first preset frequency band, and the electrode temperature are obtained respectively. The high-frequency imaginary part of impedance mainly reflects the battery's internal ohmic resistance, interface charge transfer characteristics, and the impact of temperature changes on the electrochemical process. Therefore, the battery's internal temperature can be deduced using a pre-established calibration relationship. Compared to surface temperature, internal temperature more accurately reflects the starting point of lithium battery thermal runaway risk because many overheating and over-temperature faults initially manifest as internal heat accumulation, while electrode or casing temperatures often exhibit lag. Simultaneously, the time-series data of the real part of impedance in the first preset frequency band reflects trends such as increased battery polarization, changes in mass transfer resistance, and accelerated internal reactions. Therefore, the slope of the impedance real part's descent can be obtained through time-series difference analysis to describe the rate of abnormal evolution. The electrode temperature provides a directly observable external reference for thermal anomalies. Integrating internal temperature, impedance descent rate, and external temperature into a unified observation system avoids misjudgments or delayed reporting caused by relying solely on a single temperature threshold. It can not only detect batteries that have already heated up significantly, but also identify potential signs of thermal runaway, such as "internal heating first, external heating later," making it particularly suitable for scenarios with high-rate charging and discharging, large ambient temperature fluctuations, or complex heat dissipation conditions. It improves sensitivity, stability, and lead time for overheating and over-temperature detection.

[0026] For example, step S2 involves acquiring the timing data of the imaginary part of the impedance and the electrode temperature in the second preset frequency band of the lithium battery, as well as the timing data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery. The internal temperature of the lithium battery is determined using the timing data of the imaginary part of the impedance, and the decreasing slope of the real part of the impedance is determined using the timing data of the real part of the impedance. The frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. This includes: determining the internal temperature of the lithium battery based on the timing data of the imaginary part of the impedance and a preset quadratic polynomial curve; acquiring the measured values ​​of the real part of the impedance at a first time and a second time, respectively; determining the difference between the measured values ​​at the first time and the second time, and determining the time interval based on the first time and the second time; and determining the decreasing slope of the real part of the impedance based on the difference in the measured values ​​and the time interval.

[0027] Specifically, thermal anomaly identification is broken down into two mutually corroborating dimensions: internal thermal state and impedance evolution rate, thereby improving the accuracy of identifying early risks of overheating. A mapping relationship is established between the imaginary part of impedance in the second preset frequency band and a preset quadratic polynomial curve. The internal temperature of the lithium battery is deduced from the imaginary part of impedance because the high-frequency imaginary part of the battery is extremely sensitive to temperature, and the internal temperature often reflects the degree of heat accumulation of the electrochemical reaction earlier and more accurately than the external electrode temperature. Calculation using the preset quadratic polynomial curve avoids the hysteresis and locality problems caused by directly relying on temperature sensors, allowing the system to capture the internal temperature rise trend before the temperature becomes apparent. Subsequently, the measured values ​​of the real part of impedance at the first and second time points are obtained, and the difference between them is divided by the time interval to determine the slope of the decrease in the real part of impedance, the purpose of which is to quantify the speed and direction of impedance change. For healthy batteries, the real part of impedance typically exhibits a relatively stable or slow change under normal operating conditions. However, when the battery begins to experience increased polarization, changes in the interfacial film, increased ion transport resistance, or localized heat accumulation, the rate of change of the real part of impedance becomes significantly abnormal, and the slope of its descent deviates from the normal range. Combining internal temperature with the slope of the real part of impedance's descent can form a "thermal-impedance coupling" discrimination logic: an increase in internal temperature indicates the presence of an abnormal heat source, while a faster change in the real part of impedance indicates that the stability of the electrochemical system has been disturbed. When both occur simultaneously, the credibility of overheating risk is significantly increased. This approach can not only detect batteries already in a state of rising temperature but also identify batteries whose internal thermal state has begun to deteriorate even before their temperature has significantly exceeded the limit, thus achieving hourly or even earlier early warnings. Furthermore, the method of calculating the slope using time intervals and differences is simple, real-time, and suitable for online deployment. It is easy to implement in monitoring terminals and can also be combined with temperature mapping models to form a lightweight and highly sensitive overheating precursor identification scheme.

[0028] Step S3: Determine the first joint fusion entropy increment based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature; In this embodiment, temperature data alone is insufficient to fully describe whether the battery has entered a dangerous state, because the absolute value of temperature itself may fluctuate under different operating conditions. What truly reflects the abrupt change in system state is often the change in the complexity of the coupling relationship between multiple parameters. Using the internal temperature of the lithium battery, the slope of the real part of impedance decrease, and the terminal temperature as inputs, preprocessing is performed first, including outlier removal, noise suppression, time synchronization, and missing value completion, to ensure that the multi-source data are comparable in terms of time axis and statistical distribution. Then, the fusion entropy increment is calculated based on the changes in the arrangement pattern of these parameters. The joint fusion entropy maps the degree of disorder in the arrangement structure of multiple parameters within a certain time window to a unified numerical index. When the battery is in a healthy or stable operating condition, the changes in these parameters are usually relatively regular, with low entropy values ​​and small fluctuations. However, when the battery experiences overheating, overheating, or local thermal runaway precursors, internal thermal diffusion, electrochemical reactions, and impedance changes will significantly accelerate and couple with each other, leading to a decrease in the orderliness of the parameter sequence and an increase in disorder, ultimately manifested as an increase in entropy increment. It can compress the originally scattered temperature, impedance, and thermal boundary information into a comprehensive index with strong discriminative power, thereby reducing the false alarms and false negatives caused by the single-parameter threshold method. Especially under different magnification, different ambient temperatures, and different aging levels, single parameters are easily affected by individual differences, while fusion entropy emphasizes the trend of state change, thus having better robustness and generalization, and can more accurately identify early anomalies that are "evolving towards overheating".

[0029] For example, step S3, determining the first joint fusion entropy increment based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature, includes: determining the single-parameter permutation entropy of the internal temperature of the lithium battery, the single-parameter permutation entropy of the descent slope, and the single-parameter permutation entropy of the electrode temperature based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature; and weighting and fusing the single-parameter permutation entropy of the internal temperature of the lithium battery, the single-parameter permutation entropy of the descent slope, and the single-parameter permutation entropy of the electrode temperature with a preset weight ratio to obtain the first joint fusion entropy increment.

[0030] Specifically, while internal temperature, slope of impedance drop, and terminal temperature are all related to overheating risk, they reflect different emphases: internal temperature reflects the degree of heat accumulation inside the cell, slope of impedance drop reflects the speed and trend of impedance change, and terminal temperature reflects the externally observable thermal response. Looking at only one parameter is easily affected by environmental fluctuations, sampling errors, and poor local contact, leading to false alarms or missed alarms. However, calculating the permutation entropy of each parameter separately is equivalent to measuring the complexity of each parameter's fluctuations over a certain period from the perspective of "sequence order." Under normal conditions, temperature and impedance changes are relatively stable, with fewer permutation patterns and lower entropy values; however, when overheating begins, parameter fluctuations become faster and more chaotic, permutation patterns increase, and entropy values ​​rise. Further weighted fusion according to preset proportions takes into account that different parameters are not entirely sensitive to faults. For example, internal temperature is usually closer to the actual heat source than pole temperature, so it can be given a higher weight; the descent slope reflects the rate of state change and should also have a high weight; pole temperature serves as an external confirmation signal, and its weight can be slightly lower but cannot be omitted. The first joint fusion entropy increment obtained in this way not only reflects the common anomalies among multiple parameters, but also compresses the random fluctuations of single parameters, highlighting the true trend risks. It is more robust to thermal anomalies under different operating conditions: when charging and discharging at high rates, idling at low rates, or when the ambient temperature changes significantly, a single temperature threshold is often difficult to uniformly adapt, while fusion entropy can identify the process of thermal imbalance forming by the overall change in sequence complexity, thus making it more suitable for early warning.

[0031] Step S4: Collect the timing data of the voltage, current, load status and the imaginary part of the impedance in the first preset frequency band of the lithium battery. Based on the timing data of the imaginary part of the impedance, determine the slope of the change of the imaginary part of the impedance. And determine the second joint fusion entropy value-added through the voltage, current, load status and the slope of change. In this embodiment, overcharge faults typically occur during normal charging. The danger lies in the fact that changes in voltage, current, and state of charge (SOC) are coupled with processes such as internal battery polarization, lithium deposition, intensified side reactions, and interface film damage. Therefore, simply relying on voltage exceeding limits often fails to detect the true early-stage anomalies in a timely manner. First, time-series data of battery voltage, current, load state, and the imaginary part of impedance in a first preset frequency band are collected. Voltage and current reflect external charging conditions, load state reflects the current operating condition of the battery, and the imaginary part of impedance reflects the battery's internal polarization, capacitive reactance, and dynamic response characteristics. By differentially processing the time-series data of the imaginary part of impedance, the slope of the imaginary part of impedance change can be obtained. The slope can be understood as a sensitive indicator of the speed of internal state evolution during battery charging. When the battery enters the early stage of overcharge, phenomena such as abnormal voltage rise at the end of charging, accelerated polarization changes, changes in the evolution trend of the imaginary part of impedance, and inconsistency between SOC growth and voltage response often occur. These characteristics may not seem problematic individually, but when combined, they exhibit statistical structural anomalies. Therefore, by jointly calculating the second joint fusion entropy value based on voltage, current, load status, and the slope of the imaginary part of impedance, anomalies can be identified simultaneously from two levels: whether the charging behavior is normal and whether the internal response is normal. Upgrading overcharge warning from traditional single-threshold voltage protection to multi-dimensional early detection not only enables earlier detection of issues such as charging strategy mismatch, charging termination condition failure, and SOC estimation deviation, but also provides timely warnings before significant thermal runaway or irreversible damage occurs, improving charging safety boundaries and reducing the risks of expansion, lithium plating, internal short circuits, and thermal runaway caused by overcharging.

[0032] For example, step S4, determining the slope of the change of the imaginary part of impedance based on the time-series data of the imaginary part of impedance, includes: obtaining the measured values ​​of the imaginary part of impedance at a first time and a second time respectively; determining the difference between the measured values ​​at the first time and the second time based on the measured values ​​of the imaginary part of impedance at the first time and the second time, and determining the time interval based on the first time and the second time; and determining the slope of the change of the imaginary part of impedance based on the difference between the measured values ​​and the time interval.

[0033] Specifically, the dynamic changes in the imaginary part of low-frequency impedance are used to capture the abnormal evolution of internal polarization and side reactions during charging. The imaginary part of impedance is sensitive to the internal diffusion process, interface capacitance characteristics, and charge accumulation state of the battery during charging. When the battery is in the normal charging range, the change in the imaginary part usually shows a relatively stable regularity along with the increase in SOC. However, once it enters the early stage of overcharging, phenomena such as increased lithium deposition risk, enhanced polarization, changes in mass transfer resistance, and accelerated local side reactions occur inside the battery. These changes are first reflected in the sequence of the imaginary part of impedance. By obtaining the measured values ​​of the imaginary part of impedance at the first and second moments, calculating the difference between the two, and then dividing by the time interval, the slope of change is obtained, which is used to explicitly quantify the rate of change. Compared to directly looking at the impedance value at a certain moment, the slope is better able to reflect whether the trend is abnormal, because overcharging often does not happen instantaneously, but rather first shows a dulling of the charging end response, a shift in the inflection point of the imaginary part, or an abnormal rate of change, and then gradually evolves into a continuous increase in voltage, accelerated temperature rise, and even the risk of thermal runaway. Once the slope of change deviates from the dynamic range under normal operating conditions, it means that the electrochemical reaction inside the battery is no longer under control, and a mismatch occurs between the charging process and the impedance response. This system can predict overcharge precursors by detecting dynamic anomalies in the imaginary part of the impedance before the voltage significantly exceeds the limit, thus enabling proactive control actions such as current limiting, stopping charging, or switching strategies to avoid responding only after the terminal voltage reaches a dangerous threshold. Furthermore, this slope calculation method is simple to operate, highly real-time, and easy to implement online, making it suitable for embedding in charging pile monitoring systems. Simultaneously, it does not rely on simple voltage protection logic, therefore it has better fault tolerance for SOC estimation errors, sensor drift, and short-term voltage fluctuations.

[0034] In some possible implementations, step S4, determining the second joint fusion entropy increment based on voltage, current, load state, and change slope, includes: determining the single-parameter permutation entropy of voltage, current, load state, and change slope respectively based on voltage, current, load state, and change slope; and weighting and fusing the single-parameter permutation entropy of voltage, current, load state, and change slope with a preset weight ratio to obtain the second joint fusion entropy increment.

[0035] In this embodiment, external charging behavior parameters and internal impedance evolution parameters are unified into an entropy increase evaluation framework, thereby avoiding the limitations caused by making judgments based solely on voltage or current. Voltage, current, and load status reflect the external input and operating conditions of the charging process, while the change slope reflects the response rate of the imaginary part of the internal impedance. The combination of these four parameters actually constitutes a dual observation system of "external drive - internal response". The single-parameter permutation entropy of the four parameters is calculated separately to measure the orderly change of each parameter over a period of time: Under normal charging conditions, voltage, current, and load status usually change smoothly around a predetermined strategy, and their permutation pattern is relatively stable with low entropy values; however, when entering the pre-overcharge stage, charging behavior may exhibit phenomena such as continuous application, failure of termination conditions, abnormal voltage rise at the end of charging, or decoupling between SOC and voltage response. The sequence structure of external parameters becomes more complex, and the entropy value increases; at the same time, the change slope of the imaginary part of impedance will also become abnormal due to internal polarization imbalance, further increasing the total entropy after fusion. The weighted fusion of the entropy of four single parameters according to a preset ratio is designed to consider the contribution of different parameters in overcharge identification. Voltage and current are usually the most direct overcharge signals, load status can supplement operating condition information, and the slope of change reflects the internal state evolution. Therefore, all four factors jointly determine the value of the second joint fusion entropy. Overcharge is no longer simply understood as excessive voltage, but rather a comprehensive judgment is made from multiple dimensions, including whether the charging process is abnormal, whether the internal impedance response is abnormal, and whether the operating conditions are matched. This significantly reduces false alarms and delayed alarms caused by relying solely on voltage thresholds. Especially in complex scenarios such as fast charging, constant current / constant voltage switching, SOC estimation deviation, and inconsistent charging strategies, single electrical parameters are easily distorted. Fusion entropy can identify potential risks through the joint complexity changes of multi-source timing, thus possessing stronger adaptability and generalization capabilities. It can identify abnormal charging states that have not yet exceeded limits, giving the system the opportunity to intervene before overcharging truly develops into a dangerous event, significantly improving charging process safety and battery life protection.

[0036] Step S5: Compare the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment with preset thresholds to obtain comparison results, and issue fault warnings based on the comparison results.

[0037] In this embodiment, a hierarchical early warning logic is established by comparing the results. The goodness-of-fit is primarily used to identify very early structural anomalies, corresponding to micro-short circuit risks. The first joint fusion entropy increment is primarily used to identify the accelerated evolution of thermal anomalies, corresponding to overheating risks. The second joint fusion entropy increment is primarily used to identify early deviations during the charging process, corresponding to overcharging risks. These three indicators correspond to different fault types and different time scales, avoiding mixing all anomalies in the same threshold system, which would lead to inaccurate responses. In this way, different levels of early warning signals can be output based on the comparison results. For example, a level one early warning can be used to prompt increased inspection or sampling frequency; a level two early warning can be used to trigger current limiting, load reduction, or heat dissipation intervention; and a level three early warning can be used to immediately stop charging, disconnect the circuit, or execute interlocking protection. This achieves closed-loop control from signal detection to risk management, enabling the entire method not only to identify faults but also to guide protective actions in actual operation, enhancing the system's engineering feasibility, real-time performance, and safety redundancy.

[0038] For example, in step S5, the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment are compared with preset thresholds to obtain comparison results, and fault warnings are issued based on the comparison results, including: when the goodness of fit is less than the first preset threshold, a first-level warning is triggered; when the first joint fusion entropy increment is greater than the second preset threshold range under the corresponding operating condition, a second-level warning is triggered; when the second joint fusion entropy increment is greater than the third preset threshold range under the corresponding operating condition, a third-level warning is triggered.

[0039] Specifically, comparing the goodness of fit with the first preset threshold is to identify the earliest and most concealed micro-short circuit risks. This is because micro-short circuits often first manifest as distortion in the shape of the low-frequency impedance real part relaxation curve, and the goodness of fit is a direct indicator of the degree of this distortion. When the goodness of fit is below the threshold, it indicates that the curve has significantly deviated from the power function evolution law under healthy conditions, and the first-level warning should be triggered immediately to lock the fault in its nascent stage. Secondly, comparing the first joint fusion entropy increment with the second preset threshold range is used to identify over-temperature and overheat risks. Here, an interval threshold is used instead of a single-point threshold because thermal anomalies vary under different operating conditions. Only when they exceed the normal fluctuation range under the corresponding operating condition can the anomaly be considered statistically significant. If the fusion entropy increment significantly exceeds the preset range, it indicates that the coupling complexity between internal temperature, impedance drop slope, and pole temperature has significantly increased, and the thermal anomaly is evolving at an accelerated pace, thus triggering the second-level warning. Finally, comparing the second joint fusion entropy increment with the third preset threshold range is used to determine whether the overcharge risk has entered the early abnormal state. Since charging behavior is influenced by the charging rate, load status, SOC level, and internal response, a condition-related threshold range is used to improve the accuracy of the judgment. When the fused entropy value exceeds the threshold range, a third-level warning is triggered, prompting an immediate halt to charging or a protection switch. Different faults are stratified according to time scale and degree of danger, preventing the lumping of all anomalies and improving the targeting of the warnings. Combining statistical criteria with condition thresholds allows the system to adapt to different scenarios such as high charging rates, low charging rates, and inactivity, avoiding the insufficient adaptability caused by fixed thresholds. A progressive monitoring logic is formed, from "structural anomaly—thermal anomaly—charging anomaly," enabling the system to detect, divert, and intervene early. Instead of simply outputting an "abnormal / normal" judgment, it outputs hierarchical and action-oriented warning results that directly serve subsequent control actions such as current limiting, load reduction, disconnection, and alarm interlocking, thus truly integrating risk identification into engineering safety management.

[0040] In some possible implementation methods, a graded early warning mechanism is established, targeting three core faults in lithium batteries: micro-short circuit, overheating, and overcharging, achieving early warning effects at the daily, hourly, and minute levels. The three levels of warnings are detected independently and linked sequentially to achieve early fault identification and tiered handling. The specific steps are as follows: Level 1: Micro-short circuit very early fault warning: The time-series relaxation curve of the real part of the low-frequency impedance at a single point in the 0.1~2Hz frequency band of the lithium battery is collected by the impedance detection module. The relaxation curve is fitted with a power function using the least squares method, and the goodness-of-fit R² value is calculated. When R² < 0.9, it is determined that an early micro-short circuit has occurred in the battery, triggering the first-level warning. The second-level overheating early warning system synchronously collects the imaginary part of the high-frequency impedance at a single point in the 100~1kHz frequency band, the real part of the low-frequency impedance in the 0.01~2Hz frequency band, and the electrode temperature. Based on the preset quadratic polynomial calibration curve, the internal temperature of the battery is calculated through the imaginary part of the high-frequency impedance. The slope of the decrease of the real part of the low-frequency impedance is solved by the sliding window first-order difference method. After preprocessing the three parameters of internal temperature, slope of decrease of the real part of the low-frequency impedance, and electrode temperature by outlier removal, noise filtering, time synchronization, and missing value completion, the entropy increase of the three parameters is calculated by joint fusion. When the entropy increase exceeds the preset overheating threshold range under the corresponding operating condition, the second-level warning is triggered. Level 3 overcharge fault early warning: Simultaneously collect lithium battery voltage, current, SOC value and low-frequency impedance imaginary part time series data, use sliding window first-order difference method to solve the slope of low-frequency impedance imaginary part change, after the four parameters of voltage, current, SOC and low-frequency impedance imaginary part change slope are processed by outlier removal, noise filtering, time synchronization and missing value completion preprocessing process, calculate the joint fusion entropy increase of the four parameters, when the entropy increase exceeds the preset overcharge threshold range under normal charging conditions, trigger the level 3 warning; The sampling frequency band for the real part of the low-frequency impedance is 0.1~2Hz, and the relaxation curve is a time-series curve showing the change of the real part of the low-frequency impedance over time; the formula for the power function fitting is: ; Where R1(t) is the real part of the low-frequency impedance at time t, k is the fitting coefficient, n is the power exponent, and b is a constant term; the goodness of fit R² is calculated by the following formula: ; Among them, y i This is the measured value of the real part of the low-frequency impedance. i To fit the predicted values, This represents the average of the measured values.

[0041] The sampling frequency band for the imaginary part of the high-frequency impedance is 100~1kHz; the calibration curve for calculating the internal temperature based on the imaginary part of the high-frequency impedance is a quadratic polynomial curve, and the formula is: ; in, This refers to the internal temperature of the battery. denoted as the imaginary part of the high-frequency impedance, and a, b, and c are calibration coefficients obtained from single-cell calibration tests.

[0042] Low-frequency impedance real part decreasing slope The calculation formula is: ; in, Let be the real part of the low-frequency impedance at time t. for The real part of the low-frequency impedance at time t is given, where Δt is the calculation time interval; the information entropy increase is the difference between the joint fusion entropy of the three parameters and the normal baseline entropy value, i.e. ,in For the joint fusion entropy of three parameters, This is the baseline entropy value under normal operating conditions.

[0043] The calculation process of the three-parameter joint fusion entropy is as follows: The single-parameter permutation entropy of internal temperature, the descent slope of the real part of low-frequency impedance, and pole temperature are calculated separately using the permutation entropy algorithm, and then weighted and fused according to a weight ratio of 0.4:0.3:0.3. The formula is as follows: ; in, The entropy is the internal temperature arrangement. The real part of the low-frequency impedance has a decreasing slope, which represents the arrangement entropy. The pole temperature arrangement entropy; the single-parameter arrangement entropy PE is calculated by the following formula: ; Where k is the total number of permutation patterns of the parameter time series embedding matrix, p i Let be the probability of the i-th permutation pattern, satisfying Σ1 k p i = 1.

[0044] Low-frequency impedance imaginary part slope The calculation formula is: ; in, Let be the imaginary part of the low-frequency impedance at time t. The value is the imaginary part of the low-frequency impedance at time t-Δt, where Δt is the calculation time interval.

[0045] Information entropy increase is the difference between the joint fusion entropy of the four parameters and the normal baseline entropy value, i.e. ,in The joint fusion entropy of four parameters, The entropy baseline under normal charging conditions is defined as follows: The formula for calculating the joint fusion entropy of the four parameters is: ; in, For voltage arrangement entropy, The entropy of the current arrangement. The entropy of the arrangement of charged states; The slope of the imaginary part of the low-frequency impedance change is the arrangement entropy. The single-parameter arrangement entropy PE is calculated using the following formula: ; Where k is the total number of permutation patterns of the parameter time series embedding matrix, pi Let be the probability of the i-th permutation pattern, satisfying Σ1 k p i = 1.

[0046] Parameter preprocessing includes: removing outliers using the 3σ rule, and suppressing noise through median filtering and moving average filtering.

[0047] Entropy threshold ranges are established based on operating conditions, categorized into high-rate, low-rate, and static operating conditions. For each condition, the mean μ and standard deviation σ are calculated based on healthy battery data. The over-temperature entropy warning range is defined as follows: The overcharge entropy warning range is If the value exceeds the specified range, a corresponding warning will be triggered; the baseline entropy value of the healthy battery... Calculated using the following formula: ; Where K is the number of test cycles (≥100) for the corresponding operating conditions of a healthy battery. K' represents the joint fusion entropy of the three parameters in the k-th cycle; K' represents the number of test cycles (≥100) under normal charging conditions of a healthy battery. The four-parameter joint fusion entropy is the value of the k-th charging cycle.

[0048] For example, the first-level micro-short circuit early warning system captures early impedance fluctuation anomalies by measuring the goodness of fit of the low-frequency impedance real part relaxation curve, thus achieving accurate identification of micro-short circuits. The specific process is as follows: Data acquisition: Using an AC impedance meter or an integrated impedance detection module of a BMS (Battery Management System), low-frequency impedance real part data of the lithium battery is acquired. The acquisition frequency band is set to 0.1~2Hz (this frequency band is most sensitive to micro-short circuits inside the battery and can reflect changes in the relaxation characteristics of the electrode interface). The acquisition frequency is 1~2Hz, and the time-series curve of the low-frequency impedance real part changing with time is obtained, i.e., the relaxation curve.

[0049] Power function fitting: Perform power function fitting on the collected relaxation curves. The fitting formula is as follows: ; Where R1(t) is the real part of the low-frequency impedance at time t, k is the fitting coefficient, n is the power exponent (the value of n for a healthy battery is 0.1~0.3), and b is a constant term; the least squares method is used in the fitting process to ensure that the fitting result fits the measured curve.

[0050] Goodness-of-fit calculation: Calculate the R² value of the power function fit, which measures the degree of fit between the fitted curve and the measured relaxation curve. The formula is: ; Among them, y i This is the measured value of the real part of the low-frequency impedance. i To fit the predicted values, R² is the mean of all measured values; the range of R² is [0,1]. The closer R² is to 1, the stronger the regularity of the relaxation curve and the healthier the battery is.

[0051] Warning judgment: When R² < 0.9, it indicates that the regularity of the relaxation curve of the real part of the low-frequency impedance has decreased significantly, and a micro-short circuit has appeared inside the battery (the micro-short circuit causes disordered relaxation characteristics of the electrode interface, disordered impedance fluctuation, and reduced goodness of fit), triggering the first-level micro-short circuit warning.

[0052] The second-level overheating early warning system integrates three parameters: internal temperature calculated from the imaginary part of high-frequency impedance, the descent slope of the real part of low-frequency impedance, and pole temperature. It achieves early warning of overheating by incrementally increasing the disorder of parameters through information entropy. The specific process is as follows: Data acquisition and core parameter calculation: Acquire high-frequency impedance imaginary part (frequency band 100~1kHz, sensitive to temperature but not sensitive to current state and charge state, can directly reflect internal thermal state) and low-frequency impedance real part time series data, acquisition frequency 1~2Hz; use thermocouple sensor to acquire pole temperature, acquisition frequency synchronized with impedance data.

[0053] Internal temperature calculation: Based on the calibration curve of the imaginary part of high-frequency impedance and internal temperature, the internal temperature of the battery is calculated. The calibration curve is obtained through single-cell calibration tests and is a quadratic polynomial curve, with the following formula: ; in, denoted as the imaginary part of the high-frequency impedance, and a, b, and c are calibration coefficients.

[0054] Calculation of the slope of the real part of the low-frequency impedance: The slope of the real part of the low-frequency impedance is solved using the sliding window first-order difference method. The formula is: ; Where Δt is the calculation time interval (0.5~1s, adapted to a sampling frequency of 1~2Hz), and R1(t) and R1(t-Δt) are the real parts of the low-frequency impedance at time t and t-Δt, respectively; increased temperature leads to increased ionic conductivity and decreased real part of the low-frequency impedance. The greater the absolute value of the downward slope, the more intense the heat generation.

[0055] Multi-parameter preprocessing: internal temperature Low-frequency impedance real part decreasing slope Column temperature Unified preprocessing of three-parameter time series data: Outlier removal: The 3σ principle is used to remove outliers that exceed the range of [μ-3σ, μ+3σ] (where μ is the mean of the parameters and σ is the standard deviation). Noise filtering: A combination of "median filtering + moving average filtering" is used, with a filtering window size of 10 to 20 data points, to suppress impulse noise and random noise; Time synchronization and missing value completion: The three-parameter data are aligned based on the collection timestamp, and linear interpolation is used to complete a small number of missing data to ensure sequence continuity.

[0056] Information entropy increase calculation: Single-parameter permutation entropy calculation: The permutation entropy algorithm (strong noise resistance, no discretization required) is used to calculate the single-parameter permutation entropy of the three parameters respectively; the permutation entropy parameters are set as follows: embedding dimension m=2~3, time delay τ=1, sliding window size 50~100 data points, step size 1~5 data points, to obtain (Internal temperature arrangement entropy) (Slope arrangement entropy) (Temperature arrangement entropy of poles).

[0057] Joint fusion entropy calculation: assigning weights to the sensitivity of over-temperature faults based on three parameters ( Weight 0.4, Weight 0.3, With a weight of 0.3, the weighted fusion yields the joint fusion entropy. The formula is: ; Entropy increase calculation: ,in For the joint fusion entropy of three parameters, The entropy increase is the baseline entropy value under normal operating conditions (the mean μ is obtained through statistical analysis of healthy battery data).

[0058] Warning Judgment: An entropy threshold range is established based on operating conditions (high magnification ≥1C, low magnification 0.2C~1C, static). The over-temperature warning range is... The overheat alarm range is (σ is the standard deviation of the entropy value of a healthy battery); when When the temperature falls within the warning range, a Level II overheat warning is triggered.

[0059] The third-level overcharge fault early warning integrates four parameters: voltage, current, SOC, and the slope of the imaginary part change of low-frequency impedance. It captures the parameter coupling disorder caused by overcharging through information entropy increase, thus achieving overcharge early warning. The specific process is as follows: Data acquisition and core parameter calculation: Collect lithium battery voltage U, current I, and SOC value at a frequency of 1~2Hz; synchronously collect single-point low-frequency impedance imaginary part Im(Z_low) (frequency band 0.1~2Hz) timing data at the same frequency as the above parameters.

[0060] Calculation of the slope of the imaginary part of low-frequency impedance: The sliding window first-order difference method is used to solve the problem. The formula is: ; Where Δt is the calculation time interval (5~10s), when overcharging intensifies the internal polarization of the battery and causes abnormal fluctuations in the slope of the imaginary part of the low-frequency impedance.

[0061] Multi-parameter preprocessing: preprocessing voltage U, current I, state of charge (SOC), and slope. Four-parameter time series data are processed through a preprocessing workflow (outlier removal, noise filtering, time synchronization, and missing value completion) to obtain clean time series data.

[0062] Information entropy increase calculation: Single-parameter permutation entropy calculation: Using the same permutation entropy parameters as S2, calculate the permutation entropy of all four parameters separately, and obtain... (Voltage arrangement entropy) (Current arrangement entropy) (SOC permutation entropy) (Slope arrangement entropy).

[0063] Joint fusion entropy calculation: Weights are assigned based on the sensitivity of parameters to overcharge faults (voltage 0.3, current 0.2, SOC 0.3, ...). 0.2), weighted fusion yields The formula is: ; The calculation logic of the single-parameter permutation entropy PE is consistent with that of the second-level overheating early fault warning, both using PE = -Σ1 k p i · ln p i Formula; The weight allocation of the four-parameter joint fusion entropy is based on the sensitivity coefficient of the parameters to overcharge faults. The sensitivity coefficient is calibrated through controlled variable experiments to ensure the rationality of the weight allocation.

[0064] Early warning judgment: Establish overcharge entropy threshold range and healthy charging baseline. (K'≥100, the number of charge cycles for a healthy battery); standard deviation Entropy increase calculation: ',in The four-parameter joint fusion entropy. The warning interval is... The severely overcharged range is > ;when When the charge falls into the warning range, a Level 3 overcharge warning is triggered.

[0065] Figure 3 This is a schematic diagram of a lithium battery fault early warning system provided in an embodiment of this disclosure; as shown. Figure 3 As shown, this disclosure also provides a lithium battery fault early warning system, including: a fitting module 401, used to acquire the time-series relaxation curve of the real part of the impedance in a first preset frequency band of the lithium battery, and to perform power function fitting on the time-series relaxation curve to obtain the goodness of fit; a first determination module 402, used to acquire the time-series data of the imaginary part of the impedance and the electrode temperature in a second preset frequency band of the lithium battery, and the time-series data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery, and to determine the internal temperature of the lithium battery through the time-series data of the imaginary part of the impedance and to determine the decreasing slope of the real part of the impedance through the time-series data of the real part of the impedance, wherein the frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. The system comprises: a preset frequency band; a second determining module 403, used to determine the first joint fusion entropy increment based on the internal temperature, descent slope, and electrode temperature of the lithium battery; a third determining module 404, used to collect the voltage, current, load state, and time-series data of the imaginary part of impedance in the first preset frequency band of the lithium battery, determine the change slope of the imaginary part of impedance based on the time-series data of the imaginary part of impedance, and determine the second joint fusion entropy increment through voltage, current, load state, and change slope; and an early warning module 405, used to compare the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment with preset thresholds respectively, obtain comparison results, and issue fault warnings based on the comparison results.

[0066] In some possible implementations, the fitting module 401 is specifically used to: obtain the measured value of the real part of the impedance based on the time-series relaxation curve of the real part of the impedance; determine the mean of the measured values ​​based on the measured values ​​of the real part of the impedance; and determine the goodness of fit based on the measured values ​​of the real part of the impedance and the mean of the measured values.

[0067] In some possible implementations, the first determining module 402 is specifically used to: determine the internal temperature of the lithium battery based on the timing data of the imaginary part of the impedance and a preset quadratic polynomial curve; obtain the measured values ​​of the real part of the impedance at a first time and a second time, respectively; determine the difference between the measured values ​​at the first time and the second time based on the measured values ​​of the real part of the impedance at the first time and the second time, and determine the time interval based on the first time and the second time; and determine the decreasing slope of the real part of the impedance based on the difference in the measured values ​​and the time interval.

[0068] In some possible implementations, the second determining module 403 is specifically used to: determine the single-parameter permutation entropy of the internal temperature of the lithium battery, the single-parameter permutation entropy of the downward slope, and the single-parameter permutation entropy of the electrode temperature based on the internal temperature of the lithium battery, the downward slope, and the electrode temperature; and perform weighted fusion of the single-parameter permutation entropy of the internal temperature of the lithium battery, the single-parameter permutation entropy of the downward slope, and the single-parameter permutation entropy of the electrode temperature with a preset weight ratio to obtain the first joint fusion entropy increment.

[0069] In some possible implementations, the third determining module 404 is specifically used to: obtain the measured values ​​of the imaginary part of the impedance at the first time and the second time respectively; determine the difference between the measured values ​​at the first time and the second time based on the measured values ​​of the imaginary part of the impedance at the first time and the second time, and determine the time interval based on the first time and the second time; and determine the slope of the change of the imaginary part of the impedance based on the difference of the measured values ​​and the time interval.

[0070] In some possible implementations, the third determining module 404 is specifically used to: determine the single-parameter permutation entropy of voltage, current, load state, and change slope based on voltage, current, load state, and change slope respectively; and weight and fuse the single-parameter permutation entropy of voltage, current, load state, and change slope with a preset weight ratio to obtain the second joint fusion entropy value-added.

[0071] In some possible implementations, the early warning module 405 is specifically used to: trigger a first-level early warning when the goodness of fit is less than a first preset threshold; trigger a second-level early warning when the first joint fusion entropy increment is greater than a second preset threshold range under the corresponding operating condition; and trigger a third-level early warning when the second joint fusion entropy increment is greater than a third preset threshold range under the corresponding operating condition.

[0072] This disclosure also provides a computer device, including a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of a lithium battery fault warning method.

[0073] This disclosure also provides a computer-readable storage medium storing computer-readable instructions, which, when executed by a processor, implement the steps of a lithium battery fault warning method.

[0074] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0075] The above description is merely a specific embodiment of this disclosure, enabling those skilled in the art to understand or implement it. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this disclosure. Therefore, this disclosure is not to be limited to the embodiments described herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A lithium battery fault early warning method, characterized in that, include: Obtain the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and perform power function fitting on the time-series relaxation curve to obtain the goodness of fit; The timing data of the imaginary part of the impedance and the electrode temperature in the second preset frequency band of the lithium battery, as well as the timing data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery, are obtained. The internal temperature of the lithium battery is determined by the timing data of the imaginary part of the impedance, and the falling slope of the real part of the impedance is determined by the timing data of the real part of the impedance. The frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. The first joint fusion entropy increment is determined based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature. The voltage, current, load status, and timing data of the imaginary part of impedance in the first preset frequency band of the lithium battery are collected. Based on the timing data of the imaginary part of impedance, the slope of the change of the imaginary part of impedance is determined, and the second joint fusion entropy value is determined by the voltage, the current, the load status, and the slope of change. The goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment are compared with preset thresholds to obtain comparison results, and fault warnings are issued based on the comparison results.

2. The lithium battery fault early warning method according to claim 1, characterized in that, The step of obtaining the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and performing power function fitting on the time-series relaxation curve to obtain the goodness of fit, includes: The measured value of the real part of the impedance is obtained from the time-series relaxation curve of the real part of the impedance. The mean value of the measured values ​​is determined based on the measured values ​​of the real part of the impedance. The goodness of fit is determined based on the measured value of the real part of the impedance and the mean of the measured values.

3. The lithium battery fault early warning method according to claim 1, characterized in that, The process of acquiring timing data of the imaginary part of the impedance and electrode temperature in the second preset frequency band of the lithium battery, and timing data of the real part of the impedance and electrode temperature in the first preset frequency band of the lithium battery, and determining the internal temperature of the lithium battery through the timing data of the imaginary part of the impedance and determining the decreasing slope of the real part of the impedance through the timing data of the real part of the impedance, wherein the frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery, includes: The internal temperature of the lithium battery is determined based on the timing data of the imaginary part of the impedance and the preset quadratic polynomial curve. The measured values ​​of the real part of the impedance at the first time point and the second time point are obtained respectively; Based on the measured value of the real part of the impedance at the first time point and the measured value of the real part of the impedance at the second time point, the difference between the measured values ​​at the first time point and the second time point is determined, and the time interval is determined based on the first time point and the second time point; The slope of the decrease in the real part of the impedance is determined based on the difference between the measured values ​​and the time interval.

4. The lithium battery fault early warning method according to claim 1, characterized in that, The determination of the first joint fusion entropy increment based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature includes: Based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature, the single-parameter permutation entropy of the internal temperature of the lithium battery, the single-parameter permutation entropy of the descent slope, and the single-parameter permutation entropy of the electrode temperature are determined respectively. The single-parameter arrangement entropy of the internal temperature of the lithium battery, the single-parameter arrangement entropy of the decreasing slope, and the single-parameter arrangement entropy of the electrode temperature are weighted and fused with a preset ratio to obtain the first joint fusion entropy value.

5. The lithium battery fault early warning method according to claim 1, characterized in that, Determining the slope of the change of the imaginary part of the impedance based on the time-series data of the imaginary part of the impedance includes: The measured values ​​of the imaginary part of the impedance at the first and second time points are obtained respectively. Based on the measured value of the imaginary part of the impedance at the first time point and the measured value of the imaginary part of the impedance at the second time point, the difference between the measured values ​​at the first time point and the second time point is determined, and the time interval is determined based on the first time point and the second time point; The slope of the change in the imaginary part of the impedance is determined based on the difference between the measured values ​​and the time interval.

6. The lithium battery fault early warning method according to claim 1, characterized in that, The determination of the second joint fusion entropy increment through the voltage, the current, the load state, and the change slope includes: Based on the voltage, the current, the load state, and the slope of change, determine the single-parameter permutation entropy of the voltage, the current, the load state, and the slope of change, respectively; The single-parameter arrangement entropy of the voltage, the single-parameter arrangement entropy of the current, the single-parameter arrangement entropy of the load state, and the single-parameter arrangement entropy of the change slope are weighted and fused with a preset ratio to obtain the second joint fusion entropy value.

7. The lithium battery fault early warning method according to claim 1, characterized in that, The step involves comparing the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment with preset thresholds to obtain comparison results, and then issuing a fault warning based on the comparison results, including: When the goodness of fit is less than a first preset threshold, a first-level warning is triggered; When the increase in the first joint fusion entropy is greater than the second preset threshold range under the corresponding working condition, a second-level warning is triggered. When the increase in the second joint fusion entropy exceeds the third preset threshold range under the corresponding operating condition, a third-level warning is triggered.

8. A lithium battery fault early warning system, characterized in that, include: The fitting module is used to obtain the time-series relaxation curve of the real part of the impedance in the first preset frequency band of the lithium battery, and to perform power function fitting on the time-series relaxation curve to obtain the goodness of fit. The first determining module is used to acquire timing data of the imaginary part of the impedance and the electrode temperature in the second preset frequency band of the lithium battery, as well as timing data of the real part of the impedance and the electrode temperature in the first preset frequency band of the lithium battery, and to determine the internal temperature of the lithium battery through the timing data of the imaginary part of the impedance and to determine the falling slope of the real part of the impedance through the timing data of the real part of the impedance, wherein the frequency in the impedance spectrum of the second preset frequency band of the lithium battery is higher than that in the first preset frequency band of the lithium battery. The second determining module is used to determine the first joint fusion entropy increment based on the internal temperature of the lithium battery, the descent slope, and the electrode temperature; The third determining module is used to collect the voltage, current, load status and time-series data of the imaginary part of the impedance in the first preset frequency band of the lithium battery, determine the slope of the change of the imaginary part of the impedance based on the time-series data of the imaginary part of the impedance, and determine the second joint fusion entropy value through the voltage, the current, the load status and the slope of change; The early warning module is used to compare the goodness of fit, the first joint fusion entropy increment, and the second joint fusion entropy increment with preset thresholds respectively to obtain comparison results, and to issue fault warnings based on the comparison results.

9. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the lithium battery fault warning method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the lithium battery fault warning method as described in any one of claims 1 to 7.

Images