Early warning method and system for thermal runaway of battery pack

Abstract

The invention discloses a battery pack thermal runaway early warning method and system, and relates to the technical field of battery management. The battery pack thermal runaway early warning method comprises the following steps: acquiring temperature and voltage data of each battery unit in a battery pack in real time; performing standardization processing on the temperature and voltage data to eliminate dimensional differences; synchronously extracting a space basis function and a time coefficient based on data decomposition; calculating a spatial entropy based on the spatial basis function; calculating a time entropy based on the time coefficient; the spatial entropy and the time entropy are fused into improved information entropy based on weighted parameters; outputting a prediction information entropy in response to the improved information entropy input time sequence prediction model; and triggering graded early warning based on comparison between the predicted information entropy and a preset information entropy threshold. On-line decomposition is carried out on multi-source data of a battery pack, space and time characteristics are extracted, an improved information entropy is constructed, early warning of thermal runaway is realized in combination with a trend prediction technology, and accuracy and real-time performance are improved.

Description

Technical Field

[0001] This application relates to the field of battery management technology, and in particular to a method and system for early warning of thermal runaway in battery packs. Background Technology

[0002] Lithium-ion battery packs are increasingly widely used in electric vehicles, energy storage systems, and other fields due to their high energy density and long lifespan. However, during operation, battery packs may experience thermal runaway due to internal short circuits, overcharging, or mechanical damage, leading to abnormal temperature increases and even fires or explosions, seriously threatening personal safety and equipment stability. Therefore, developing efficient and real-time thermal runaway early warning technology has become a key requirement in the field of battery management.

[0003] However, existing thermal runaway early warning technologies have the following problems:

[0004] 1. Methods based on physical models involve large amounts of computation and have poor real-time performance;

[0005] 2. Data-driven methods are highly dependent on data and lack universality;

[0006] 3. Multi-source fusion technology has bottlenecks in feature extraction and processing efficiency;

[0007] 4. Existing solutions mostly focus on anomaly detection and fail to effectively achieve early warning of thermal runaway. Summary of the Invention

[0008] To address the aforementioned issues, this application provides a method for early warning of thermal runaway in battery packs. This method involves online decomposition of multi-source data from the battery pack, extraction of spatial and temporal features, construction of an improved information entropy, and integration with trend prediction technology to achieve early warning of thermal runaway, thereby improving accuracy and real-time performance. Correspondingly, a battery pack thermal runaway early warning system is also provided, which can implement methods for early warning of thermal runaway in battery packs under different conditions.

[0009] The first technical solution adopted in this application is: a method for early warning of thermal runaway in a battery pack, comprising the following steps:

[0010] Real-time acquisition of temperature and voltage data of each battery cell in the battery pack; standardization processing of the temperature and voltage data to eliminate dimensional differences;

[0011] Spatial basis functions and time coefficients are extracted synchronously based on data decomposition; the spatial basis functions characterize the spatial distribution characteristics of temperature; and the time coefficients characterize the temperature variation trend over time.

[0012] Spatial entropy is calculated based on the spatial basis function; temporal entropy is calculated based on the time coefficient; the spatial entropy and the temporal entropy are fused into an improved information entropy based on weighted parameters.

[0013] In response to the improved information entropy input time-series prediction model outputting predicted information entropy, a graded early warning is triggered based on the comparison between the predicted information entropy and a preset information entropy threshold.

[0014] In an optional embodiment, the data decomposition includes the following steps:

[0015] The temperature data is subjected to spatiotemporal decomposition to generate spatial basis functions and time coefficients of the spatial basis functions.

[0016] In an optional embodiment, the calculation of the spatial entropy includes the following steps:

[0017] The difference between the quantized space basis functions and the initial space basis functions;

[0018] The differences are normalized into a spatial probability distribution; the spatial entropy is calculated based on the spatial probability distribution.

[0019] In an optional embodiment, the calculation of the time entropy includes the following steps:

[0020] The time coefficients are differentiated to generate a derivative sequence;

[0021] The probability density distribution of the derivative sequence is obtained based on the derivative sequence; the time entropy is calculated based on the probability density distribution.

[0022] In an optional embodiment, the method further includes acquiring the charge / discharge state of the battery pack and adjusting the weighting parameter based on the charge / discharge state.

[0023] In an optional embodiment, the tiered early warning includes a three-level response mechanism:

[0024] Level 1 warning: When the predicted information entropy exceeds the first danger threshold but is lower than the second danger threshold, a status monitoring command is triggered;

[0025] Level 2 warning: When the predicted information entropy exceeds the second danger threshold but is below the third danger threshold, a preventive intervention instruction is triggered;

[0026] Level 3 warning: When the predicted information entropy exceeds the third danger threshold, an emergency shutdown and alarm command will be triggered.

[0027] In an optional embodiment, the preset information entropy threshold is dynamically adjusted based on battery aging parameters; the size of the preset information entropy threshold is adjusted based on the battery aging parameters.

[0028] In an optional embodiment, an anomaly detection function is also included, comprising the following steps:

[0029] Based on historical normal operation data, an improved information entropy probability density function is constructed using kernel density estimation.

[0030] Set the confidence level and calculate the anomaly detection threshold based on the probability density function;

[0031] An anomaly exists if the improved information entropy is greater than the anomaly detection threshold.

[0032] In an optional embodiment, anomaly localization is performed in response to the presence of an anomaly, the anomaly localization including the following steps:

[0033] Calculate the contribution of each spatial basis function to the change in spatial entropy;

[0034] The battery cells corresponding to the spatial basis functions whose contribution exceeds the positioning threshold are marked as thermal runaway risk sources.

[0035] The second technical solution adopted in this application is: providing a battery pack thermal runaway early warning system, including:

[0036] Sensor modules are distributed and mounted on the surface of each battery cell to acquire temperature and voltage data;

[0037] A calculation module, connected to the sensor module, receives temperature and voltage data and outputs an early warning result; the calculation module is configured to execute the battery pack thermal runaway early warning method as described above;

[0038] An alarm module is connected to the computing module to receive the early warning result; the alarm module's operating status is adjusted based on the early warning result.

[0039] Due to the adoption of the above technical solution, this application has at least one of the following beneficial effects compared with the prior art:

[0040] 1. By decomposing multi-source data of the battery pack online, extracting spatial and temporal features, constructing improved information entropy, and combining it with trend prediction technology, early warning of thermal runaway can be achieved, improving accuracy and real-time performance.

[0041] 2. By adjusting the weighted parameters based on the charging and discharging status of the battery pack, and by dynamically adjusting the preset information entropy threshold based on battery aging parameters, the system can maintain high early warning accuracy throughout the entire battery life cycle, thus enhancing its adaptability.

[0042] 3. By using data decomposition technology to extract spatial basis functions and time coefficients online, and then calculating spatial entropy and time entropy based on these, the reliance on large amounts of high-quality data is reduced, thus lowering computational complexity.

[0043] 4. By collecting temperature and voltage data of each battery cell in the battery pack in real time and performing standardization processing to eliminate dimensional differences, and then extracting spatial basis functions and time coefficients based on data decomposition, the changes in the internal state of the battery pack can be captured quickly and accurately. Attached Figure Description

[0044] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0045] in:

[0046] Figure 1 A schematic flowchart of a battery pack thermal runaway early warning method provided in an embodiment of this application;

[0047] Figure 2 This is a schematic diagram of the framework of a battery thermal runaway early warning system provided in an embodiment of this application. Detailed Implementation

[0048] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It is understood that the specific embodiments described herein are only for explaining this application and not for limiting it. Furthermore, it should be noted that, for ease of description, only the parts related to this application are shown in the accompanying drawings, not all structures. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0049] The terms "first," "second," etc., used in this application are used to distinguish different objects, not to describe a specific order. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the listed steps or units, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to these processes, methods, products, or apparatuses.

[0050] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0051] Existing technologies for early warning of thermal runaway in battery packs have significant drawbacks, mainly manifested in high computational complexity, poor real-time performance, and a high dependence on high-quality data leading to insufficient universality and a lack of early warning capabilities. In view of this, this application provides a method for early warning of thermal runaway in battery packs. This method involves online decomposition of multi-source data from the battery pack to extract spatial and temporal features, constructing an improved information entropy, and combining this with trend prediction technology to achieve early warning of thermal runaway, thereby improving accuracy and real-time performance.

[0052] like Figure 1 As shown, Figure 1 A flowchart illustrating a battery pack thermal runaway early warning method according to an embodiment of this application includes the following steps:

[0053] S1: Real-time acquisition of temperature and voltage data of each battery cell in the battery pack; each battery cell is equipped with a temperature sensor and a voltage sensor, and the sensors acquire the temperature and voltage data of each battery cell in real time at a sampling frequency of 1Hz and transmit it to the computing module.

[0054] Some battery packs do not have temperature sensors installed in every battery cell, but voltage and current sensors must be installed to obtain data on the charging and discharging of the battery cells; in another embodiment, the current and voltage data of each battery cell in the battery pack are collected in real time; real-time temperature data is obtained based on the acquired current and voltage data.

[0055] Temperature and voltage data are standardized to eliminate dimensional differences; for temperature data, the following formula is used for standardization:

[0056]

[0057] Among them, T norm To standardize the temperature, T represents the current temperature value. min T is the minimum current value in the historical data of the current battery cell. max This refers to the maximum current value in the historical data of the current battery cell; for example, if the historical temperature range of a certain battery cell is 20℃-30℃, and the currently measured temperature is 25.5℃, then...

[0058] Voltage data is also processed using the same standardized method to ensure consistency in subsequent calculations, which will not be elaborated further here.

[0059] Standardizing temperature and voltage data eliminates the impact of unit differences between different physical quantities, which is beneficial for subsequent data processing and feature extraction. Standardized data is easier for algorithms to understand and process, and can avoid problems such as unstable model training or biased prediction results caused by inconsistent scales of dependent variables.

[0060] S2: Extract spatial basis functions and time coefficients simultaneously based on data decomposition; perform spatiotemporal dimension decomposition on temperature data to generate spatial basis functions and time coefficients of spatial basis functions; temperature is spatiotemporally coupled data, and its spatial distribution and temporal evolution need to be decomposed simultaneously.

[0061] Spatial basis functions characterize the spatial distribution of temperature; time coefficients characterize the trend of temperature variation over time; the decomposition of temperature data is described in detail below:

[0062] Temperature data is collected from temperature sensors on the battery pack. KL (Karhunen-Loève) decomposition is used to decompose the temperature data into spatial basis functions and time coefficients of the spatial basis functions, separating spatial and temporal dynamics. The spatial basis functions reflect the spatial distribution characteristics of temperature, and the time coefficients of the spatial basis functions reflect the trend of temperature change over time.

[0063] Spatial basis functions describe the temperature distribution characteristics at the spatial location of the battery pack at time point t.

[0064] Calculation: Assume the temperature data is represented in matrix form: T represents the time period, and N represents the sensor data.

[0065] For Y k Perform KL decomposition: Y k =Φ k ∑ k A k , Φ k It is a spatial basis function matrix containing N spatial basis vectors, A k This is a time coefficient matrix; a detailed description is provided below with reference to a specific embodiment:

[0066] Assume a 3×3 battery array consisting of 9 lithium-ion battery cells for an electric vehicle. Each battery cell is equipped with a temperature sensor that collects temperature data for 10 seconds. Starting from the 5th second, battery 5 experiences an internal short circuit, and its temperature rapidly rises to 35°C. The other batteries experience a slight temperature increase due to heat conduction.

[0067] Y k Matrix: 9*10.

[0068] time 1 2 3 ··· 10 1 25.1 25.2 25.1 25.4 2 25.1 25.3 25.2 253 3 25.1 25.1 25.1 25.1 ··· 9 25.2 25.1 25.4 25.1 25.1

[0069] For Y kPerform KL decomposition, Φ k Given a 9x9 matrix, extract Φ by reducing its order. k The first two spatial basis vectors are as follows:

[0070]

[0071] Φ1 k It represents a uniform temperature distribution, reflecting the overall temperature field under normal conditions.

[0072] Φ2 k The element corresponding to the battery (the 5th one) is significantly larger, reflecting the local high temperature in the abnormal region.

[0073] Time coefficient A k It is a 10*10 matrix representing the time coefficients of the spatial basis functions;

[0074]

[0075] This indicates a uniform temperature distribution within the battery pack, meaning that under normal conditions, all battery temperatures are similar, and fluctuations are only caused by noise.

[0076] The high-temperature characteristics of the battery indicate that the anomaly is mainly concentrated in battery 5, while the surrounding batteries are slightly heated due to heat conduction; this reflects spatial heterogeneity, that is, the anomaly causes uneven local temperature distribution.

[0077] S3: Calculate spatial entropy based on spatial basis functions; spatial basis functions reflect the spatial distribution pattern of battery pack temperature; under normal operation, the temperature distribution is relatively uniform, and the fluctuation of spatial basis functions is small; under abnormal conditions (such as internal short circuits), the temperature in a certain area rises, and the spatial basis functions will show greater spatial heterogeneity; the calculation of spatial entropy includes the following steps:

[0078] The difference between the quantized spatial basis functions and the initial spatial basis functions; that is, determining the spatial basis functions Φ in an initial state. i 0 , Φ i 0 Given the spatial basis functions obtained under the initial normal operating conditions of the battery pack, calculate the current spatial basis function Φ for each new time point t. i k ; Calculate the difference between the current space basis functions and the initial space basis functions.

[0079] The differences are normalized to a spatial probability distribution. For each element in the spatial basis vector, its difference value is normalized so that the sum of all elements is 1. The difference normalization is achieved based on the following formula:

[0080]

[0081] Spatial entropy is calculated based on spatial probability distribution. The formula for calculating spatial entropy is as follows:

[0082]

[0083] Here, +1 ensures that the spatial entropy value is non-negative. This indicates that the upper limit of the summation is 2, meaning that the probability density function is divided into two intervals or states.

[0084] Φ1 obtained in step S2 k To ensure a uniform temperature distribution, let Φ1 k Let Φ2 be the initial space basis function; k Let be the spatial basis functions at the current time; then

[0085] Normalizing the difference values ​​at each location i of the battery yields the spatial probability distribution:

[0086]

[0087] We know that p(k) = [0.05, 0.05, 0.05, 0.10, 0.38, 0.10, 0.05, 0.05, 0.05] T .

[0088]

[0089] The formula divides the probability density function into two intervals: a normal region with low probability and an abnormal region with high probability. It then uses the two highest probability values ​​for calculation, specifically probabilities of 0.38 and 0.10.

[0090] h s (k)=0.38·log2(0.38)+0.10·log2(0.10)+1≈0.47.

[0091] Time entropy is calculated based on a time coefficient. Time entropy is the derivative of the time coefficient and reflects the rate of temperature change over time. The calculation of time entropy includes the following steps:

[0092] Differentiate the time coefficient to generate a derivative sequence;

[0093] Obtain the probability density distribution of the derivative sequence based on the derivative sequence; quantify the disorder of temperature change using the derivative; calculate the time entropy based on the probability density distribution.

[0094] The probability density function of the derivative is calculated based on kernel density estimation, and then the time entropy value is calculated; the calculation formula is shown below:

[0095]

[0096] The specific steps for calculating time entropy will not be elaborated here.

[0097] Spatial entropy and temporal entropy are fused based on weighted parameters to form improved information entropy; the formula for calculating improved information entropy is as follows:

[0098] H(k)=αh s (k)+(1-α)h t (k)

[0099] Where H(k) is the improved information entropy, h s (k) is the spatial entropy, h t (k) represents the time entropy, and α is the weighting parameter.

[0100] The influence ratio of spatial entropy and temporal entropy on improving information entropy can be adjusted by modifying the weighting parameters. This also includes acquiring the charge / discharge state of the battery pack and adjusting the weighting parameters based on the charge / discharge state. For example, when the battery is charging, temperature changes are more pronounced, and local temperature differences are more easily highlighted; in this case, the weight of spatial entropy is increased, and the weighting parameter is adjusted to 0.6. When the battery is discharging, voltage fluctuations are more significant, and temporal dynamic changes are more drastic; in this case, the weight of temporal entropy is increased, and the weighting parameter is adjusted to 0.4. In other embodiments, the size of the weighting parameters and the adjustment factors can be selected differently, and no limitations are imposed.

[0101] S4: In response to the improved information entropy input time-series prediction model, output the predicted information entropy; input the calculated information entropy sequence into a pre-trained Long Short-Term Memory (LSTM) network to predict the information entropy value for the next 10 minutes; the LSTM network configuration is as follows: number of layers: 2 layers; number of hidden units: 50; training data: historical normal and abnormal samples.

[0102] In other embodiments, the sampling time and number of samples may be selected separately, and no limitation is made thereto.

[0103] A tiered early warning system is triggered based on a comparison between the predicted information entropy and a preset information entropy threshold; the tiered early warning system includes a three-level response mechanism:

[0104] Level 1 warning: When the predicted information entropy exceeds the first danger threshold but is lower than the second danger threshold, a status monitoring command is triggered;

[0105] Level 2 warning: When the predicted information entropy exceeds the second danger threshold but is below the third danger threshold, a preventive intervention instruction is triggered;

[0106] Level 3 warning: When the predicted information entropy exceeds the third danger threshold, an emergency shutdown and alarm command will be triggered.

[0107] In this embodiment, the first danger threshold is 1.2 times the average information entropy; the second danger threshold is 1.5 times the average information entropy; and the third danger threshold is 2.0 times the average information entropy. In other embodiments, the specific values ​​of the first danger threshold, the second danger threshold, and the third danger threshold can be selected separately, and no limitation is made in this regard.

[0108] The preset information entropy threshold is dynamically adjusted based on battery aging parameters; the value of the preset information entropy threshold is adjusted based on battery aging parameters; considering the impact of battery aging on abnormal characteristics, the threshold is adjusted according to the battery capacity decay rate after every 100 charge-discharge cycles. For example: if the capacity decays by 5%, the threshold is increased by 2%; adjustment formula:

[0109] New threshold = Original threshold × (1 + 0.02 × attenuation percentage)

[0110] Ensure the early warning system remains accurate throughout the battery's entire lifespan.

[0111] The battery pack thermal runaway early warning method also includes an anomaly detection function, comprising the following steps:

[0112] Based on historical normal operation data, an improved information entropy probability density function is constructed using kernel density estimation.

[0113] Set the confidence level and calculate the anomaly detection threshold based on the probability density function;

[0114] An anomaly is considered to exist if the improved information entropy is greater than the anomaly detection threshold; otherwise, an anomaly is not considered to exist.

[0115] In response to the existence of an anomaly, an anomaly localization process is performed, which includes the following steps:

[0116] Calculate the contribution of each spatial basis function to the change in spatial entropy;

[0117] Battery cells whose contribution exceeds the location threshold are marked as thermal runaway risk sources.

[0118] The entropy contribution function is the contribution of each spatial location to the maximum information entropy, and the calculation formula is as follows:

[0119]

[0120] z = [x, y] T

[0121] Once an anomaly is detected, the coordinates corresponding to the maximum information entropy change can be considered as the location of the anomaly.

[0122] This application also provides a battery thermal runaway early warning system, such as Figure 2 As shown, Figure 2A schematic diagram of the framework of a battery thermal runaway early warning system provided in an embodiment of this application includes:

[0123] Sensor modules are distributed and mounted on the surface of each battery cell to acquire temperature and voltage data.

[0124] A calculation module is connected to a sensor module to receive temperature and voltage data and output early warning results; the calculation module is configured to execute the battery pack thermal runaway early warning method as described in the above embodiment.

[0125] An alarm module is connected to a computing module to receive early warning results; the alarm module's operating status is adjusted based on the early warning results; in this embodiment, the alarm module includes LED indicator lights: displaying green (normal), yellow (level 1 warning), orange (level 2 warning), and red (level 3 warning); and a buzzer: emitting an audible alarm during level 3 warnings.

[0126] The following describes the workflow of the battery thermal runaway early warning system:

[0127] The sensor acquires temperature and voltage data at a frequency of 1 Hz;

[0128] After the data is preprocessed by the processor, the information entropy value is calculated.

[0129] The information entropy sequence is input into the LSTM model for prediction;

[0130] Based on the comparison between the prediction results and the threshold, a corresponding early warning is triggered;

[0131] LED lights indicate the current status, and a buzzer sounds an alarm in case of an emergency.

[0132] In the several embodiments provided in this application, it should be understood that the disclosed methods and devices can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.

[0133] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0134] Furthermore, the functional units in the various embodiments of this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0135] The above description is merely an embodiment of this application and does not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A method for early warning of thermal runaway in a battery pack, characterized in that, Includes the following steps: Real-time acquisition of temperature and voltage data of each battery cell in the battery pack; standardization processing of the temperature and voltage data to eliminate dimensional differences; Spatial basis functions and time coefficients are extracted synchronously based on data decomposition; the spatial basis functions characterize the spatial distribution characteristics of temperature; and the time coefficients characterize the temperature variation trend over time. Calculate spatial entropy based on the spatial basis functions; calculate temporal entropy based on the time coefficients; The spatial entropy and the temporal entropy are fused together based on weighted parameters to form an improved information entropy; In response to the improved information entropy input time-series prediction model, the predicted information entropy is output. A tiered early warning is triggered based on the comparison between the predicted information entropy and the preset information entropy threshold.

2. The battery pack thermal runaway early warning method according to claim 1, characterized in that, The data decomposition includes the following steps: The temperature data is subjected to spatiotemporal decomposition to generate spatial basis functions and time coefficients of the spatial basis functions.

3. The battery pack thermal runaway early warning method according to claim 1, characterized in that, The calculation of the spatial entropy includes the following steps: The difference between the quantized space basis functions and the initial space basis functions; The differences are normalized into a spatial probability distribution; the spatial entropy is calculated based on the spatial probability distribution.

4. The battery pack thermal runaway early warning method according to claim 1, characterized in that, The calculation of the time entropy includes the following steps: The time coefficients are differentiated to generate a derivative sequence; The probability density distribution of the derivative sequence is obtained based on the derivative sequence; the time entropy is calculated based on the probability density distribution.

5. The battery pack thermal runaway early warning method according to claim 1, characterized in that, It also includes acquiring the charge / discharge state of the battery pack; and adjusting the weighting parameter based on the charge / discharge state.

6. The battery pack thermal runaway early warning method according to claim 1, characterized in that, The tiered early warning system includes a three-level response mechanism: Level 1 warning: When the predicted information entropy exceeds the first danger threshold but is lower than the second danger threshold, a status monitoring command is triggered; Level 2 warning: When the predicted information entropy exceeds the second danger threshold but is below the third danger threshold, a preventive intervention instruction is triggered; Level 3 warning: When the predicted information entropy exceeds the third danger threshold, an emergency shutdown and alarm command will be triggered.

7. The battery pack thermal runaway early warning method according to claim 6, characterized in that, The preset information entropy threshold is dynamically adjusted based on battery aging parameters; the size of the preset information entropy threshold is adjusted based on the battery aging parameters.

8. The battery pack thermal runaway early warning method according to claim 1, characterized in that, It also includes anomaly detection functionality, comprising the following steps: Based on historical normal operation data, an improved information entropy probability density function is constructed using kernel density estimation. Set the confidence level and calculate the anomaly detection threshold based on the probability density function; An anomaly exists if the improved information entropy is greater than the anomaly detection threshold.

9. The battery pack thermal runaway early warning method according to claim 8, characterized in that, In response to the existence of an anomaly, an anomaly localization is performed, which includes the following steps: Calculate the contribution of each spatial basis function to the change in spatial entropy; The battery cells corresponding to the spatial basis functions whose contribution exceeds the positioning threshold are marked as thermal runaway risk sources.

10. A battery pack thermal runaway early warning system, characterized in that, include: Sensor modules are distributed and mounted on the surface of each battery cell to acquire temperature and voltage data; A computing module is connected to the sensor module to receive temperature and voltage data and output early warning results; The computing module is configured to execute the battery pack thermal runaway early warning method as described in any one of claims 1-9; An alarm module is connected to the computing module to receive the early warning result; the alarm module's operating status is adjusted based on the early warning result.

Images