Thermal runaway prediction method for power battery, vehicle and storage medium

Abstract

Provided in the present application are a thermal runaway prediction method for a power battery, a vehicle and a storage medium. The method is applied to the field of vehicles, and comprises: acquiring a state parameter of each battery cell of a power battery under test; on the basis of the state parameter of each battery cell, calculating a plurality of risk indicators of the battery cell from different dimensions, the risk indicators being used for measuring the risk of occurrence of abrupt thermal runaway in the battery cell; on the basis of the plurality of risk indicators and respective weight factors corresponding to the plurality of risk indicators, determining a risk score of occurrence of abrupt thermal runaway in each battery cell; and, on the basis of the risk score, determining an abrupt thermal runaway prediction result of the power battery. The method can accurately predict abrupt thermal runaway of power batteries, thereby facilitating accurate early warning for abrupt thermal runaway of power batteries.

Description

A method for predicting thermal runaway of a power battery, a vehicle, and a storage medium.

[0001] This application claims priority to Chinese Patent Application No. 2024116197687, filed on November 13, 2024, entitled "A Method for Predicting Thermal Runaway of a Power Battery, a Vehicle, and a Storage Medium", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of vehicles, and more specifically, to a method for predicting thermal runaway of a power battery, a vehicle, and a storage medium in the field of vehicles. Background Technology

[0003] Thermal runaway in power batteries is mainly classified into two categories: gradual thermal runaway and abrupt thermal runaway. Gradual thermal runaway, with its evolution cycle of several weeks or months and significant characteristics such as battery voltage or self-discharge rate, allows for accurate early warning. However, abrupt thermal runaway, lacking obvious warning signs before the accident (voltage, SOC, capacity, self-discharge rate, etc., appear normal) and developing rapidly—from slight symptoms to thermal runaway within hours or days—makes accurate early warning impossible. Therefore, accurately predicting abrupt thermal runaway in power batteries has become a pressing technical problem. Summary of the Invention

[0004] This application provides a method for predicting thermal runaway of a power battery, a vehicle, and a storage medium. The method can accurately predict sudden thermal runaway of a power battery, thereby facilitating accurate early warning of sudden thermal runaway of a power battery.

[0005] In a first aspect, this application provides a method for predicting thermal runaway of a power battery. The method includes: acquiring state parameters of a single cell of the power battery under test; calculating multiple risk indicators of the single cell from different dimensions based on the state parameters of the single cell; wherein the risk indicators are used to measure the risk of sudden thermal runaway of the single cell; determining a risk score for sudden thermal runaway of the single cell based on the multiple risk indicators and their respective weighting factors; and determining a prediction result for sudden thermal runaway of the power battery based on the risk score.

[0006] The above technical solution achieves accurate prediction of sudden thermal runaway by meticulously monitoring and comprehensively analyzing the state parameters of individual battery cells. Considering that related technologies often use the state parameters of the entire battery pack for risk prediction, but the state parameters of the entire battery pack may lead to the neglect of minor differences or deterioration in individual cells, making it difficult to accurately predict sudden thermal runaway, this embodiment uses individual battery cells within the power battery as the monitoring and analysis object. By acquiring the state parameters of each individual cell and calculating multiple risk indicators for each cell from different dimensions, multi-dimensional risk indicators can more comprehensively reflect the health status of individual cells. By comprehensively considering these indicators, potential risk signs can be captured, improving the accuracy of risk assessment. Through weighted scoring, the importance of each risk indicator to the risk score of an individual cell can be quantified, making the risk assessment more objective and scientific. The risk scores of each individual cell provide reliable data references for accurately determining the prediction results of sudden thermal runaway of the power battery. Therefore, accurate prediction of sudden thermal runaway of the power battery can be achieved.

[0007] In some embodiments, the weight factors corresponding to each of the plurality of risk indicators are determined as follows: A plurality of training samples are obtained; wherein each training sample includes state parameters of a sample power battery, and each training sample is labeled with a sample tag, the sample tag representing whether the sample power battery has experienced abrupt thermal runaway under the state parameters; sample features of each training sample are determined; wherein the sample features include: a plurality of risk indicators of a single cell within the sample power battery calculated from different dimensions; a supervised training risk prediction model is performed based on the sample features and the sample tags to obtain the weight factors corresponding to each of the plurality of risk indicators; wherein the weight factors corresponding to each of the plurality of risk indicators enable the difference between the risk prediction result output by the risk prediction model and the sample tags labeled on the input training samples to be less than or equal to a preset difference.

[0008] In some embodiments, the plurality of risk indicators include any combination of the following: a single cell voltage outlier index, used to characterize whether the voltage of a single cell of the power battery is abnormal; a differential voltage index, used to characterize whether the voltage change of the single cell during charging and discharging is abnormal; a voltage entropy index, used to characterize whether the consistency of the voltage distribution of the single cell of the power battery is abnormal; and a single cell internal resistance index, used to characterize whether the internal resistance of the single cell of the power battery is abnormal.

[0009] In some embodiments, the state parameters of the individual battery cell include: the sampled voltage of the individual battery cell; the individual battery cell voltage outlier index is determined by: calculating the average voltage and voltage variance of the individual battery cells in the power battery based on the sampled voltage of each individual battery cell in the power battery; and calculating the individual battery cell voltage outlier index corresponding to each individual battery cell based on the sampled voltage of each individual battery cell in the power battery, the average voltage value, and the voltage variance value.

[0010] In some embodiments, the state parameters of the individual battery cell include: the sampling voltage, sampling current, and charge / discharge state of the individual battery cell; the differential voltage index is determined as follows: if the charge / discharge state of the individual battery cell is a charging state, then based on the sampling voltage and sampling current of the individual battery cell in the charging state, the differential voltage of the individual battery cell during the charging process is calculated, and within a preset monitoring period, the first number of times the rate of change of the differential voltage of the individual battery cell during the charging process exceeds a preset rate of change is determined; if the charge / discharge state of the individual battery cell is a discharging state, then based on the sampling voltage and sampling current of the individual battery cell in the discharging state, the differential voltage of the individual battery cell during the discharging process is calculated, and within a preset monitoring period, the second number of times the rate of change of the differential voltage of the individual battery cell during the discharging process exceeds a preset rate of change is determined; the differential voltage index is determined based on the first number and the second number.

[0011] In some embodiments, the state parameters of the individual battery cell include: the sampled voltage of the individual battery cell; the voltage entropy index is determined by: calculating the voltage entropy value of each individual battery cell based on the sampled voltage of each individual battery cell in the power battery within a preset time window; calculating the average entropy value and the standard deviation of the entropy value based on the voltage entropy value of each individual battery cell; calculating an anomaly coefficient based on the voltage entropy value, the average entropy value, and the standard deviation of the entropy value of each individual battery cell, and using the anomaly coefficient as the voltage entropy index.

[0012] In some embodiments, the step of calculating anomaly coefficients based on the voltage entropy value, the average entropy value, and the standard deviation of the entropy value of each individual cell, and using the anomaly coefficients as a voltage entropy index, includes: for each individual cell, calculating the difference between the voltage entropy value of the individual cell and the average entropy value, and using the quotient obtained by dividing the difference by the standard deviation of the entropy value as the anomaly coefficient of the individual cell.

[0013] In some embodiments, the state parameters of the individual battery cell include: the acquisition voltage, acquisition current, open-circuit voltage, and polarization voltage of the individual battery cell; the internal resistance index of the individual battery cell is determined by: constructing an equivalent circuit model of the individual battery cell; wherein, the equivalent circuit model is used to estimate the internal resistance of the individual battery cell; the acquisition voltage, acquisition current, open-circuit voltage, and polarization voltage of the individual battery cell are substituted into the equivalent circuit model to estimate the internal resistance of the individual battery cell, and the estimated internal resistance is used as the internal resistance index of the individual battery cell.

[0014] In some embodiments, determining the risk score of the single cell exhibiting sudden thermal runaway based on the plurality of risk indicators and the weighting factors corresponding to each of the plurality of risk indicators includes: determining the risk score corresponding to each of the plurality of risk indicators based on the plurality of risk indicators of the single cell; and weighting the risk scores corresponding to each of the plurality of risk indicators based on the weighting factors corresponding to each of the plurality of risk indicators to obtain the risk score of the single cell exhibiting sudden thermal runaway.

[0015] Secondly, this application provides a thermal runaway prediction device for a power battery. The device includes: an acquisition module for acquiring state parameters of each individual cell in the power battery under test; a calculation module for calculating multiple risk indicators of the individual cell from different dimensions based on the state parameters of the individual cell; wherein the risk indicators are used to measure the risk of sudden thermal runaway of the individual cell; a risk score determination module for determining a risk score of sudden thermal runaway of the individual cell based on the multiple risk indicators and the weighting factors corresponding to the multiple risk indicators; and a prediction module for determining a prediction result of sudden thermal runaway of the power battery based on the risk score.

[0016] Thirdly, this application provides a vehicle, comprising: a memory for storing executable program code; and a processor for calling and running the executable program code from the memory, causing the vehicle to perform the methods described in the first aspect or any of the embodiments of the first aspect.

[0017] Fourthly, this application provides a computer program product comprising: computer program code, which, when run on a computer, causes the computer to perform the methods described in the first aspect or any of the embodiments of the first aspect.

[0018] Fifthly, this application provides a computer-readable storage medium storing computer program code that, when executed on a computer, causes the computer to perform the methods described in the first aspect or any of the embodiments of the first aspect. Attached Figure Description

[0019] Figure 1 is a schematic flowchart of a thermal runaway prediction method for a power battery provided in an embodiment of this application;

[0020] Figure 2 is a schematic diagram of the weight factors of each risk indicator obtained through training according to an embodiment of this application;

[0021] Figure 3 is a schematic diagram of the structure of a thermal runaway prediction device for a power battery provided in an embodiment of this application;

[0022] Figure 4 is a structural schematic diagram of a vehicle provided in an embodiment of this application. Embodiments of the present invention

[0023] The technical solutions in this application will be clearly and thoroughly described below with reference to the accompanying drawings. In the description of the embodiments of this application, unless otherwise stated, " / " means "or," for example, A / B can mean A or B. "And / or" in the text is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, and B existing alone. Furthermore, in the description of the embodiments of this application, "multiple" refers to two or more than two.

[0024] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature.

[0025] Thermal runaway in a power battery refers to an uncontrollable heat release phenomenon caused by an abnormal increase in temperature under certain conditions. Thermal runaway can be triggered by factors such as internal short circuits, overcharging, external heating, and mechanical damage. Based on its development process, thermal runaway can be divided into two categories: gradual thermal runaway and abrupt thermal runaway. Gradual thermal runaway refers to a situation where the battery temperature gradually rises, reaching a critical point after a relatively long period, thus triggering thermal runaway. Abrupt thermal runaway refers to a situation where the battery temperature rises sharply within a short period, rapidly reaching a runaway state.

[0026] This application's research reveals that gradual thermal runaway, due to its evolution cycle of several weeks or months and the significant characteristics of the battery such as voltage or self-discharge rate, can be accurately predicted in advance. However, abrupt thermal runaway, lacking obvious warning signs before the accident (voltage, remaining charge, capacity, self-discharge rate, etc. appear normal) and developing rapidly—from slight symptoms to thermal runaway in just a few hours or days—makes accurate prediction impossible. Therefore, accurately predicting abrupt thermal runaway in power batteries has become an urgent technical problem to be solved.

[0027] Based on this, and to at least solve the aforementioned technical problems, this application provides a method for predicting thermal runaway of a power battery, applicable to vehicles equipped with power batteries. This method, while ensuring the safety of the power battery, can detect the evolutionary risks of the power battery in advance. It amplifies the minute differences and deterioration levels of individual cells within the power battery, and through multi-dimensional analysis of the risk of abrupt thermal runaway, achieves accurate prediction of abrupt thermal runaway of the power battery.

[0028] Figure 1 is a schematic flowchart of a thermal runaway prediction method for a power battery provided in an embodiment of this application.

[0029] For example, as shown in Figure 1, the thermal runaway prediction method for a power battery includes:

[0030] Step 101: Obtain the state parameters of each individual cell in the power battery to be tested.

[0031] Step 102: Calculate multiple risk indicators for a single cell from different dimensions based on the state parameters of the single cell; among them, the risk indicators are used to measure the risk of a single cell experiencing sudden thermal runaway.

[0032] Step 103: Determine the risk score of sudden thermal runaway in a single cell based on multiple risk indicators and their respective weighting factors.

[0033] Step 104: Determine the prediction result of sudden thermal runaway of the power battery based on the risk score.

[0034] In the embodiment shown in Figure 1, precise prediction of sudden thermal runaway is achieved through meticulous monitoring and comprehensive analysis of the state parameters of individual cells in the power battery. Considering that related technologies often use the state parameters of the entire battery pack for risk prediction, but the state parameters of the entire battery pack may lead to the neglect of minor differences or deterioration in individual cells, making it difficult to accurately predict sudden thermal runaway, this embodiment uses individual cells within the power battery as the monitoring and analysis object. By acquiring the state parameters of each individual cell and calculating multiple risk indicators for each cell from different dimensions, multi-dimensional risk indicators can more comprehensively reflect the health status of individual cells. By comprehensively considering these indicators, potential risk signs can be captured, improving the accuracy of risk assessment. Through weighted scoring, the importance of each risk indicator to the risk score of an individual cell can be quantified, making the risk assessment more objective and scientific. The risk scores of each individual cell provide reliable data references for accurately determining the prediction results of sudden thermal runaway in the power battery. Therefore, this embodiment can achieve precise prediction of sudden thermal runaway in the power battery.

[0035] The specific implementation of each step in the embodiment shown in Figure 1 is described below:

[0036] In step 101, the power battery under test is the power battery for which abrupt thermal runaway prediction is to be performed. This power battery can be understood as the power battery in the vehicle under test. The state parameters can be: relevant parameters of each individual cell collected during the operation of the power battery. These state parameters can include: the collected voltage, collected current, charge / discharge state, and insulation resistance of each individual cell, etc., and each state parameter corresponds to a collection time point t. In specific implementation, during the operation of the power battery, the state parameters of each individual cell can be collected periodically, and the state parameters of each individual cell collected at different collection time points t can be saved to provide detailed reference data for subsequent abrupt thermal runaway prediction.

[0037] For example, step 101 above can specifically involve: obtaining the state parameters of each individual cell in the power battery under test within the current time period. The current time period can be set according to actual needs, such as within the current week, the current month, or the current six months.

[0038] Optionally, to facilitate easier tracking of the thermal runaway evolution of the power battery, step 101 above can be specifically described as: obtaining the state parameters of each individual cell in the power battery under test throughout its entire life cycle. Here, the entire life cycle can be understood as the time period from the time the power battery was manufactured to the current time.

[0039] In step 102, multiple risk indicators for each individual battery cell are calculated from different dimensions based on the state parameters of each cell. These multiple risk indicators are used to measure the risk of sudden thermal runaway in each individual battery cell from different perspectives. In this step, each individual battery cell inside the power battery is taken as the object of risk analysis. For each individual battery cell, multi-dimensional risk indicators can be calculated, which can more comprehensively reflect the health status of the individual battery cell.

[0040] For example, multiple risk indicators include any combination of the following: individual cell voltage outlier, differential voltage, voltage entropy, and individual cell internal resistance. These risk indicators are explained below:

[0041] The cell voltage outlier index is used to characterize whether the voltage of a single cell in a power battery is abnormal. For example, it uses mathematical statistics to identify which cells in a power battery have voltage values ​​that deviate from the normal voltage range of most cells, thus helping to identify cells with potential voltage anomalies.

[0042] In some embodiments, the state parameters of a single battery cell include: the sampled voltage of the single battery cell. The determination of this single-cell voltage outlier index includes the following steps S11 to S12:

[0043] S11: Calculate the average voltage and voltage variance of each cell in the power battery based on the collected voltage of each cell.

[0044] Specifically, the average voltage of a single battery cell can be calculated using the following formula:

[0045]

[0046] in, This represents the average voltage, and N represents the total number of individual cells in the power battery. This represents the sampling voltage of the i-th individual battery cell.

[0047] After calculating the average voltage of a single battery cell, the voltage variance of the individual cells in the power battery is calculated based on the average voltage and the sampled voltage of each cell. For example, the voltage variance can be calculated using the following formula:

[0048]

[0049] in, This represents the voltage variance value. This represents the average voltage, and N represents the total number of individual cells in the power battery. This represents the sampling voltage of the i-th individual battery cell.

[0050] S12: Calculate the individual cell voltage outlier index for each individual cell based on the collected voltage, average voltage value, and voltage variance value of each individual cell in the power battery.

[0051] Specifically, for each individual cell, the absolute value of the difference between the sampled voltage and the average voltage can be calculated, and the quotient obtained by dividing this absolute value by the voltage variance value can be used as the individual cell voltage outlier index. For example, the individual cell voltage outlier index can be calculated using the following formula:

[0052]

[0053] in, This represents the outlier index of the voltage of the i-th individual cell. This represents the sampled voltage of the i-th individual battery cell. This represents the average voltage. This represents the voltage variance value.

[0054] In practical implementation, during the operation of the power battery, the state parameters of each individual cell can be periodically collected and stored at different collection time points t. Therefore, when calculating the individual cell voltage outlier index, the index can be calculated based on the collected voltages of each individual cell at the same collection time point.

[0055] Understandably, a higher single-cell voltage outlier indicates a greater difference between the voltage of that single cell and the voltages of other cells in the battery pack. If the voltage of some single cells deviates significantly from that of other cells, it may indicate poor health and potential failure risks. Sudden thermal runaway can occur when internal faults occur within a single cell, such as internal short circuits, which may initially manifest as abnormal cell voltage. Therefore, monitoring the single-cell voltage outlier index can provide a reliable reference for predicting sudden thermal runaway.

[0056] The following is an introduction to the relevant content of the differential voltage index:

[0057] The differential voltage index is used to characterize whether the voltage change of a single battery cell is abnormal during charging and discharging. The differential voltage index identifies abnormal behavior of a single battery cell during charging and discharging by analyzing the rate of change of voltage with capacity or time.

[0058] For example, the state parameters of a single battery cell include: the sampled voltage, sampled current, and charge / discharge state of the single battery cell. The determination method for the differential voltage index includes the following steps S21 to S23:

[0059] S21: If the charging / discharging state of a single cell is the charging state, then based on the collected voltage and current of the single cell in the charging state, calculate the differential voltage of the single cell during the charging process, and within the preset monitoring time, determine the first number of times that the rate of change of the differential voltage of the single cell during the charging process exceeds the preset rate of change.

[0060] Specifically, during the charging process, the voltage and current of individual battery cells can be periodically collected. The collection frequency can be set according to specific needs, such as once per second or once every 60 seconds. The differential voltage during charging can be expressed as dQ / dV, where dQ / dV represents the differential of voltage change with capacity. dQ / dV = I*dt / dV, where I refers to the collected current at a certain time t (usually measured in amperes, A). dt is an infinitesimally small time interval used to describe a continuously changing process. dV is the change in voltage within the time interval dt. dQ is the change in charge within the time interval dt, and dQ can be approximated as I*dt.

[0061] Understandably, if the voltage and current of a single battery cell are collected every 60 seconds, a differential voltage can be calculated every 60 seconds. Based on this, during charging, the rate of change of the differential voltage can be determined by the absolute value of the difference between two adjacent calculated differential voltages. If this rate of change is greater than a preset rate of change, this instance is recorded. Finally, the first time the rate of change of the differential voltage during charging exceeds the preset rate of change is accumulated within a preset monitoring period. This preset monitoring period can be set according to actual needs, such as one day or one charge-discharge cycle. The preset rate of change can be pre-set to measure whether the change in differential voltage is normal.

[0062] S22: If the charging / discharging state of a single cell is the discharging state, then based on the collected voltage and collected current of the single cell in the discharging state, calculate the differential voltage of the single cell during the discharging process, and within the preset monitoring time, determine the second number of times that the rate of change of the differential voltage of the single cell during the discharging process exceeds the preset rate of change.

[0063] Specifically, the method for calculating the differential voltage of a single cell during the discharge process is roughly the same as the method for calculating the differential voltage of a single cell during the charging process, and will not be repeated here to avoid repetition.

[0064] When determining the second number, the rate of change of the differential voltage can be determined based on the absolute value of the difference between two adjacent differential voltages calculated during the discharge process. If the rate of change is greater than the preset rate of change, the case where the rate of change is greater than the preset rate of change is recorded. Finally, the second number of times the rate of change of the differential voltage during the discharge process exceeds the preset rate of change is accumulated within the preset monitoring time.

[0065] It should be noted that in this embodiment, since the rate of change of the differential voltage is calculated based on the absolute value of the difference between two adjacent differential voltage calculations, the rate of change of the differential voltage is always positive.

[0066] S23: Determine the differential voltage index based on the first and second counts.

[0067] The first count characterizes the degree of abnormality in the rate of change of the differential voltage during charging, while the second count characterizes the degree of abnormality in the rate of change of the differential voltage during discharging. Therefore, in this embodiment, the first and second counts corresponding to a single cell are directly used as the differential voltage index. Alternatively, the sum of the first and second counts can be used as the differential voltage index.

[0068] If the first count exceeds the calibration count, it indicates that the rate of change of the differential voltage in the individual cell exceeds the preset rate of change multiple times during charging. If the second count exceeds the calibration count, it indicates that the rate of change of the differential voltage in the individual cell exceeds the preset rate of change multiple times during discharging. Whether the first or second count exceeds the calibration count, it indicates that the internal problems of the individual cell may be continuously worsening, and the risk of thermal runaway is significantly increased.

[0069] The following section introduces the relevant content of the voltage entropy index:

[0070] The voltage entropy index is used to characterize whether the voltage distribution of a single cell in a power battery is abnormal. The voltage entropy index can be seen as a quantification of the voltage distribution of a single cell within a preset time window, similar to the concept of entropy in information theory. The voltage entropy index can be used to evaluate the voltage stability or fluctuation of a single cell within a preset time window.

[0071] For example, the state parameters of a single battery cell include: the sampled voltage of the single battery cell; the determination method of the voltage entropy index includes the following steps S31 to S33:

[0072] S31: Calculate the voltage entropy value of each individual cell in the power battery based on the collected voltage within a preset time window.

[0073] The preset time window can be set according to actual needs, such as one day; this embodiment does not impose a specific limitation on this. For each individual cell, the voltage entropy value of the individual cell can be calculated based on the various collected voltages (V_1, V_2, ..., V_n) collected within the preset time window.

[0074] For example, for each individual cell, the voltage entropy value of that individual cell can be calculated as follows:

[0075]

[0076] in, This represents the voltage entropy value of the individual battery cell. M represents the number of different voltage values ​​of the individual battery cell collected within a preset time window (e.g., one day). In other words, there are a total of M different voltage values ​​collected for the individual battery cell within the preset time window. This represents the probability that the m-th sampled voltage of the single cell occurs among all sampled voltages (V_1, V_2, ..., V_n).

[0077] S32: Calculate the average entropy value and the standard deviation of entropy value based on the voltage entropy value of each individual cell.

[0078] Understandably, the voltage entropy value of all individual cells in the power battery has been calculated using the aforementioned formula. The average entropy value is obtained by summing the voltage entropy values ​​of all individual cells and then dividing by the total number of individual cells. Based on the standard deviation formula, the standard deviation of the entropy value is calculated using the average entropy value and the voltage entropy value of each individual cell.

[0079] S33: Calculate the anomaly coefficient based on the voltage entropy value, average entropy value, and standard deviation of each individual cell, and use the anomaly coefficient as a voltage entropy index.

[0080] Specifically, the anomaly coefficient characterizes the degree of deviation of the voltage entropy value of each individual cell from the average entropy value. By monitoring the voltage entropy index of each individual cell, it is possible to quickly identify cases where the voltage entropy index differs significantly from that of other individual cells. This indicates that the individual cell may have potential problems, such as aging, failure, or thermal runaway.

[0081] For example, the above calculation of the anomaly coefficient based on the voltage entropy value, average entropy value, and standard deviation of each individual cell, and the use of the anomaly coefficient as a voltage entropy index, includes: for each individual cell, calculating the difference between the voltage entropy value and the average entropy value of the individual cell, and using the quotient obtained by dividing the difference by the standard deviation of the entropy value as the anomaly coefficient of the individual cell.

[0082] Specifically, for each individual cell, the anomaly coefficient is calculated as follows: (Voltage entropy value of the individual cell - Average entropy value) / Standard deviation of entropy value. Based on the calculation method for voltage entropy, since both the voltage entropy value and the average entropy value are negative, the anomaly coefficient can be either positive or negative. If the anomaly coefficient of an individual cell is negative, it indicates that the voltage entropy value of that individual cell is lower than the average entropy value. If the anomaly coefficient of an individual cell is positive, it indicates that the voltage entropy value of that individual cell is higher than the average entropy value. However, regardless of whether the anomaly coefficient is positive or negative, the larger the absolute value of the difference between the voltage entropy value and the average entropy value of an individual cell, the greater the deviation of the voltage entropy value from the average entropy value. If the anomaly coefficient of an individual cell is positive, the larger the anomaly coefficient, the greater the risk of sudden thermal runaway. If the anomaly coefficient of an individual cell is negative, the larger the anomaly coefficient, the lower the risk of sudden thermal runaway.

[0083] In practical implementation, a calibration anomaly coefficient range can be defined. The upper limit of the calibration anomaly coefficient range is a positive number, and the lower limit is a negative number, typically being opposites of each other. If the anomaly coefficient of a single cell falls outside the calibration anomaly coefficient range, it can be preliminarily determined that the single cell may be at risk of sudden thermal runaway. For example, if the anomaly coefficient of a single cell is greater than the upper limit of the calibration anomaly coefficient range or less than the lower limit, it indicates that the single cell may be at risk of sudden thermal runaway.

[0084] The following section introduces the relevant content regarding the internal resistance index of a single unit:

[0085] The internal resistance index of a single cell is used to characterize whether the internal resistance of a single cell in a power battery is abnormal. It is an indicator that describes the magnitude of the internal resistance of a single cell and directly affects the performance of the power battery, including power output capability, efficiency, and thermal management.

[0086] An abnormally high internal resistance in a single battery cell may indicate aging, internal damage, or poor contact. Excessively high internal resistance can lead to a large voltage drop during high-current charging and discharging, and may cause the cell to overheat, potentially resulting in thermal runaway. Conversely, excessively low internal resistance may indicate changes in the cell's internal structure, such as diaphragm damage or internal short circuits, which can also lead to thermal runaway. In other words, both excessively low and excessively high internal resistance can pose a risk of sudden thermal runaway in a single battery cell. In practical implementation, a pre-defined normal internal resistance range—one that will not lead to a sudden risk of thermal runaway—can be established. If the internal resistance of a single battery cell deviates from this normal range, it can be preliminarily determined that the cell may be at risk of sudden thermal runaway.

[0087] For example, the state parameters of a single battery cell include: the sampling voltage, sampling current, open-circuit voltage, and polarization voltage. The open-circuit voltage (OCV) of a single battery cell refers to the voltage between the positive and negative electrodes of the cell when there is no external load (i.e., no current flowing through it). OCV is a function of the cell's state of charge (SOC) and typically varies with SOC. OCV is relatively fixed for specific types of single cells (e.g., lithium-ion batteries) and can be obtained by consulting standard OCV-SOC curves. The sampling voltage of a single battery cell refers to the voltage measured during battery operation; it is affected by internal resistance and load current. The polarization voltage of a single battery cell refers to the voltage drop caused by non-ideal factors in the electrochemical reaction process (such as concentration polarization, activation polarization, ohmic polarization, etc.). Polarization voltage is typically related to current density, electrode materials, electrolyte properties, and temperature. The determination of the single-cell internal resistance index includes the following methods S41 to S42:

[0088] S41: Construct an equivalent circuit model of a single battery cell; whereby the equivalent circuit model is used to estimate the internal resistance of a single battery cell.

[0089] Specifically, the equivalent circuit model expression for a single battery cell is as follows:

[0090]

[0091] in, Indicates the voltage being collected. Indicates open-circuit voltage. Indicates the current being collected. Indicates polarization voltage. This represents the internal resistance of a single battery cell. In practical implementation, the polarization voltage... The expression is as follows:

[0092]

[0093] It represents the concentration polarization voltage, which characterizes the polarization voltage caused by changes in the ion concentration in the electrolyte solution. It represents the activation polarization voltage, which characterizes the polarization voltage caused by the limitation of charge transfer rate. It represents the ohmic polarization voltage, which characterizes the polarization voltage generated due to the resistance encountered by electrons and ions during their movement. , and All of these can be obtained in advance through artificial equivalence. Therefore, the expression for the equivalent circuit model of a single battery cell can be further expressed as:

[0094]

[0095] S42: Substitute the acquisition voltage, acquisition current, open circuit voltage, and polarization voltage of a single cell into the above equivalent circuit model to estimate the internal resistance of the single cell, and use the estimated internal resistance as the single cell's internal resistance index.

[0096] From the equivalent circuit model of the single battery cell described above, it can be seen that when the sampling voltage, sampling current, open-circuit voltage, and polarization voltage of the single battery cell are known, the internal resistance of the single battery cell can be calculated. The calculated internal resistance The internal resistance of a single battery cell is an indicator of its internal resistance.

[0097] In step 103, the weight factor corresponding to each risk indicator represents the degree of influence of the risk indicator on the prediction result of sudden thermal runaway of the power battery. The greater the degree of influence, the larger the weight factor corresponding to the risk indicator.

[0098] In some embodiments, the weighting factors corresponding to each risk indicator can be set based on experience.

[0099] In other embodiments, the weight factors corresponding to each risk indicator can also be obtained through supervised training. The implementation method of obtaining each weight factor through supervised training is described below:

[0100] For example, the methods for determining the weighting factors corresponding to each of the multiple risk indicators include the following S51 to S53:

[0101] S51: Obtain several training samples.

[0102] Each training sample includes the state parameters of the sample battery and is labeled with a tag indicating whether the sample battery experienced abrupt thermal runaway under the aforementioned state parameters. Based on the different tags, the training samples can be divided into fault samples and normal samples. Fault samples are labeled with a first tag indicating that the sample battery experienced abrupt thermal runaway under the aforementioned state parameters. Normal samples are labeled with a second tag indicating that the sample battery did not experience abrupt thermal runaway under the aforementioned state parameters. In this embodiment, the training samples are mainly used to allow the risk prediction model to learn the changing patterns of the state parameters when the battery experiences abrupt thermal runaway; therefore, the training samples used can primarily be fault samples.

[0103] The difference between the sample power battery and the power battery to be tested in step 101 above is that the sample power battery is the power battery used in supervised training of the risk prediction model, and whether this power battery has experienced sudden thermal runaway is known. The power battery to be tested in step 101 above is a power battery for which sudden thermal runaway prediction is required; whether this power battery has experienced sudden thermal runaway is unknown and needs to be predicted based on subsequent steps.

[0104] The state parameters of the sample power battery can be historical state parameters from a large number of power batteries, representing the actual state parameters of the power battery within a historical time period. Alternatively, the state parameters of the sample power battery can also be simulated state parameters, such as the state parameters simulating a power battery experiencing sudden thermal runaway. The state parameters of the sample power battery may include, but are not limited to: the collected voltage, collected current, charge / discharge state, and collected time point of each individual cell in the sample power battery.

[0105] Each training sample is labeled with a sample label, which indicates whether the training sample has experienced a sudden thermal runaway under the recorded state parameters. The sample label can be binary (e.g., 0 indicates no sudden thermal runaway and 1 indicates sudden thermal runaway) or multi-class (e.g., different levels of sudden thermal runaway).

[0106] S52: Determine the sample features for each training sample.

[0107] The sample features include multiple risk indicators for individual cells within the sample power battery, calculated from different dimensions. These multiple risk indicators may include the aforementioned: cell voltage outlier index, differential voltage index, voltage entropy index, and cell internal resistance index. The sample features can also be understood as risk indicators for the training samples.

[0108] Specifically, for each training sample, an outlier characteristic analysis of the individual cell voltage can be performed. By collecting voltage signals, the average voltage and voltage variance of the individual cells in the sample power battery can be calculated.

[0109] By combining the average voltage value and the voltage variance value, the degree of voltage deviation of each individual cell is analyzed, which is the individual cell voltage outlier index.

[0110] Optionally, after calculating the outlier index of each cell in different sample power batteries, clustering methods can be used to statistically analyze the outlier index of each cell. For example, the cells can be grouped according to the outlier index, so that the cells within the same group are as similar as possible, while the cells in different groups are as different as possible.

[0111] Since the occurrence of abrupt thermal runaway in different sample power batteries can be determined based on sample labels, the group containing the individual cells of the sample power batteries that experienced abrupt thermal runaway can be identified from the clustered groups as outlier groups. Then, the average cell voltage outlier index and the standard deviation of the cell voltage outlier index for each cell within the outlier group are calculated. Based on the distribution of the cell voltage outlier index within the outlier group, the outlier range is determined. For example, the highest cell voltage outlier index or a certain statistic (such as the average cell voltage outlier index plus a certain multiple of the standard deviation of the cell voltage outlier index) within the outlier group can be used as the upper limit of the outlier range. The lowest cell voltage outlier index or a certain statistic (such as the average cell voltage outlier index minus a certain multiple of the standard deviation of the cell voltage outlier index) within the outlier group can be used as the lower limit of the outlier range.

[0112] This anomaly range can be used to measure whether the voltage outlier of a single battery cell is abnormal. For example, if the voltage outlier of a single battery cell falls within this anomaly range, it indicates that the voltage outlier is abnormal, and consequently, that the voltage of that single battery cell is abnormal. If the voltage outlier of a single battery cell falls outside the anomaly range, it indicates that the voltage outlier is normal, and consequently, that the voltage of that single battery cell is normal.

[0113] It should be noted that in the above example, the anomaly range is obtained based on the clustering results. Clustering analysis can reveal the commonalities and patterns hidden in the training samples, thus obtaining the anomaly range of the individual cell voltage outlier index. However, in practical implementations, the anomaly range can also be pre-defined based on experience. By determining the anomaly range, a reference can be provided for the subsequent supervised training of the risk prediction model, helping the risk prediction model to learn the characteristic patterns of individual cells whose individual cell voltage outlier index is outside the anomaly range, thereby improving the model's prediction accuracy and robustness.

[0114] For each training sample, differential voltage characteristic analysis can be performed. This involves calculating the differential voltage by collecting charging / discharging states, voltage, and current signals. A long-term monitoring mechanism is established to analyze sudden increases in differential voltage during charging or sudden decreases during discharging. The number of times the rate of change of differential voltage for each individual cell exceeds the calibrated rate of change is recorded. Individual cells with a rate exceeding the calibrated rate are marked as abnormal, indicating a potential risk of abrupt thermal runaway. This number includes the first occurrence during the charging process and the second occurrence during the discharging process.

[0115] For each training sample, voltage entropy characteristic analysis can be performed. By collecting voltage signals, the voltage entropy value, average entropy value, and standard deviation of each individual cell are calculated in a window defined on a daily basis. Based on the voltage entropy value, average entropy value, and standard deviation of each individual cell, the anomaly coefficient of each individual cell is calculated. When the anomaly coefficient is outside the calibrated anomaly coefficient range, the individual cell is marked as an anomaly, indicating that the individual cell may have a sudden thermal runaway risk.

[0116] For each training sample, the internal resistance estimation characteristics can be analyzed to estimate the internal resistance of each individual cell. Individual cells with internal resistance outside the normal range are marked as abnormal, indicating that the individual cell may have a risk of sudden thermal runaway.

[0117] By labeling anomalies, a reference can be provided for subsequent supervised training of the risk prediction model. This helps the model learn the state parameter variation patterns of individual cells with anomalies occurring more frequently than the calibration count, cells with anomaly coefficients outside the calibration anomaly coefficient range, and cells with internal resistance outside the normal range, thereby improving the model's prediction accuracy and robustness. In practice, the calibration count, normal internal resistance range, and calibration anomaly coefficient range can all be obtained by analyzing the commonalities and patterns implicit in the training samples using the aforementioned clustering analysis. This data-driven approach helps reduce subjective biases when determined manually based on experience.

[0118] S53: Based on sample features and sample labels, a supervised training risk prediction model is developed to obtain the weight factors corresponding to each of the multiple risk indicators.

[0119] The weighting factors for each of the multiple risk indicators ensure that the difference between the risk prediction model's output and the labeled samples of the input training samples is less than or equal to a preset difference. The risk prediction model outputs a prediction of whether a sudden thermal runaway has occurred. The preset difference can be set according to the desired prediction accuracy; ideally, it can be set to 0, meaning the trained weighting factors ensure that the risk prediction model's output matches the labeled samples of the input training samples.

[0120] Specifically, a gradient boosting framework (XGBOOST) can be constructed to supervise the risk prediction model. This involves training a risk prediction model using the XGBOOST algorithm to predict the risk of sudden thermal runaway in power batteries. In the implementation, sample features are input into the training risk prediction model. During training, the model parameters are continuously adjusted until the difference between the risk prediction result output by the model and the sample labels of the input training samples is less than or equal to a preset difference. The model parameters adjusted during training include the weight factors of each sample feature (i.e., multiple risk indicators). When the difference between the risk prediction result output by the risk prediction model and the sample labels of the input training samples is less than or equal to the preset difference, the weight factors in the model parameters at this point can be extracted. These extracted weight factors are then used as the weight factors for each of the multiple risk indicators. The weight factors for each of the multiple risk indicators determine the degree of influence of the input risk indicators on the risk prediction result output by the risk prediction model. The higher the weight factor, the greater the contribution of that risk indicator to the model's prediction result.

[0121] Referring to Figure 2, which is a schematic diagram of the weighting factors of various risk indicators obtained during training, an XGBOOST supervised risk prediction model is constructed. This model is trained on four sample features—cell voltage outlier index, differential voltage index, voltage entropy index, and cell internal resistance index—using sample labels. The training objective is to enable the risk prediction model to predict whether a power battery will experience sudden thermal runaway based on these sample features. During training, XGBOOST automatically adjusts the weighting factors of each sample feature to optimize prediction performance. After training, the weighting factors of each sample feature can be extracted from the risk prediction model. These weighting factors are the weighting factors of each risk indicator. By extracting weighting factors during training, the influence of each risk indicator on the model's prediction results is determined. This enhances the interpretability of the obtained weighting factors for each risk indicator and helps to better understand and manage the thermal runaway risk of power batteries.

[0122] In this embodiment, when predicting sudden thermal runaway of the power battery under test, it is unnecessary to input multiple risk indicators of individual cells within the battery into the risk prediction model. Instead, these risk indicators can be directly weighted based on the weight factors corresponding to each of the pre-trained risk indicators. This method avoids deploying a complex risk prediction model on the vehicle side, eliminates the need for complex model inference processes, simplifies the calculation process, and improves real-time performance and computational efficiency. Furthermore, the vehicle side only needs to store the weight factors corresponding to each risk indicator, without deploying the entire risk prediction model, thus saving vehicle-side storage space. By directly using the weight factors, the impact of each risk indicator on the prediction results can be more intuitively understood, enhancing the interpretability of the prediction results. If problems arise with the prediction results, the weight factors and risk indicators can be directly checked, facilitating debugging and problem localization.

[0123] For example, the above-mentioned determination of the risk score of the single cell exhibiting sudden thermal runaway based on multiple risk indicators and their respective weighting factors includes the following steps S61 to S62:

[0124] S61: Based on multiple risk indicators of a single battery cell, determine the risk score corresponding to each of the multiple risk indicators.

[0125] Specifically, there is a mapping relationship between risk indicators and their corresponding risk scores. The higher the risk of sudden thermal runaway represented by a risk indicator, the higher its corresponding risk score. This mapping relationship can be pre-defined, so after obtaining multiple risk indicators for a single battery cell, the risk score corresponding to each risk indicator can be determined by querying this mapping relationship.

[0126] A higher single-cell voltage outlier indicates a greater difference between the voltage of that single cell and the voltages of other cells in the battery pack, thus indicating a higher risk of sudden thermal runaway in that single cell. Therefore, the single-cell voltage outlier is positively correlated with its corresponding risk score.

[0127] The differential voltage index includes the first number of times the rate of change of differential voltage during charging exceeds a preset rate of change, and the second number of times the rate of change of differential voltage during discharging exceeds a preset rate of change. Understandably, a larger first or second number indicates that the internal problems of the individual cell may be continuously worsening, and the risk of thermal runaway is significantly increased. Therefore, the differential voltage index and its corresponding risk score are positively correlated, specifically: the first number and its corresponding risk score are positively correlated, and the second number and its corresponding risk score are also positively correlated, or the number of times the first and second numbers are positively correlated with their corresponding risk scores.

[0128] The voltage entropy index, also known as the anomaly coefficient, can be either positive or negative. When the voltage entropy index is positive, a larger index indicates a greater deviation of the voltage entropy value of the individual cell from the average entropy value, suggesting a potential risk of sudden thermal runaway. In this case, the voltage entropy index and its corresponding risk score are positively correlated. Conversely, when the voltage entropy index is negative, a smaller index indicates a greater deviation of the voltage entropy value of the individual cell from the average entropy value, also suggesting a potential risk of sudden thermal runaway. In this case, the voltage entropy index and its corresponding risk score are negatively correlated.

[0129] Both excessively high and excessively low internal resistance parameters in a monomer can lead to the risk of sudden thermal runaway. Therefore, the absolute value of the difference between the monomer's internal resistance parameter and a standard internal resistance parameter can be set to be positively correlated with the risk score corresponding to that internal resistance parameter. The standard internal resistance parameter can be a pre-defined parameter that will not pose a risk of sudden thermal runaway.

[0130] S62: Based on the weighting factors corresponding to each of the multiple risk indicators, the risk scores corresponding to each of the multiple risk indicators are weighted to obtain the risk score of sudden thermal runaway in a single cell.

[0131] Specifically, the risk score for each risk indicator is multiplied by its corresponding weighting factor, and then summed to obtain the risk score for a single battery cell experiencing sudden thermal runaway. For example, the risk score for each single battery cell experiencing sudden thermal runaway can be calculated using the following formula. :

[0132]

[0133] in, , , , These represent the risk scores corresponding to the individual voltage outlier index, the differential voltage index, the voltage entropy index, and the individual internal resistance index, respectively. , , , These represent the weighting factors corresponding to the individual cell voltage outlier index, the differential voltage index, the voltage entropy index, and the individual cell internal resistance index, respectively.

[0134] In step 104, the prediction result of sudden thermal runaway of the power battery is determined based on the risk score of sudden thermal runaway of each individual cell.

[0135] It is understood that the risk score of each individual cell in the power battery can be obtained through step 103 above. Based on this, the number of individual cells with a risk score greater than a preset threshold can be counted, and the preset threshold can be pre-calibrated. If the number of individual cells with a risk score greater than the preset threshold is greater than or equal to the preset number, then the predicted result of sudden thermal runaway of the power battery is determined to be that sudden thermal runaway is about to occur. The preset number can be set according to actual needs, such as one or more, and this embodiment does not specifically limit it.

[0136] In practical implementation, if the predicted result of sudden thermal runaway of the power battery indicates that it is about to occur, a risk warning can be issued to alert the driver. The risk warning can include the underlying cause that triggered it, helping the driver or relevant maintenance personnel to take appropriate corrective measures based on this cause. The underlying cause can include the identification information of individual battery cells with a risk score exceeding a preset threshold. This allows maintenance personnel to directly locate the problematic individual battery cell based on this identification information and perform repairs accordingly, minimizing the possibility of sudden thermal runaway and ensuring battery safety.

[0137] In summary, this embodiment addresses the industry pain point of inaccurate prediction of sudden thermal runaway by providing a method for predicting sudden thermal runaway in power batteries. This method covers multiple micro-short-circuit failure modes in power batteries, using individual cells within the battery as the monitoring and analysis object. This approach amplifies the minute differences and deterioration levels within individual cells, employs unsupervised learning algorithms such as clustering to discover commonalities and patterns hidden in the training samples, and incorporates expert knowledge supervision. Through machine learning algorithms training existing fault samples, it achieves risk prediction for power batteries, improving the recall and precision rates for detecting sudden thermal runaway.

[0138] Figure 3 is a schematic diagram of the structure of a thermal runaway prediction device for a power battery provided in an embodiment of this application.

[0139] For example, as shown in Figure 3, the thermal runaway prediction device 300 for a power battery includes: an acquisition module 301, used to acquire the state parameters of each individual cell in the power battery to be tested; a calculation module 302, used to calculate multiple risk indicators of the individual cell from different dimensions based on the state parameters of the individual cell; wherein the risk indicators are used to measure the risk of the individual cell experiencing sudden thermal runaway; a risk score determination module 303, used to determine the risk score of the individual cell experiencing sudden thermal runaway based on the multiple risk indicators and the weighting factors corresponding to each of the multiple risk indicators; and a prediction module 304, used to determine the prediction result of sudden thermal runaway of the power battery based on the risk score.

[0140] In some embodiments, the thermal runaway prediction device for power batteries further includes: a weight factor determination module, configured to: acquire a plurality of training samples; wherein each training sample includes state parameters of a sample power battery, each training sample is labeled with a sample label, the sample label characterizing whether the sample power battery has experienced abrupt thermal runaway under the state parameters; determine the sample features of each training sample; wherein the sample features include: multiple risk indicators of individual cells within the sample power battery calculated from different dimensions; and, based on the sample features and the sample label, supervisedly train a risk prediction model to obtain weight factors corresponding to each of the multiple risk indicators; wherein the weight factors corresponding to each of the multiple risk indicators enable the difference between the risk prediction result output by the risk prediction model and the sample label labeled on the input training sample to be less than or equal to a preset difference.

[0141] In some embodiments, the plurality of risk indicators include any combination of the following: a single cell voltage outlier index, used to characterize whether the voltage of a single cell of the power battery is abnormal; a differential voltage index, used to characterize whether the voltage change of the single cell during charging and discharging is abnormal; a voltage entropy index, used to characterize whether the consistency of the voltage distribution of the single cell of the power battery is abnormal; and a single cell internal resistance index, used to characterize whether the internal resistance of the single cell of the power battery is abnormal.

[0142] In some embodiments, the state parameters of the individual battery cell include: the sampled voltage of the individual battery cell; the calculation module 302 includes a submodule for calculating the individual battery cell voltage outlier index, used to calculate the average voltage and voltage variance of the individual battery cells in the power battery based on the sampled voltage of each individual battery cell in the power battery; and to calculate the individual battery cell voltage outlier index corresponding to each individual battery cell based on the sampled voltage of each individual battery cell in the power battery, the average voltage value and the voltage variance value.

[0143] In some embodiments, the state parameters of the individual battery cell include: the sampling voltage, sampling current, and charge / discharge state of the individual battery cell; the calculation module 302 includes a differential voltage index calculation submodule, used to calculate the differential voltage of the individual battery cell during the charging process based on the sampling voltage and sampling current of the individual battery cell during the charging process if the charge / discharge state of the individual battery cell is the charging state, and determine, within a preset monitoring period, the first number of times the rate of change of the differential voltage of the individual battery cell during the charging process exceeds a preset rate of change; if the charge / discharge state of the individual battery cell is the discharging state, calculate the differential voltage of the individual battery cell during the discharging process based on the sampling voltage and sampling current of the individual battery cell during the discharging state, and determine, within a preset monitoring period, the second number of times the rate of change of the differential voltage of the individual battery cell during the discharging process exceeds a preset rate of change; and determine the differential voltage index based on the first number and the second number.

[0144] In some embodiments, the state parameters of the individual battery cell include: the sampled voltage of the individual battery cell; the calculation module 302 includes a voltage entropy index calculation submodule, used to calculate the voltage entropy value of each individual battery cell based on the sampled voltage of each individual battery cell in the power battery within a preset time window; calculate the average entropy value and the standard deviation of the entropy value based on the voltage entropy value of each individual battery cell; calculate the anomaly coefficient based on the voltage entropy value, the average entropy value, and the standard deviation of the entropy value of each individual battery cell, and use the anomaly coefficient as the voltage entropy index.

[0145] In some embodiments, the voltage entropy index calculation submodule is specifically used to: for each of the individual cells, calculate the difference between the voltage entropy value of the individual cell and the average entropy value, and use the quotient obtained by dividing the difference by the standard deviation of the entropy value as the anomaly coefficient of the individual cell.

[0146] In some embodiments, the state parameters of the single cell include: the acquisition voltage, acquisition current, open circuit voltage, and polarization voltage of the single cell; the calculation module 302 includes a single cell internal resistance index calculation submodule, used to: construct an equivalent circuit model of the single cell; wherein, the equivalent circuit model is used to estimate the internal resistance of the single cell; substitute the acquisition voltage, acquisition current, open circuit voltage, and polarization voltage of the single cell into the equivalent circuit model to estimate the internal resistance of the single cell, and use the estimated internal resistance as the single cell internal resistance index.

[0147] In some embodiments, the risk score determination module 303 is specifically used to: determine the risk score corresponding to each of the multiple risk indicators based on the multiple risk indicators of the single cell; and perform weighted processing on the risk scores corresponding to each of the multiple risk indicators based on the weighting factors corresponding to each of the multiple risk indicators to obtain the risk score of the single cell experiencing the sudden thermal runaway.

[0148] Figure 4 is a structural schematic diagram of a vehicle provided in an embodiment of this application.

[0149] For example, as shown in FIG4, the vehicle 400 includes a memory 401 and a processor 402, wherein the memory 401 stores executable program code 4011, and the processor 402 is used to call and execute the executable program code 4011 to perform a thermal runaway prediction method for a power battery.

[0150] Furthermore, embodiments of this application also protect an apparatus that may include a memory and a processor, wherein the memory stores executable program code, and the processor is used to call and execute the executable program code to perform a thermal runaway prediction method for a power battery provided in embodiments of this application.

[0151] This embodiment can divide the device into functional modules based on the above method example. For example, each module can correspond to a separate function, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware. It should be noted that the module division in this embodiment is illustrative and only represents one logical functional division. In actual implementation, there may be other division methods.

[0152] When each functional module is divided according to its corresponding function, the device may further include an acquisition module, a calculation module, a risk score determination module, and a prediction module. It should be noted that all relevant content regarding the steps involved in the above method embodiments can be referenced to the functional descriptions of the corresponding functional modules, and will not be repeated here.

[0153] It should be understood that the device provided in this embodiment is used to execute the above-described method for predicting thermal runaway of a power battery, and therefore can achieve the same effect as the above-described implementation method.

[0154] When using an integrated unit, the device may include a processing module and a storage module. When the device is applied to a vehicle, the processing module can be used to control and manage the vehicle's movements. The storage module can be used to support the vehicle in executing relevant program code.

[0155] The processing module may be a processor or a controller, which can implement or execute various exemplary logic blocks, modules, and circuits shown in conjunction with the disclosure of this application. The processor may also be a combination of functions that implement computing capabilities, such as a combination of one or more microprocessors, a combination of digital signal processing (DSP) and a microprocessor, etc., and the storage module may be a memory.

[0156] In addition, the device provided in the embodiments of this application may specifically be a chip, component or module. The chip may include a connected processor and a memory. The memory is used to store instructions. When the processor calls and executes the instructions, the chip can execute a thermal runaway prediction method for a power battery provided in the above embodiments.

[0157] This embodiment also provides a computer-readable storage medium storing computer program code. When the computer program code is run on a computer, the computer executes the above-described related method steps to implement the thermal runaway prediction method for a power battery provided in the above embodiment.

[0158] This embodiment also provides a computer program product that, when run on a computer, causes the computer to perform the aforementioned related steps to realize the thermal runaway prediction method for a power battery provided in the above embodiment.

[0159] In this embodiment, the device, computer-readable storage medium, computer program product, or chip are all used to execute the corresponding methods provided above. Therefore, the beneficial effects they can achieve can be referred to the beneficial effects in the corresponding methods provided above, and will not be repeated here.

[0160] Through the above description of the embodiments, those skilled in the art will understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above.

[0161] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus 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 device, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between devices or units may be electrical, mechanical, or other forms.

[0162] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for predicting thermal runaway of a power battery, the method comprising: Obtain the state parameters of each individual cell in the power battery under test; Based on the state parameters of the individual battery cell, multiple risk indicators of the individual battery cell are calculated from different dimensions; wherein, the risk indicators are used to measure the risk of the individual battery cell undergoing sudden thermal runaway; Based on the multiple risk indicators and their respective weighting factors, the risk score of the single cell exhibiting sudden thermal runaway is determined. Based on the risk score, the prediction result for the sudden thermal runaway of the power battery is determined.

2. The method of claim 1, wherein, The weighting factors for each of the multiple risk indicators are determined as follows: A number of training samples are obtained; wherein each training sample includes state parameters of the sample power battery, and each training sample is labeled with a sample label, the sample label indicating whether the sample power battery has experienced abrupt thermal runaway under the state parameters; Determine the sample features for each training sample; wherein the sample features include: multiple risk indicators of individual cells within the sample power battery calculated from different dimensions; Based on the sample features and the sample labels, a supervised training risk prediction model is developed to obtain weight factors corresponding to multiple risk indicators. The weight factors corresponding to the multiple risk indicators are such that the difference between the risk prediction result output by the risk prediction model and the sample labels of the input training samples is less than or equal to a preset difference.

3. The method of claim 1 or 2, wherein, The multiple risk indicators include any combination of the following: The cell voltage outlier index is used to characterize whether the voltage of a single cell in the power battery is abnormal. The differential voltage index is used to characterize whether the voltage change of the individual battery cell is abnormal during charging and discharging. Voltage entropy index is used to characterize whether the consistency of voltage distribution in the individual cells of the power battery is abnormal. The single-cell internal resistance index is used to characterize whether the internal resistance of the single cell of the power battery is abnormal.

4. The method of claim 3, wherein, The state parameters of the individual battery cell include: the sampled voltage of the individual battery cell; the outlier index of the individual battery cell voltage is determined in the following way: Based on the collected voltage of each individual cell in the power battery, calculate the average voltage and voltage variance of each individual cell in the power battery. Based on the collected voltage of each individual cell in the power battery, the average voltage value, and the voltage variance value, the individual cell voltage outlier index corresponding to each individual cell is calculated.

5. The method of claim 3, wherein, The state parameters of the individual battery cell include: the sampling voltage, sampling current, and charge / discharge state of the individual battery cell; the differential voltage index is determined in the following manner: If the charging / discharging state of the single cell is the charging state, then based on the collected voltage and collected current of the single cell in the charging state, the differential voltage of the single cell during the charging process is calculated, and within a preset monitoring time, the first number of times the rate of change of the differential voltage of the single cell during the charging process exceeds the preset rate of change is determined. If the charging / discharging state of the single cell is the discharging state, then based on the collected voltage and collected current of the single cell in the discharging state, the differential voltage of the single cell during the discharging process is calculated, and within a preset monitoring time, the second number of times the rate of change of the differential voltage of the single cell during the discharging process exceeds the preset rate of change is determined. The differential voltage index is determined based on the first number and the second number.

6. The method of claim 3, wherein, The state parameters of the individual battery cell include: the sampled voltage of the individual battery cell; the voltage entropy index is determined in the following way: Based on the collected voltage of each individual cell in the power battery within a preset time window, calculate the voltage entropy value of each individual cell; Calculate the average entropy value and the standard deviation of the entropy value based on the voltage entropy value of each individual cell; An anomaly coefficient is calculated based on the voltage entropy value, the average entropy value, and the standard deviation of the entropy value for each individual cell, and the anomaly coefficient is used as a voltage entropy index.

7. The method of claim 6, wherein, The step of calculating an anomaly coefficient based on the voltage entropy value, the average entropy value, and the standard deviation of the entropy value for each individual battery cell, and using the anomaly coefficient as a voltage entropy index, includes: For each individual cell, the difference between the voltage entropy value of the individual cell and the average entropy value is calculated, and the quotient obtained by dividing the difference by the standard deviation of the entropy value is used as the anomaly coefficient of the individual cell.

8. The method of claim 3, the state parameter of the monobloc cell comprising: The sampling voltage, sampling current, open-circuit voltage, and polarization voltage of the individual battery cell; the internal resistance of the individual cell is determined in the following way: Construct an equivalent circuit model for a single battery cell; wherein, the equivalent circuit model is used to estimate the internal resistance of the single battery cell; The sampling voltage, sampling current, open-circuit voltage, and polarization voltage of the individual battery cell are substituted into the equivalent circuit model to estimate the internal resistance of the individual battery cell, and the estimated internal resistance is used as the individual internal resistance index of the individual battery cell.

9. The method according to any one of claims 1 to 8, wherein, The step of determining the risk score of the single cell experiencing sudden thermal runaway based on the multiple risk indicators and their respective weighting factors includes: Based on multiple risk indicators of the individual battery cell, the risk scores corresponding to each of the multiple risk indicators are determined; Based on the weighting factors corresponding to each of the multiple risk indicators, the risk scores corresponding to each of the multiple risk indicators are weighted to obtain the risk score of the single cell experiencing sudden thermal runaway.

10. The method of claim 4, wherein, The step of calculating the single-cell voltage outlier index for each single cell based on the collected voltage of each individual cell in the power battery, the average voltage value, and the voltage variance value includes: For each individual cell in the power battery, the absolute value of the difference between the collected voltage of the individual cell and the average voltage is calculated, and the quotient obtained by dividing the absolute value by the voltage variance value is used as the individual cell voltage outlier index.

11. The method according to any one of claims 1 to 10, wherein, The acquisition of state parameters of each individual cell in the power battery under test includes: Obtain the state parameters of each individual cell in the power battery under test throughout its entire life cycle; wherein, the entire life cycle refers to the time period from the time the power battery was manufactured to the current time.

12. The method according to any one of claims 1 to 10, wherein, The step of determining the prediction result of sudden thermal runaway of the power battery based on the risk score includes: The number of individual battery cells with a risk score greater than a preset threshold is counted. If the number of individual battery cells with a statistical risk score greater than a preset score threshold is greater than or equal to a preset number, then the predicted result of the sudden thermal runaway of the power battery is determined to be that sudden thermal runaway is about to occur.

13. A thermal runaway prediction device for a power battery, the device comprising: The acquisition module is used to acquire the state parameters of each individual cell in the power battery under test. The calculation module is used to calculate multiple risk indicators of the single cell from different dimensions based on the state parameters of the single cell; wherein, the risk indicators are used to measure the risk of the single cell undergoing sudden thermal runaway; The risk score determination module is used to determine the risk score of the single cell experiencing the sudden thermal runaway based on the multiple risk indicators and the weighting factors corresponding to each of the multiple risk indicators. The prediction module is used to determine the prediction result of sudden thermal runaway of the power battery based on the risk score.

14. A vehicle, the vehicle comprising: Memory, used to store executable program code; A processor for calling and running the executable program code from the memory, causing the vehicle to perform the method as described in any one of claims 1 to 12.

15. A computer-readable storage medium storing a computer program that, when executed, implements the method as claimed in any one of claims 1 to 12.

Images