A thermal runaway warning system and method for power batteries of hydrogen fuel cell vehicles

Abstract

The present invention relates to the field of power battery safety technology, and specifically to a power battery thermal runaway warning system and method for hydrogen fuel cell vehicles, the system comprising multiple acquisition modules, a temperature prediction module, a warning level classification module and an alarm module; the temperature prediction module predicts the model temperature of each single cell through a temperature prediction model based on the data of each acquisition module; the warning level classification module combines the driving temperature and model temperature of each single cell, predicts through the warning level classification model, obtains the warning level information of each single cell and sends it to the alarm module; the alarm module automatically performs the corresponding level of emergency operation according to the warning level information of each single cell. The present invention not only overcomes the defect that the current power battery thermal runaway warning system mostly uses the model temperature obtained by model prediction as the judgment basis, but also realizes more comprehensive and accurate monitoring and warning of the thermal runaway state of the power battery.

Description

Technical Field

[0001] The present invention relates to the field of power battery safety technology, and in particular to a power battery thermal runaway warning system and method for hydrogen fuel cell vehicles. Background Art

[0002] The demand for power batteries in modern society is growing, including hydrogen fuel cell vehicles, renewable energy storage and other fields. However, due to the large size and high energy storage density of power batteries, if their temperature is not effectively controlled, it may cause thermal runaway of the power batteries, causing serious safety hazards, and even threatening personnel safety and causing property losses. The traditional thermal runaway warning method mainly predicts the battery temperature through a "black box" model that lacks explanation, such as a neural network, and then warns of thermal runaway. For the purpose of control and easy implementation, the surface temperature of the battery is usually used to represent the temperature, and the dimensions considered are limited. At the same time, the current thermal runaway warning system for power batteries mostly uses the model temperature predicted by the model as the basis for judgment. For example, a multi-model fusion power battery thermal runaway warning system and method disclosed in the Chinese patent with announcement number CN118144571B, that is, a multi-model fusion technology is used to predict and warn the thermal runaway risk of power batteries, and its core is the temperature value calculated based on the model. However, for vehicles under special working conditions of actual driving, the prediction effect of this method is greatly weakened, because it is difficult for the model to fully and real-time reflect the temperature changes under these extreme conditions. This limitation significantly restricts the accuracy and reliability of the thermal runaway warning system. Therefore, finding an efficient and reliable power battery thermal runaway warning method has become an urgent need in the field of power battery safety technology. Summary of the invention

[0003] In view of the deficiencies in the above-mentioned prior art, the present invention provides a power battery thermal runaway warning system and method for hydrogen fuel cell vehicles, aiming to provide an innovative technical solution to overcome the defects in the prior art that the power battery temperature is predicted by only considering the surface temperature of the battery and the current power battery thermal runaway warning system mostly uses the model temperature obtained by model prediction as the basis for judgment, so as to achieve more comprehensive and accurate monitoring and warning of the thermal runaway state of the power battery.

[0004] The present invention introduces a temperature prediction model based on a physical information neural network and a K-means clustering algorithm, which not only realizes the accurate prediction of the model temperature of a single cell, but also conducts a comprehensive analysis of the difference between the driving temperature and the model temperature. By performing multi-level classification on the temperature deviation and change rate of each single cell at the current moment, a more comprehensive and accurate monitoring and early warning of the thermal runaway state of the power battery is achieved, thereby providing a strong guarantee for the safe operation of the vehicle.

[0005] The present invention provides a power battery thermal runaway warning system for a hydrogen fuel cell vehicle, comprising:

[0006] A battery module, in which a plurality of single cells are evenly arranged;

[0007] A plurality of voltage sensors are respectively arranged in the battery module and correspond to the single cells one by one, and are used to detect the voltage of the single cells; in addition, the voltage sensors are also arranged on the total circuit of the battery module, and are used to detect the total voltage of the battery module;

[0008] A plurality of temperature sensors, which are respectively arranged in the battery module and correspond one to one with the single cells, and are used to detect the driving temperature of the single cells in the battery module;

[0009] A current sensor, which is arranged on the total loop of the battery module and is used to obtain the total current of the battery module;

[0010] Multiple acquisition modules are connected to the battery module and the driving computer to collect battery module data and driving condition data at a sampling frequency of once every N seconds, wherein the multiple acquisition modules specifically include a first acquisition module, a second acquisition module and a third acquisition module;

[0011] A data processing module, used for processing the data collected by each collection module;

[0012] The temperature prediction module obtains the processed data of each acquisition module and predicts the model temperature of each single battery at the same time through the temperature prediction model;

[0013] The warning level classification module obtains the processed data of the first acquisition module and the model temperature of each single battery at the same time, and predicts through the warning level classification model to obtain the warning level information of each single battery, wherein the specific construction steps of the warning level classification model are as follows:

[0014] S21, obtaining second and third historical working data and performing data processing, wherein the second historical working data is the driving temperature of the single battery of the vehicle within a period of time before and after the thermal runaway occurs, and the third historical working data is the model temperature of the single battery output after the battery module and driving condition data of the same period as the second historical working data are input into the temperature prediction model;

[0015] S22, dividing the second and third historical working data after data processing into a training set and a test set according to a proportion;

[0016] S23, inputting the training set data into the clustering-based warning level classification model for training to obtain an initial warning level classification model;

[0017] S24, inputting the test set data into the initial warning level classification model for evaluation, adjusting the model parameters according to the evaluation results to obtain the warning level classification model, wherein the evaluation method uses the silhouette coefficient to evaluate the classification results of each cross-validation test data;

[0018] S25, inputting the warning level classification model into the warning level classification module for prediction of warning level information of each single battery during actual driving of the vehicle;

[0019] The alarm module obtains the warning level information of each single battery and transmits it to the vehicle infotainment system, the driver display terminal and the remote monitoring platform, and automatically performs the corresponding level of emergency operation according to the warning level information of each single battery.

[0020] Preferably, the data collected by each collection module is:

[0021] The first acquisition module is used to collect the serial number and driving temperature of each single battery in the battery module in real time;

[0022] The second acquisition module is used to collect the total voltage of the battery module, the total current of the battery module and the single cell voltage in real time;

[0023] The third acquisition module is used to obtain the mileage, motor torque, accelerator pedal travel and ambient temperature during vehicle driving.

[0024] Preferably, the temperature prediction model is a temperature prediction model based on a physical information neural network, and the specific construction steps are as follows:

[0025] S11, obtaining first historical operating data and performing data processing, wherein the first historical operating data is obtained from battery module data and driving condition data of a vehicle that has not experienced thermal runaway, including driving temperature, single cell voltage, total voltage, total current, ambient temperature, mileage, motor torque, and accelerator pedal travel data of a single cell;

[0026] S12, dividing the first historical working data after data processing into a training set and a test set according to a ratio;

[0027] S13, input the training set data into the temperature prediction model based on the physical information neural network for training, and obtain an initial temperature prediction model, wherein the temperature prediction model based on the physical information neural network is obtained by introducing the thermal balance equation of the single cell into the loss function of the model training, so that the temperature prediction model based on the physical information neural network is constrained by the laws of physics, and thus the model temperature of each single cell in the battery module can be predicted. The loss function is as follows:

[0028] ,

[0029] ,

[0030] ,

[0031] ,

[0032] In the formula, Indicates data loss. represents the equation loss; represents the initial condition loss, the loss function is related to the initial condition, that is, no current is applied to the battery cell and the driving temperature of the battery is equal to the ambient temperature; represents the total loss of the neural network, represents the number of training samples, represents the true value of the i-th sample, represents the predicted value of the i-th sample, Indicates time, and represents the weight coefficient of the relevant item, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell at time t=0, represents the ambient temperature of the i-th training sample of the single cell at time t=0; m is the number of samples added to the training at time t=0, and m<n;

[0033] S14, inputting the test set data into the initial temperature prediction model for evaluation, adjusting the model parameters according to the evaluation results to obtain the temperature prediction model, wherein the evaluation method uses the mean relative error MRE to evaluate the prediction results of each cross-validation test data;

[0034] S15, inputting the temperature prediction model into a temperature prediction module for predicting the model temperature of each single battery at the same time during the actual driving process of the vehicle.

[0035] Preferably, the clustering-based warning level classification model in step S23 is a warning level classification model based on a K-means clustering algorithm, and the specific construction steps are as follows:

[0036] S231, calculating each series of data of each single battery at the current moment;

[0037] S232, taking different series of data of each single battery at the current moment as an independent sample set, and using K-means clustering algorithm to divide all single batteries into different clusters, wherein the cluster k of the K-means clustering algorithm is 10;

[0038] S233, calculating the characteristic value of each cluster in each series of data, wherein the characteristic value of each cluster is the mean of the sample data in the cluster;

[0039] S234, determining the initial warning level corresponding to each single cell in the cluster according to the characteristic value of each cluster in each series of data and the preset warning level judgment condition;

[0040] S235. The initial warning levels of each single cell determined in different series of data are integrated, and the warning level of each single cell is determined using the "highest warning level principle", where the "highest warning level principle" is: for each single cell, compare its initial warning levels obtained in different series of data, and take the highest initial warning level as the warning level of the single cell.

[0041] Preferably, each series of data of each single cell at the current moment in step S231 includes temperature deviation and change rate, and the specific calculation steps are: based on the driving temperature and model temperature of each single cell, the temperature deviation of each single cell at the previous moment in the storage module is retrieved at the same time, and the temperature deviation and change rate of each single cell at the current moment are calculated, wherein the temperature deviation calculation formula of the single cell at the current moment is as follows:

[0042] ,

[0043] In the formula, Indicates the temperature deviation of the single cell at the current moment. Indicates the driving temperature of the single battery. Indicates the model temperature of the single cell;

[0044] The formula for calculating the rate of change of a single battery at the current moment is as follows:

[0045] ,

[0046] In the formula, R represents the change rate of the single battery at the current moment, and Indicates the temperature deviation of the single cell at time n and time n-1;

[0047] The specific steps of step S232 are: according to the temperature deviation of each single battery at the current moment, using the K-means clustering algorithm to divide all the single batteries into several different clusters; according to the change rate of each single battery at the current moment, using the K-means clustering algorithm to divide all the single batteries into several different clusters;

[0048] The specific steps of step S233 for calculating the characteristic value of each cluster in each series of data are:

[0049] In the clusters divided based on the temperature deviation of each single cell at the current moment, the temperature deviation characteristic value calculation formula of each cluster is as follows:

[0050] ;

[0051] Among the clusters divided based on the change rate of each single battery at the current moment, the calculation formula of the change rate characteristic value of each cluster is as follows:

[0052] ,

[0053] In the formula, represents the temperature deviation eigenvalue in each cluster, represents the temperature deviation of the ith single cell in each cluster, represents the rate of change eigenvalue in each cluster, represents the change rate of the i-th single cell in each cluster.

[0054] Preferably, the preset warning level judgment condition includes a warning level judgment condition based on temperature deviation and a warning level judgment condition based on change rate, wherein the warning level judgment condition based on temperature deviation is specifically:

[0055] 1a), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the first level;

[0056] 1b), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the second level;

[0057] 1c), if the eigenvalue of a cluster satisfies , then the initial warning level of all single cells in the cluster is defined as the third level;

[0058] The specific conditions for judging the warning level based on the change rate are as follows:

[0059] 2a), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the first level;

[0060] 2b) If the characteristic value of a cluster meets the preset threshold Preset Threshold , the initial warning level of all single cells in the cluster is defined as the second level;

[0061] 2c), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the third level.

[0062] Preferably, the preset threshold value in the preset warning level judgment condition is , and the corresponding thermal runaway occurrence time in each warning level is determined by experts based on experience from a large amount of test data;

[0063] In addition, when the warning level is defined as the first level, the vehicle is about to experience thermal runaway; when the warning level is defined as the second level, the vehicle may experience thermal runaway; when the warning level is defined as the third level, the vehicle's operating status is normal.

[0064] Preferably, the alarm module transmits the warning level information of each single battery through a wireless communication protocol, wherein the alarm module specifically includes:

[0065] An information transmission unit, used to transmit the warning level information of each single battery to the in-vehicle infotainment system and the driver display terminal, so as to warn the driver visually and audibly;

[0066] A remote communication unit, used to send the warning level information of each single battery to the remote monitoring platform through a cellular network or a satellite communication system for real-time monitoring and emergency response;

[0067] The emergency processing unit automatically performs the corresponding level of emergency operations based on the warning level information of each single battery to ensure driving safety, wherein the emergency operations corresponding to each warning level in the emergency processing unit are specifically:

[0068] When the warning level is the first level, the emergency processing unit automatically controls the vehicle to park in a safe area; at the same time, the backup power supply is started and switched to the emergency cooling mode to minimize the battery temperature and wait for rescue safely;

[0069] When the warning level is the second level, the emergency processing unit automatically reduces the battery output power and limits the vehicle speed; at the same time, the backup power supply is started, the auxiliary cooling system is turned on or the output of the existing cooling system is enhanced to reduce the battery temperature;

[0070] When the warning level is the third level, the emergency processing unit does not perform any emergency operation, but only records the current status for monitoring and data analysis; at the same time, it continues to monitor the battery status to ensure that it remains within the normal operating range.

[0071] Preferably, the power battery thermal runaway warning system for hydrogen fuel cell vehicles is further provided with a storage module for storing the temperature deviation of each single battery at each moment;

[0072] In addition, the data processing mentioned in the data processing module, step S11 and step S21 refers to processing operations including data cleaning, feature selection, feature conversion, data scaling and data encoding on the collected data, wherein data scaling specifically refers to normalization processing of the data;

[0073] When the emergency processing unit receives multiple emergency operation instructions of different levels at the same time, the emergency processing unit will directly execute the emergency operation instruction of the highest level.

[0074] The present invention also provides a thermal runaway warning method for a power battery of a hydrogen fuel cell vehicle, which is applied to a thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle, and comprises the following steps:

[0075] S31, after the vehicle is started, the first, second and third acquisition modules acquire real-time battery module data and driving condition data at a sampling frequency of once every N seconds and send them to the data processing module;

[0076] S32, the data processing module processes the data collected by each collection module and sends the data to the temperature prediction module and the warning level classification module;

[0077] S33, the temperature prediction module obtains the processed data of each acquisition module, predicts the model temperature of each single battery at the same time through the temperature prediction model, and sends it to the warning level classification module;

[0078] S34, the warning level classification module obtains the processed data of the first acquisition module and the model temperature of each single battery at the same time, and predicts through the warning level classification model to obtain the warning level information of each single battery;

[0079] S35, the alarm module obtains the warning level information of each single battery and transmits it to the vehicle infotainment system, the driver display terminal and the remote monitoring platform, and automatically performs the corresponding level of emergency operation according to the warning level information of each single battery.

[0080] Compared with the prior art, the present invention has the following beneficial effects:

[0081] 1. The temperature prediction model based on physical information neural network adopted in the present invention introduces the thermal balance equation of the single cell into the loss function, so that the training of the neural network is constrained by physical laws, which can more accurately reflect the temperature changes inside the battery and significantly improve the accuracy of temperature prediction.

[0082] 2. The present invention uses the K-means clustering algorithm to comprehensively analyze the difference between the driving temperature and the model temperature. By performing multi-level classification on the temperature deviation and change rate of each single battery at the current moment, the potential safety hazards in the battery status are more accurately identified. At the same time, the accurate classification of the warning level is effectively ensured, further improving the reliability and effectiveness of the warning.

[0083] 3. The present invention divides the warning level into three levels by accurately analyzing the temperature deviation and change rate. Each level is equipped with corresponding emergency treatment measures. This hierarchical warning mechanism effectively reduces false alarms and missed alarms, and significantly improves the practical efficiency and reliability of the warning system.

[0084] 4. The present invention utilizes the K-means clustering algorithm to comprehensively consider the difference between the driving temperature and the model temperature predicted by the temperature prediction model based on the physical information neural network, so as to determine the warning level of each single cell. This not only overcomes the defects that the existing technology predicts the power battery temperature by only considering the surface temperature of the battery, and the current thermal runaway warning system of the power battery mostly uses the model temperature predicted by the model as the basis for judgment, but also better reflects the actual operating status and potential risks of the single cell battery than directly clustering the temperature data using the K-means clustering algorithm alone. It can also reduce the influence of outliers on the clustering results to a certain extent, thereby improving the robustness of the algorithm. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0086] Figure 1 A schematic block diagram of a thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle;

[0087] Figure 2 Specifically construct a flow chart for the warning level classification model based on K-means clustering algorithm;

[0088] Figure 3 The present invention is a flow chart of a thermal runaway warning method for power batteries of hydrogen fuel cell vehicles. DETAILED DESCRIPTION

[0089] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0090] In the description of the present invention, it should be noted that the terms "upper", "lower", "inner", "outer", etc. indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific position, be constructed and operated in a specific position, and therefore cannot be understood as limiting the present invention; the terms "first", "second", and "third" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance. In addition, unless otherwise clearly specified and limited, the term "connection" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be an indirect connection through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0091] like Figure 1 As shown, the present invention provides a power battery thermal runaway warning system for hydrogen fuel cell vehicles, comprising:

[0092] A battery module, in which a plurality of single cells are evenly arranged;

[0093] A plurality of voltage sensors are respectively arranged in the battery module and correspond to the single cells one by one, and are used to detect the voltage of the single cells; in addition, the voltage sensors are also arranged on the total circuit of the battery module, and are used to detect the total voltage of the battery module;

[0094] A plurality of temperature sensors, which are respectively arranged in the battery module and correspond one to one with the single cells, and are used to detect the driving temperature of the single cells in the battery module;

[0095] The current sensor is arranged on the total loop of the battery module and is used to obtain the total current of the battery module.

[0096] The multiple acquisition modules involved in the present invention are connected to the battery module and the driving computer, and collect the battery module data and the driving condition data at a sampling frequency of once every N seconds.

[0097] In the embodiment of the present application, the multiple acquisition modules specifically include:

[0098] The first acquisition module is used to collect the serial number and driving temperature of each single battery in the battery module in real time;

[0099] The second acquisition module is used to collect the total voltage of the battery module, the total current of the battery module and the single cell voltage in real time;

[0100] The third acquisition module is used to obtain the mileage, motor torque, accelerator pedal travel and ambient temperature during vehicle driving.

[0101] The data processing module involved in the present invention is used for processing the data collected by each collection module.

[0102] The temperature prediction module of the present invention obtains the processed data of each acquisition module and predicts the model temperature of each single battery at the same time through the temperature prediction model.

[0103] For example, the temperature prediction module processes the data collected by each collection module at 3 minutes and 20 seconds and inputs it into the temperature prediction model, and the temperature prediction model also outputs the model temperature of the single cell at 3 minutes and 20 seconds accordingly.

[0104] In the embodiment of the present application, the temperature prediction model is a temperature prediction model based on a physical information neural network, and the specific construction steps are as follows:

[0105] S11. Acquire first historical operating data and perform data processing, wherein the first historical operating data is obtained from battery module data and driving condition data of a vehicle that has never experienced thermal runaway, including driving temperature, single cell voltage, total voltage, total current, ambient temperature, mileage, motor torque and accelerator pedal travel data of single cells.

[0106] It should be noted that, considering the impact of seasons and battery health status on battery thermal runaway, in order to make the warning results more accurate and reduce the probability of false alarms, the first historical working data is obtained from the battery module data and driving condition data of a real car for one year obtained from the National Monitoring and Management Platform for New Energy Vehicles. In addition, due to the differences in various components such as the structure and model of different battery packs and the series and parallel connection of battery cells, the time span of the first historical working data in the embodiment of the present application can be adjusted according to actual needs.

[0107] S12. Divide the first historical working data after data processing into a training set and a test set according to a proportion.

[0108] S13, inputting the training set data into the temperature prediction model based on the physical information neural network for training to obtain an initial temperature prediction model.

[0109] Preferably, the temperature prediction model based on the physical information neural network in step S13 is obtained by introducing the thermal balance equation of the single cell into the loss function of the model training, so that the temperature prediction model based on the physical information neural network is constrained by physical laws, and thus the model temperature of each single cell in the battery module can be predicted, wherein the loss function construction steps are as follows:

[0110] S131. Construct and verify the heat generation balance equation of a lithium-ion battery cell, where the equation is as follows:

[0111] ,

[0112] The verification equation is as follows:

[0113] ,

[0114] In the formula, Represents the heat capacity of a single cell. Indicates the quality of a single battery. represents the driving temperature of the ith training sample of the single battery, t represents the time, represents the total current of the ith training sample of a single cell, represents the open circuit voltage of the i-th training sample of the single cell, represents the single cell voltage of the i-th training sample of the single cell battery, represents the convection heat coefficient, Represents the surface area of ​​a single cell. represents the ambient temperature of the i-th training sample of the single cell at time t, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell, and Represents the weight coefficient of the relevant item of the i-th training sample of the single cell.

[0115] S132. Based on the verified heat generation balance equation of lithium-ion battery cells, a loss function that takes into account the physical heat generation law of the single cell battery is derived. The loss function is as follows:

[0116] ,

[0117] ,

[0118] ,

[0119] ,

[0120] In the formula, Indicates data loss, represents the equation loss; represents the initial condition loss, the loss function is related to the initial condition, that is, no current is applied to the battery cell and the driving temperature of the battery is equal to the ambient temperature; represents the total loss of the neural network, represents the number of training samples, represents the true value of the i-th sample, represents the predicted value of the i-th sample, Indicates time, and represents the weight coefficient of the relevant item, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell at time t=0, represents the ambient temperature of the i-th training sample of the single cell at time t=0; m is the number of samples added to the training at time t=0, and m<n.

[0121] It should be noted that n represents the total number of training samples of the single cell, but not all of the n samples need to conform to the initial loss. Only the m samples input at time t=0 need to be constrained, so m<n.

[0122] In an embodiment of the present application, the temperature prediction model based on the physical information neural network in step S13 defines a fully connected neural network, which includes a loss function that takes into account the physical heat generation law of the single cell battery, and uses the driving temperature, total voltage, total current, ambient temperature, mileage, motor torque and accelerator pedal travel data of the single cell battery as input items, and uses the model temperature data of the single cell battery as output item for training.

[0123] In addition, in the embodiment of the present application, the tanh activation function is used to activate the fully connected neural network of the temperature prediction model based on the physical information neural network in step S13, wherein the tanh activation function is as follows:

[0124] ,

[0125] In the formula, The first samples.

[0126] S14, input the test set data into the initial temperature prediction model for evaluation, and adjust the model parameters according to the evaluation results to obtain the temperature prediction model, wherein the evaluation method uses the mean relative error MRE to evaluate the prediction results of each cross-validation test data, and the mean relative error MRE formula is as follows:

[0127] ,

[0128] In the formula, MRE represents the mean relative error, represents the number of samples, represents the true value of the i-th sample, Represents the predicted value of the i-th sample.

[0129] It should be noted that the smaller the MRE value is, the smaller the relative error between the predicted value and the true value is, and the better the performance of the model is. In this embodiment, the final MRE of complete training and testing is the average MRE of 10 cross-validations.

[0130] S15, inputting the temperature prediction model into a temperature prediction module for predicting the model temperature of each single battery at the same time during the actual driving process of the vehicle.

[0131] The warning level classification module of the present invention obtains the processed data of the first acquisition module and the model temperature of each single battery at the same time, and predicts through the warning level classification model to obtain the warning level information of each single battery.

[0132] In this application, the specific steps of constructing the warning level classification model are as follows:

[0133] S21. Acquire the second and third historical working data and perform data processing, wherein the second historical working data is the driving temperature of the single cell of the vehicle in a period of time before and after thermal runaway occurs, and the third historical working data is the model temperature of the single cell output after the battery module and driving condition data of the same period as the second historical working data are input into the temperature prediction model.

[0134] It should be noted that in the embodiment of the present application, in the step S21, the second historical working data is the driving temperature of the single cell battery of the vehicle within 15 minutes before the occurrence of thermal runaway and within 5 minutes after the occurrence of thermal runaway. In actual application, the specific value of "a period of time before and after the occurrence of thermal runaway" can be set according to actual needs.

[0135] Preferably, the data processing module, step S11 and the data processing mentioned in step S21 refer to processing operations including data cleaning, feature selection, feature conversion, data scaling and data encoding on the collected data, wherein data scaling specifically refers to normalization of the data, and the normalization formula is as follows:

[0136] ,

[0137] In the formula, Represents the original value, Indicates the normalized value.

[0138] In this embodiment, data cleaning, feature selection, and feature conversion are to ensure that the input data has the same format and distribution as the training data so that the model can process the data correctly; data scaling is to scale the features of the input data so that they have a similar scale; and data encoding is to encode the data so that it is suitable for the requirements of the model.

[0139] S22. Divide the second and third historical working data after data processing into a training set and a test set according to a proportion.

[0140] It should be noted that in order to ensure the stability and reproducibility of the training model, a 10-fold cross-validation was performed in the present application, and the training set and test set in steps S12 and S22 accounted for 80% and 20% of the entire sample, respectively.

[0141] S23. Input the training set data into the clustering-based warning level classification model for training to obtain an initial warning level classification model.

[0142] like Figure 2 As shown, the clustering-based warning level classification model in step S23 is a warning level classification model based on the K-means clustering algorithm, and the specific construction steps are as follows:

[0143] S231, calculating each series of data of each single battery at the current moment.

[0144] In the embodiment of the present application, each series of data of each single cell at the current moment in step S231 includes a temperature deviation and a change rate, and the specific calculation steps are: based on the driving temperature and the model temperature of each single cell, the temperature deviation of each single cell at the previous moment in the storage module is retrieved at the same time, and the temperature deviation and change rate of each single cell at the current moment are calculated, wherein the temperature deviation calculation formula of the single cell at the current moment is as follows:

[0145] ,

[0146] In the formula, Indicates the temperature deviation of the single cell at the current moment. Indicates the driving temperature of the single battery. Indicates the model temperature of the single cell;

[0147] The formula for calculating the rate of change of a single battery at the current moment is as follows:

[0148] ,

[0149] In the formula, R represents the change rate of the single battery at the current moment, and Indicates the temperature deviation of the single cell at time n and time n-1.

[0150] The present invention is also provided with a storage module for storing the temperature deviation of each single battery at each moment.

[0151] It should be noted that the "current moment" and "previous moment" mentioned in this application are not the well-known moments, i.e., 15 minutes. In this application, each acquisition module collects data at a sampling frequency of once N seconds, so the current moment and the previous moment are N seconds apart.

[0152] In the embodiment of the present application, the temperature deviation of each single cell, that is, the difference between the running temperature of the single cell and the model temperature, is used to reflect the temperature deviation of the single cell in actual operation. When the running temperature of the single cell is significantly higher than the model temperature, it indicates that an abnormal exothermic reaction or poor heat dissipation may have occurred inside the single cell, which will increase the risk of thermal runaway. Therefore, the greater the temperature deviation, the more heat accumulated inside the battery, and the shorter the triggering time of thermal runaway may be.

[0153] S232, taking different series of data of each single battery at the current moment as an independent sample set, and using K-means clustering algorithm to divide all single batteries into different clusters.

[0154] In the embodiment of the present application, the step S232 is specifically as follows: according to the temperature deviation of each single cell at the current moment, all single cells are divided into several different clusters using the K-means clustering algorithm; according to the change rate of each single cell at the current moment, all single cells are also divided into several different clusters using the K-means clustering algorithm.

[0155] It should be noted that two series of data are set in the embodiment of the present application, namely, temperature deviation and change rate. Among the several clusters divided based on temperature deviation, the single cells in each cluster show similar characteristics in temperature deviation; among the several clusters divided based on change rate, the single cells in each cluster also show similar characteristics in change rate. In actual applications, users can set multiple series of data according to their needs.

[0156] In the embodiment of the present application, the cluster k of the K-means clustering algorithm in step S232 is 10. In actual use, the number of clusters can be set according to needs.

[0157] S233, calculating the eigenvalue of each cluster in each series of data, wherein the eigenvalue of each cluster is the mean of the sample data in the cluster.

[0158] In the embodiment of the present application, among the several clusters divided based on the temperature deviation of each single cell at the current moment, the temperature deviation characteristic value calculation formula of each cluster is as follows:

[0159] ;

[0160] Based on the change rate of each single battery at the current moment, the calculation formula for the change rate characteristic value of each cluster is as follows:

[0161] ,

[0162] In the formula, represents the temperature deviation eigenvalue in each cluster, represents the temperature deviation of the ith single cell in each cluster, represents the rate of change eigenvalue in each cluster, represents the change rate of the i-th single cell in each cluster.

[0163] S234: Determine the initial warning level corresponding to each single cell in the cluster according to the characteristic value of each cluster in each series of data and the preset warning level judgment condition.

[0164] In the embodiment of the present application, the preset warning level judgment condition includes a warning level judgment condition based on temperature deviation and a warning level judgment condition based on change rate, wherein the warning level judgment condition based on temperature deviation is specifically:

[0165] 1a), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the first level;

[0166] 1b), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the second level;

[0167] 1c), if the eigenvalue of a cluster satisfies , then the initial warning level of all single cells in the cluster is defined as the third level;

[0168] The specific conditions for judging the warning level based on the change rate are as follows:

[0169] 2a), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the first level;

[0170] 2b), if the eigenvalue of a cluster satisfies Preset Threshold , the initial warning level of all single cells in the cluster is defined as the second level;

[0171] 2c), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the third level.

[0172] It should be noted that the preset threshold value in the preset warning level judgment condition is The corresponding thermal runaway occurrence time in each warning level is determined by experts based on experience from a large amount of test data. It can be set according to needs in actual application.

[0173] S235. The initial warning levels of each single cell determined in different series of data are integrated, and the warning level of each single cell is determined using the "highest warning level principle", where the "highest warning level principle" is: for each single cell, compare its initial warning levels obtained in different series of data, and take the highest initial warning level as the warning level of the single cell.

[0174] It should be noted that since the clustering results of each series of data are independent, that is, each single cell may be in a different position in the temperature deviation cluster and the change rate cluster. For example, the initial warning level of a single cell is the second level in the warning level judgment condition based on temperature deviation, but the initial warning level is the first level in the warning level judgment condition based on change rate. Therefore, in the embodiment of the present application, by adopting the "highest warning level principle" to determine the warning level of each single cell, it not only effectively ensures that even if the single cell shows a low risk in one dimension, but if the risk in another dimension is high, it will also receive full attention and warning, but also ensures that the warning level of each single cell is based on a comprehensive evaluation of its overall performance, thereby improving the prediction accuracy.

[0175] Preferably, when the warning level is defined as the first level, the vehicle is about to experience thermal runaway; when the warning level is defined as the second level, the vehicle may experience thermal runaway; when the warning level is defined as the third level, the vehicle is operating normally.

[0176] S24, inputting the test set data into the initial warning level classification model for evaluation, and adjusting the model parameters according to the evaluation results to obtain the warning level classification model.

[0177] Preferably, in step S24, the silhouette coefficient is used to evaluate the classification effect of each cross-validation test data, and the silhouette coefficient calculation method is as follows:

[0178] ,

[0179] In the formula, is the silhouette coefficient, For sample i The average distance to other samples in the same cluster, For sample i The average distance to all samples of any other cluster j The minimum value in .

[0180] It should be noted that the value range of the silhouette coefficient is between [-1, 1]. The larger the value, the better the clustering effect, and a negative value indicates a very poor clustering effect.

[0181] S25, inputting the warning level classification model into the warning level classification module for predicting the warning level information of each single battery during the actual driving process of the vehicle.

[0182] It should be noted that K-means clustering is an unsupervised learning method. It does not require pre-labeled data for training. It groups data according to its intrinsic characteristics. On this basis, this embodiment combines the actual grouping situation of the model and expert experience, and selects the best classification model through multiple groupings to achieve the best warning classification effect of the warning model.

[0183] The alarm module of the present invention obtains the warning level information of each single battery and transmits it to the vehicle infotainment system, the driver display terminal and the remote monitoring platform, and automatically performs the corresponding level of emergency operation according to the warning level information of each single battery, specifically including:

[0184] An information transmission unit, used to transmit the warning level information of each single battery to the in-vehicle infotainment system and the driver display terminal, so as to warn the driver visually and audibly;

[0185] A remote communication unit, used to send the warning level information of each single battery to the remote monitoring platform through a cellular network or a satellite communication system for real-time monitoring and emergency response;

[0186] The emergency processing unit automatically performs emergency operations of corresponding levels based on the warning level information of each single battery to ensure driving safety.

[0187] Preferably, the alarm module transmits the warning level information of each single battery via a wireless communication protocol.

[0188] In an embodiment of the present application, the alarm module transmits the warning level information of each single battery to the vehicle infotainment system, the driver display terminal and the remote monitoring platform through a wireless communication protocol, ensuring that the warning signal can be conveyed to the driver and remote monitoring personnel in a timely and comprehensive manner, thereby achieving rapid response and emergency handling.

[0189] In this application, the emergency operations corresponding to each warning level in the emergency processing unit are specifically:

[0190] When the warning level is the first level, the emergency processing unit automatically controls the vehicle to park in a safe area; at the same time, the backup power supply is started and switched to the emergency cooling mode to minimize the battery temperature and wait for rescue safely;

[0191] When the warning level is the second level, the emergency processing unit automatically reduces the battery output power and limits the vehicle speed; at the same time, the backup power supply is started, the auxiliary cooling system is turned on or the output of the existing cooling system is enhanced to reduce the battery temperature;

[0192] When the warning level is the third level, the emergency processing unit does not perform any emergency operation, but only records the current status for monitoring and data analysis; at the same time, it continues to monitor the battery status to ensure that it remains within the normal operating range.

[0193] Preferably, when the emergency processing unit receives multiple emergency operation instructions of different levels at the same time, the emergency processing unit will directly execute the emergency operation instruction of the highest level.

[0194] For example, when the emergency processing unit receives emergency operation instructions of the second level and the third level at the same time, the emergency processing unit will directly execute the emergency operation instruction of the second level.

[0195] like Figure 3 As shown, the present invention also provides a thermal runaway warning method for a power battery of a hydrogen fuel cell vehicle, which is applied to a thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle, and includes the following steps:

[0196] S31, after the vehicle is started, the first, second and third acquisition modules acquire real-time battery module data and driving condition data at a sampling frequency of once every N seconds and send them to the data processing module;

[0197] S32, the data processing module processes the data collected by each collection module and sends the data to the temperature prediction module and the warning level classification module;

[0198] S33, the temperature prediction module obtains the processed data of each acquisition module, predicts the model temperature of each single battery at the same time through the temperature prediction model, and sends it to the warning level classification module;

[0199] S34, the warning level classification module obtains the processed data of the first acquisition module and the model temperature of each single battery at the same time, and predicts through the warning level classification model to obtain the warning level information of each single battery;

[0200] S35, the alarm module obtains the warning level information of each single battery and transmits it to the vehicle infotainment system, the driver display terminal and the remote monitoring platform, and automatically performs the corresponding level of emergency operation according to the warning level information of each single battery.

[0201] Finally, it should be noted that the above is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle, characterized in that: include: A battery module, in which a plurality of single cells are evenly arranged; A plurality of voltage sensors are respectively arranged in the battery module and correspond to the single cells one by one, and are used to detect the voltage of the single cells; in addition, the voltage sensors are also arranged on the total circuit of the battery module, and are used to detect the total voltage of the battery module; A plurality of temperature sensors, which are respectively arranged in the battery module and correspond one to one with the single cells, and are used to detect the driving temperature of the single cells in the battery module; A current sensor, which is arranged on the total loop of the battery module and is used to obtain the total current of the battery module; Multiple acquisition modules are connected to the battery module and the driving computer to collect battery module data and driving condition data at a sampling frequency of once every N seconds, wherein the multiple acquisition modules specifically include a first acquisition module, a second acquisition module and a third acquisition module; A data processing module, used for processing the data collected by each collection module; The temperature prediction module obtains the processed data of each acquisition module and predicts the model temperature of each single battery at the same time through the temperature prediction model; The warning level classification module obtains the processed data of the first acquisition module and the model temperature of each single battery at the same time, and predicts through the warning level classification model to obtain the warning level information of each single battery, wherein the specific construction steps of the warning level classification model are as follows: S21, obtaining second and third historical working data and performing data processing, wherein the second historical working data is the driving temperature of the single battery of the vehicle within a period of time before and after the thermal runaway occurs, and the third historical working data is the model temperature of the single battery output after the battery module and driving condition data of the same period as the second historical working data are input into the temperature prediction model; S22, dividing the second and third historical working data after data processing into a training set and a test set according to a proportion; S23, inputting the training set data into the clustering-based warning level classification model for training to obtain an initial warning level classification model; S24, inputting the test set data into the initial warning level classification model for evaluation, adjusting the model parameters according to the evaluation results to obtain the warning level classification model, wherein the evaluation method uses the silhouette coefficient to evaluate the classification results of each cross-validation test data; S25, inputting the warning level classification model into the warning level classification module for prediction of warning level information of each single battery during actual driving of the vehicle; The alarm module obtains the warning level information of each single battery and transmits it to the vehicle infotainment system, the driver display terminal and the remote monitoring platform, and automatically performs the corresponding level of emergency operation according to the warning level information of each single battery.

2. A power battery thermal runaway warning system for a hydrogen fuel cell vehicle according to claim 1, characterized in that: The specific data collected by each acquisition module are: The first acquisition module is used to collect the serial number and driving temperature of each single battery in the battery module in real time; The second acquisition module is used to collect the total voltage of the battery module, the total current of the battery module and the single cell voltage in real time; The third acquisition module is used to obtain the mileage, motor torque, accelerator pedal travel and ambient temperature during vehicle driving.

3. A thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle according to claim 2, characterized in that: The temperature prediction model is a temperature prediction model based on a physical information neural network, and the specific construction steps are as follows: S11, obtaining first historical operating data and performing data processing, wherein the first historical operating data is obtained from battery module data and driving condition data of a vehicle that has not experienced thermal runaway, including driving temperature, single cell voltage, total voltage, total current, ambient temperature, mileage, motor torque, and accelerator pedal travel data of a single cell; S12, dividing the first historical working data after data processing into a training set and a test set according to a ratio; S13, input the training set data into the temperature prediction model based on the physical information neural network for training, and obtain an initial temperature prediction model, wherein the temperature prediction model based on the physical information neural network is obtained by introducing the thermal balance equation of the single cell into the loss function of the model training, so that the temperature prediction model based on the physical information neural network is constrained by the laws of physics, and thus the model temperature of each single cell in the battery module can be predicted. The loss function is as follows: ， ， ， ， In the formula, Indicates data loss. represents the equation loss; represents the initial condition loss, the loss function is related to the initial condition, that is, no current is applied to the battery cell and the driving temperature of the battery is equal to the ambient temperature; represents the total loss of the neural network, represents the number of training samples, represents the true value of the i-th sample, represents the predicted value of the i-th sample, Indicates time, and represents the weight coefficient of the relevant item, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell, represents the physical equation used to constrain the neural network for the i-th training sample of the single cell at time t=0, represents the ambient temperature of the i-th training sample of the single cell at time t=0; m is the number of samples added to the training at time t=0, and m<n; S14, inputting the test set data into the initial temperature prediction model for evaluation, adjusting the model parameters according to the evaluation results to obtain the temperature prediction model, wherein the evaluation method uses the mean relative error MRE to evaluate the prediction results of each cross-validation test data; S15, inputting the temperature prediction model into a temperature prediction module for predicting the model temperature of each single battery at the same time during the actual driving process of the vehicle.

4. A power battery thermal runaway warning system for a hydrogen fuel cell vehicle according to claim 3, characterized in that: The clustering-based warning level classification model in step S23 is a warning level classification model based on the K-means clustering algorithm, and the specific construction steps are as follows: S231, calculating each series of data of each single battery at the current moment; S232, taking different series of data of each single battery at the current moment as an independent sample set, and using K-means clustering algorithm to divide all single batteries into different clusters, wherein the cluster k of the K-means clustering algorithm is 10; S233, calculating the characteristic value of each cluster in each series of data, wherein the characteristic value of each cluster is the mean of the sample data in the cluster; S234, determining the initial warning level corresponding to each single cell in the cluster according to the characteristic value of each cluster in each series of data and the preset warning level judgment condition; S235. Based on the initial warning levels of each single cell determined in different series of data, the warning level of each single cell is determined by adopting the "highest warning level principle", wherein the "highest warning level principle" is: for each single cell, compare its initial warning levels obtained in different series of data, and take the highest initial warning level as the warning level of the single cell.

5. A thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle according to claim 4, characterized in that: In step S231, each series of data of each single battery at the current moment includes temperature deviation and change rate. The specific calculation steps are: based on the driving temperature and model temperature of each single battery, the temperature deviation of each single battery at the previous moment in the storage module is retrieved at the same time, and the temperature deviation and change rate of each single battery at the current moment are calculated. The temperature deviation calculation formula of the single battery at the current moment is as follows: ， In the formula, Indicates the temperature deviation of the single cell at the current moment. Indicates the driving temperature of the single battery. Indicates the model temperature of the single cell; The formula for calculating the rate of change of a single battery at the current moment is as follows: ， In the formula, R represents the change rate of the single battery at the current moment, and Indicates the temperature deviation of the single cell at time n and time n-1; The specific steps of step S232 are: according to the temperature deviation of each single battery at the current moment, using the K-means clustering algorithm to divide all the single batteries into several different clusters; according to the change rate of each single battery at the current moment, using the K-means clustering algorithm to divide all the single batteries into several different clusters; The specific steps of step S233 for calculating the characteristic value of each cluster in each series of data are: In the clusters divided based on the temperature deviation of each single cell at the current moment, the temperature deviation characteristic value calculation formula of each cluster is as follows: ， Among the clusters divided based on the change rate of each single battery at the current moment, the calculation formula of the change rate characteristic value of each cluster is as follows: ， In the formula, represents the temperature deviation eigenvalue in each cluster, represents the temperature deviation of the ith single cell in each cluster, represents the rate of change eigenvalue in each cluster, represents the change rate of the i-th single cell in each cluster.

6. A thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle according to claim 5, characterized in that: The preset warning level judgment condition includes a warning level judgment condition based on temperature deviation and a warning level judgment condition based on change rate, wherein the warning level judgment condition based on temperature deviation is specifically: 1a), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the first level; 1b), if the eigenvalue of a cluster satisfies , then the initial warning level of all single cells in the cluster is defined as the second level; 1c), if the eigenvalue of a cluster satisfies , then the initial warning level of all single cells in the cluster is defined as the third level; The specific conditions for judging the warning level based on the change rate are as follows: 2a), if the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the first level; 2b) If the characteristic value of a cluster meets the preset threshold Preset Threshold , then the initial warning level of all single cells in the cluster is defined as the second level; 2c) If the eigenvalue of a cluster satisfies , the initial warning level of all single cells in the cluster is defined as the third level.

7. A thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle according to claim 6, characterized in that: The preset threshold value in the preset warning level judgment condition , and the corresponding thermal runaway occurrence time in each warning level is determined by experts based on experience from a large amount of test data; In addition, when the warning level is defined as the first level, the vehicle is about to experience thermal runaway; when the warning level is defined as the second level, the vehicle may experience thermal runaway; when the warning level is defined as the third level, the vehicle's operating status is normal.

8. A power battery thermal runaway warning system for a hydrogen fuel cell vehicle according to claim 7, characterized in that: The alarm module transmits the warning level information of each single battery through a wireless communication protocol, wherein the alarm module specifically includes: An information transmission unit, used to transmit the warning level information of each single battery to the in-vehicle infotainment system and the driver display terminal, so as to warn the driver visually and audibly; A remote communication unit, used to send the warning level information of each single battery to the remote monitoring platform through a cellular network or a satellite communication system for real-time monitoring and emergency response; The emergency processing unit automatically performs the corresponding level of emergency operations based on the warning level information of each single battery to ensure driving safety, wherein the emergency operations corresponding to each warning level in the emergency processing unit are specifically: When the warning level is the first level, the emergency processing unit automatically controls the vehicle to park in a safe area; at the same time, the backup power supply is started and switched to the emergency cooling mode to minimize the battery temperature and wait for rescue safely; When the warning level is the second level, the emergency processing unit automatically reduces the battery output power and limits the vehicle speed; at the same time, the backup power supply is started, the auxiliary cooling system is turned on or the output of the existing cooling system is enhanced to reduce the battery temperature; When the warning level is the third level, the emergency processing unit does not perform any emergency operation, but only records the current status for monitoring and data analysis; at the same time, it continues to monitor the battery status to ensure that it remains within the normal operating range.

9. A power battery thermal runaway warning system for a hydrogen fuel cell vehicle according to claim 8, characterized in that: The power battery thermal runaway warning system for hydrogen fuel cell vehicles is also provided with a storage module for storing the temperature deviation of each single battery at each time; In addition, the data processing mentioned in the data processing module, step S11 and step S21 refers to processing operations including data cleaning, feature selection, feature conversion, data scaling and data encoding on the collected data, wherein data scaling specifically refers to normalization processing of the data; When the emergency processing unit receives multiple emergency operation instructions of different levels at the same time, the emergency processing unit will directly execute the emergency operation instruction of the highest level.

10. A thermal runaway warning method for a power battery of a hydrogen fuel cell vehicle, applied to a thermal runaway warning system for a power battery of a hydrogen fuel cell vehicle as claimed in any one of claims 1 to 9, characterized in that: The following steps are involved: S31, after the vehicle is started, the first, second and third acquisition modules acquire real-time battery module data and driving condition data at a sampling frequency of once every N seconds and send them to the data processing module; S32, the data processing module processes the data collected by each collection module and sends the data to the temperature prediction module and the warning level classification module; S33, the temperature prediction module obtains the processed data of each acquisition module, predicts the model temperature of each single battery at the same time through the temperature prediction model, and sends it to the warning level classification module; S34, the warning level classification module obtains the processed data of the first acquisition module and the model temperature of each single battery at the same time, and predicts through the warning level classification model to obtain the warning level information of each single battery; S35, the alarm module obtains the warning level information of each single battery and transmits it to the vehicle infotainment system, the driver display terminal and the remote monitoring platform, and automatically performs the corresponding level of emergency operation according to the warning level information of each single battery.

Images