An electric vehicle battery thermal runaway alarm system and method

Abstract

The present invention discloses a battery thermal runaway alarm system and method for electric vehicles. The battery warning system monitors the temperature of battery cells, pressure within the battery pack, busbar current, and busbar voltage in different operating modes of the vehicle. The smoke sensor monitors the smoke concentration within the battery pack. The collected data is transmitted to the battery management system, which transmits the smoke concentration data to the vehicle controller. The vehicle controller analyzes the smoke concentration data and determines whether to issue a first thermal runaway warning instruction. The battery management system transmits the data collected by the battery warning system to the vehicle terminal. The vehicle terminal transmits the data to the cloud server. The cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and a trained thermal runaway warning model. The battery management system determines whether to issue a thermal runaway alarm based on the two runaway warning instructions. This improves alarm reliability and user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of pure electric vehicle battery monitoring, and in particular to an electric vehicle battery thermal runaway alarm system and method. Background Art

[0002] The statements in this section merely mention background art related to the present invention and do not necessarily constitute prior art.

[0003] As the number of electric vehicles on the market increases, the safety issues of electric vehicles are becoming increasingly serious.

[0004] The root cause of fires and explosions in electric vehicles is thermal runaway caused by battery failure. This problem not only hinders the widespread adoption and commercial application of electric vehicles but also poses a serious threat to public life, property, and traffic safety. Currently, when an electric vehicle battery experiences thermal runaway, it will generate alarms for high temperature, temperature differential, and pressure differential. However, there is currently no accurate, rapid, and personalized battery thermal runaway alarm system.

[0005] China's utility model patent CN215752027U - Power battery thermal runaway warning system and vehicle, collects the voltage and temperature of single cells through voltage sensors and temperature sensors to achieve power battery thermal runaway warning. However, there are problems such as single alarm conditions, slow alarm data processing speed, low alarm sensitivity, poor reliability, and false alarms.

[0006] Chinese utility model patent CN212313296U - A battery pack thermal runaway warning system, which provides thermal runaway warning by collecting smoke concentration and single cell information inside the battery pack. Although it can solve the problem of false battery alarms, it has problems such as single data, unreliable analysis results, and time-consuming analysis process, which leaves users with too little time to stay away from dangerous battery packs.

[0007] The inventors found that the existing battery thermal runaway warning system not only has the problems of single data collection, unreliable alarm results, low alarm sensitivity, and long calculation time for the alarm analysis process, which means very little time is gained for users, but also has the problem of excessive alarm sensitivity. For example, the data generated by different models of electric vehicle battery packs are different, and the data generated by the same model of electric vehicle battery packs with different years of use are also different. If only the data collected at the current time point is considered, the alarm sensitivity may be too high, and an alarm may be issued when it should not be, which will seriously affect the user experience. Summary of the Invention

[0008] To address the deficiencies of the prior art, the present invention provides an electric vehicle battery thermal runaway alarm system and method. The system provides a rapid early warning of thermal runaway failures in the electric vehicle battery system, with the battery management system issuing an alarm signal 30 minutes before thermal runaway in the passenger compartment, buying more time for users. The system uses a variety of data for analysis to improve the reliability of the alarm. At the same time, the system ensures that the sensitivity of the thermal runaway alarm is not too high, thereby improving the user experience.

[0009] In a first aspect, the present invention provides a method for alarming thermal runaway of an electric vehicle battery;

[0010] An electric vehicle battery thermal runaway alarm method, applied to a battery management system (BMS), includes:

[0011] Transmitting the smoke concentration data to the vehicle control unit (VCU); enabling the VCU to analyze the smoke concentration data and determine whether to issue a first thermal runaway warning instruction; and receiving the analysis result of the VCU;

[0012] The data collected by the battery warning system in different operating modes is transmitted to the cloud server through the vehicle terminal T-BOX (TelematicsBOX); the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; and the analysis results of the cloud server are received through the vehicle terminal.

[0013] Based on the two analysis results, determine whether to issue a thermal runaway alarm.

[0014] In a second aspect, the present invention provides a battery management system BMS;

[0015] A battery management system BMS, comprising:

[0016] The first transmission module is configured to: transmit smoke concentration data to a vehicle control unit (VCU); enable the VCU to analyze the smoke concentration data and determine whether to issue a first thermal runaway warning instruction; and receive analysis results from the VCU;

[0017] The second transmission module is configured to transmit data collected by the battery warning system in different operating modes to the cloud server via the vehicle-mounted terminal T-BOX (Telematics Box); so that the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle-mounted terminal and the trained thermal runaway warning model; and receives the analysis results of the cloud server via the vehicle-mounted terminal;

[0018] The output module is configured to determine whether to issue a thermal runaway alarm according to the two analysis results.

[0019] In a third aspect, the present invention provides an electric vehicle battery thermal runaway alarm method;

[0020] An electric vehicle battery thermal runaway alarm method, applied to a vehicle controller, includes:

[0021] Obtain smoke concentration data;

[0022] Analyze smoke concentration data to determine whether to issue the first thermal runaway warning instruction;

[0023] Send the first analysis result to the battery management system BMS;

[0024] The auxiliary battery management system combines the first analysis result and the second analysis result to determine whether to issue a thermal runaway alarm;

[0025] Among them, the second analysis result is sent to the battery management system BMS by the cloud server through the vehicle terminal; the second analysis result is determined by the cloud server whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; the data uploaded by the vehicle terminal is data collected by different sensors in different working modes collected by the battery warning system.

[0026] In a fourth aspect, the present invention provides a vehicle controller;

[0027] Vehicle controller, including:

[0028] A first acquisition module is configured to: acquire smoke concentration data;

[0029] A first analysis module is configured to: analyze the smoke concentration data to determine whether to issue a first thermal runaway warning instruction;

[0030] A first sending module is configured to: send the first analysis result to the battery management system BMS;

[0031] a first determination module configured to: determine whether to issue a thermal runaway alarm by combining the first analysis result and the second analysis result with the auxiliary battery management system;

[0032] Among them, the second analysis result is sent to the battery management system BMS by the cloud server through the vehicle terminal; the second analysis result is determined by the cloud server whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; the data uploaded by the vehicle terminal is data collected by different sensors in different working modes collected by the battery warning system.

[0033] In a fifth aspect, the present invention provides an electric vehicle battery thermal runaway alarm method;

[0034] An electric vehicle battery thermal runaway alarm method, applied to a cloud server, comprising:

[0035] Acquire data in different working modes and modalities;

[0036] Determining a second analysis result based on data from different operating modes and modalities and a trained thermal runaway warning model; the second analysis result indicates whether to issue a second thermal runaway warning instruction;

[0037] Sending the determined second analysis result to the battery management system BMS via the vehicle terminal;

[0038] The auxiliary battery management system combines the first analysis result and the second analysis result to determine whether to issue a thermal runaway alarm;

[0039] Among them, data of different working modes and different modalities are collected by different sensors and uploaded to the cloud server in sequence through the battery management system BMS and the on-board terminal; the first analysis result is obtained by analyzing the smoke concentration data by the vehicle controller VCU; the first analysis result refers to whether the first thermal runaway warning instruction is issued; the smoke concentration data is collected by the smoke sensor and uploaded to the vehicle controller VCU through the battery management system BMS.

[0040] In a sixth aspect, the present invention provides a cloud server;

[0041] A cloud server, comprising:

[0042] A second acquisition module is configured to: acquire data in different working modes and different modalities;

[0043] a second analysis module configured to determine a second analysis result based on data from different operating modes and modalities and a trained thermal runaway warning model; the second analysis result indicating whether to issue a second thermal runaway warning instruction;

[0044] A second sending module is configured to send the determined second analysis result to the battery management system BMS via the vehicle terminal;

[0045] a second determination module configured to: determine whether to issue a thermal runaway alarm by combining the first analysis result and the second analysis result with the auxiliary battery management system;

[0046] Among them, data of different working modes and different modalities are collected by different sensors and uploaded to the cloud server in sequence through the battery management system BMS and the on-board terminal; the first analysis result is obtained by analyzing the smoke concentration data by the vehicle controller VCU; the first analysis result refers to whether the first thermal runaway warning instruction is issued; the smoke concentration data is collected by the smoke sensor and uploaded to the vehicle controller VCU through the battery management system BMS.

[0047] In a seventh aspect, the present invention also provides an electric vehicle battery thermal runaway alarm system;

[0048] An electric vehicle battery thermal runaway alarm system includes: a battery management system BMS, a vehicle controller VCU and a cloud server;

[0049] The battery management system (BMS) transmits smoke concentration data to the vehicle control unit (VCU). The VCU analyzes the smoke concentration data and determines whether to issue the first thermal runaway warning instruction.

[0050] The battery management system (BMS) transmits data collected by the battery warning system regarding different operating modes to the vehicle-mounted terminal (T-BOX). The vehicle-mounted terminal then transmits the data to the cloud server. The cloud server then determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle-mounted terminal and the trained thermal runaway warning model.

[0051] The battery management system BMS determines whether to issue a thermal runaway alarm based on the two runaway warning instructions.

[0052] In an eighth aspect, the present invention provides an electronic device;

[0053] An electronic device, comprising:

[0054] a memory for non-transitory storage of computer-readable instructions; and

[0055] a processor for executing said computer-readable instructions,

[0056] When the computer-readable instructions are executed by the processor, the method described in the first, third or fifth aspect above is executed.

[0057] In a ninth aspect, the present invention further provides a storage medium that non-temporarily stores computer-readable instructions, wherein when the non-temporary computer-readable instructions are executed by a computer, the instructions of the method described in the first, third or fifth aspect are executed.

[0058] In a tenth aspect, the present invention further provides a computer program product, comprising a computer program, which is used to implement the method described in the first, third or fifth aspect when running on one or more processors.

[0059] Compared with the prior art, the present invention has the following beneficial effects:

[0060] (1) Through the collaborative warning of the battery warning system and the smoke sensor, a rapid warning of thermal runaway failure of the electric vehicle battery system is achieved. The battery management system sends an alarm signal 30 minutes before thermal runaway in the passenger compartment, buying more time for users;

[0061] (2) Use temperature, pressure, current, voltage, and smoke data for analysis to improve the reliability and sensitivity of the alarm;

[0062] (3) By combining historical data of batteries of the same model and batch with the battery mechanism model and artificial intelligence early warning model, the sensitivity of the thermal runaway alarm is ensured not to be too high, thereby improving the user experience.

[0063] (4) The present invention combines a battery early warning system and a smoke sensor to issue a thermal runaway sound and light alarm at least 30 minutes before thermal runaway occurs, thus solving the problem of issuing a thermal runaway alarm signal 5 minutes before thermal runaway occurs in the passenger compartment, as required by the safety requirements for power batteries used in electric vehicles. This expands the functionality of electric vehicles and improves their safety.

[0064] (5) It replaces the conventional battery thermal runaway alarm (there is no special battery thermal runaway alarm system). After monitoring the alarm information such as battery high temperature alarm, battery temperature difference alarm, battery pressure difference alarm, etc., it reports the battery thermal runaway. The present invention has high accuracy and strong reliability. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The accompanying drawings, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0066] Figure 1 This is a schematic diagram of the internal connections of the electric vehicle battery thermal runaway alarm system according to the first embodiment of the present application;

[0067] Figure 2 This is a schematic diagram of the thermal runaway alarm strategy of Example 1 of the present application;

[0068] Figure 3 This is a schematic diagram of a battery warning system according to Example 1 of the present application;

[0069] Figure 4 This is a flow chart of the interaction between the BMS and the smoke sensor in the non-sleep state of Example 1 of the present application;

[0070] Figure 5 This is a flow chart of the interaction between the BMS and the smoke sensor in the dormant state according to the first embodiment of the present application;

[0071] Among them, 001 represents different vehicle modes. Driving mode, slow charging mode, and fast charging mode correspond to the BMS non-sleep state, and static power-down mode corresponds to the BMS sleep state.

[0072] 002 represents the BMS ignition signal, including K15 ignition, slow charge ignition, and fast charge ignition signals;

[0073] 003 means BMS enters working state after being awakened;

[0074] 004 indicates thermal runaway alarm monitoring information, including warning information monitored by the battery warning system and smoke alarm information;

[0075] 005 means that the thermal runaway alarm system in the BMS integrates the monitoring information of the battery early warning system and the smoke alarm information to confirm the thermal runaway alarm and upload it to the vehicle network through the vehicle CAN;

[0076] 006 indicates that the vehicle VCU will issue an audible and visual alarm after receiving the battery thermal runaway alarm information. DETAILED DESCRIPTION

[0077] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those skilled in the art to which the present invention belongs.

[0078] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0079] In the absence of conflict, the embodiments of the present invention and the features thereof may be combined with each other.

[0080] All data in this embodiment is obtained in compliance with laws and regulations and based on the consent of the user, and is used legally.

[0081] Example 1: This embodiment provides an electric vehicle battery thermal runaway alarm method;

[0082] like Figure 1 As shown, a thermal runaway alarm method for an electric vehicle battery is applied to a battery management system (BMS), comprising:

[0083] Transmitting the smoke concentration data to the vehicle control unit (VCU); enabling the VCU to analyze the smoke concentration data and determine whether to issue a first thermal runaway warning instruction; and receiving the analysis result of the VCU;

[0084] The data collected by the battery warning system in different operating modes is transmitted to the cloud server through the vehicle terminal T-BOX (TelematicsBOX); the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; and the analysis results of the cloud server are received through the vehicle terminal.

[0085] Based on the two analysis results, determine whether to issue a thermal runaway alarm.

[0086] Furthermore, the battery warning system and smoke sensor are both installed in the battery pack;

[0087] The battery early warning system is responsible for monitoring the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the vehicle under different operating modes;

[0088] The smoke sensor is responsible for monitoring the smoke concentration inside the battery pack of the vehicle under different operating modes;

[0089] The battery early warning system and the smoke sensor transmit the collected data to the battery management system BMS (Battery Management System).

[0090] By collecting multiple data points, the above-mentioned technical solution solves the existing technical problem of single-source data collection and inaccurate analysis results, achieving a more reliable result from the analysis of multiple data points. Furthermore, the vehicle controller analyzes smoke concentration data to issue a first thermal runaway warning instruction; the cloud server analyzes data collected by the battery warning system to issue a second thermal runaway warning instruction. By comprehensively considering these two thermal runaway warning instructions, a final thermal runaway alarm is issued, avoiding false thermal runaway alarms, improving the reliability and stability of the thermal runaway alarm, and controlling the sensitivity of the thermal runaway alarm from being too high or too low.

[0091] When the vehicle is in driving, AC charging, DC charging, or stationary power-down mode, the battery warning system monitors the temperature, current, voltage, and pressure of the battery system throughout its life cycle. It also monitors the abnormal smoke concentration alarm information in real time in conjunction with the battery management system (BMS) of the battery system assembly controller. It issues a thermal runaway alarm signal at least 30 minutes before thermal runaway occurs, and the vehicle issues an audible and visual alarm to alert passengers. Figure 2 shown.

[0092] Furthermore, the battery early warning system includes: a temperature sensor, a pressure sensor, a current sensor and a voltage sensor.

[0093] Furthermore, the vehicle controller VCU analyzes the smoke concentration data to determine whether to issue a first thermal runaway warning instruction; specifically, the steps include:

[0094] (11) When the vehicle is not in sleep mode (driving, AC charging, DC charging), the smoke sensor operates in continuous working mode, monitoring the smoke concentration, pulse width modulation (PWM) frequency and duty cycle in the battery pack; the battery management system (BMS) reads the PWM frequency and duty cycle monitored by the smoke sensor in real time;

[0095] (12) The smoke sensor determines whether the smoke concentration in the battery pack exceeds the set threshold. If so, the PWM frequency and duty cycle corresponding to the time when the concentration exceeds the set threshold are uploaded to the battery management system BMS;

[0096] (13) The battery management system BMS determines whether the received frequency is equal to the first frequency threshold, and if not, enters (14); if yes, enters (15);

[0097] (14) determining whether the received frequency is equal to a second frequency threshold; if so, reporting a fault in the smoke sensor itself based on the PWM frequency and duty cycle; if not, continuing to determine whether the received frequency is equal to the second frequency threshold;

[0098] (15) Determine whether the current duty cycle is greater than a first duty cycle threshold value. If so, determine to issue a first thermal runaway warning instruction, send the first thermal runaway warning instruction to the vehicle controller VCU, and the vehicle controller VCU issues a high-voltage disconnect instruction to actively stop battery charging or battery discharging; if not, report that the smoke sensor itself has failed based on the PWM frequency and duty cycle.

[0099] Furthermore, the vehicle controller VCU analyzes the smoke concentration data to determine whether to issue a first thermal runaway warning instruction; specifically, the steps include:

[0100] (21) When the vehicle is in dormant (parked) state, the smoke sensor operates in a low-power mode and monitors the smoke concentration in the battery pack. The smoke sensor determines whether the smoke concentration in the battery pack exceeds a set threshold. If so, the smoke sensor outputs a high level to the battery management system BMS.

[0101] (22) The battery management system BMS continuously determines whether the wake-up signal Wake-up is at a high level; if not, it determines that the battery management system BMS is in the initialization stage and the smoke sensor has not sent a wake-up signal; if so, the battery management system BMS sends a signal Request from BMS to the smoke sensor at a high level, and at the same time, wakes up the vehicle controller VCU through the network;

[0102] (23) The smoke sensor receives the Request from BMS as a high level and sends the PWM frequency and duty cycle to the battery management system BMS;

[0103] (24) The battery management system BMS receives the PWM frequency and duty cycle, and determines whether the frequency is the first threshold. If not, it proceeds to (25); if so, it proceeds to (26);

[0104] (25) Determine whether the frequency is a second threshold value. If so, report that the smoke sensor itself has failed based on the PWM frequency and duty cycle. If not, continue to determine whether the frequency is a second threshold value.

[0105] (26) Determine whether the duty cycle reaches the duty cycle alarm threshold. If so, issue the first thermal runaway warning instruction; if not, report that the smoke sensor itself has failed based on the PWM frequency and duty cycle; set Request from BMS to a low level, and the smoke sensor enters a low power consumption mode.

[0106] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0107] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0108] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0109] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0110] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then the internal resistance of this single cell is determined to be abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0111] Furthermore, the input values of the self-discharge model are the battery state of charge (SOC) and the battery cell voltage. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extract the data fragments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process; 2) Single charge pressure difference calculation: calculate SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC respectively. ≤20% of the mean pressure difference ΔV after denoising of the data segments; 3) based on the time series query, the pressure difference of the charging SOC interval segments corresponding to the time interval requirements (5d~6d, 10d~11d, 15d~16d) is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated; 4) If the set number of continuous charging segments does not trigger the current model, it is determined to be normal, otherwise it is determined to be a battery system self-discharge abnormality; if the set number of continuous charging segments does not meet the calculation requirements, no output is given; the output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0112] Furthermore, the internal short circuit model has as input the bus current, the battery state of charge (SOC), and the battery cell voltage, and outputs the cell code and the internal short circuit abnormality score.

[0113] Furthermore, the input values of the temperature rise model are the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the data of the charge / discharge conditions and 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0114] Furthermore, the trained thermal runaway warning model; the training process includes:

[0115] Build a convolutional neural network;

[0116] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0117] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0118] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0119] The beneficial effect of the above technical solution is that the model is trained using the historical monitoring data of batteries after electric vehicles of the same model and batch come off the production line. The resulting thermal runaway warning model is targeted. For the battery to be monitored, compared with the training data of batteries of different models and batches, the model trained with the historical performance of its "brother" batteries can more accurately predict the performance of the current batteries of the same model and batch.

[0120] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0121] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0122] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0123] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0124] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0125] When the vehicle is in driving mode, the BMS is awakened by the K15 vehicle ignition hard-wire signal. The BMS works in non-sleep mode and interacts with the smoke sensor in real time. If an abnormal smoke concentration alarm is detected and the battery warning system issues a thermal runaway warning, the thermal runaway alarm is confirmed.

[0126] When the vehicle is in slow charging mode, the BMS is awakened by the slow charging ignition hard-wire signal. The BMS works in non-sleep mode and interacts with the smoke sensor in real time. If an abnormal smoke concentration alarm is detected and the battery warning system issues a thermal runaway warning, the thermal runaway alarm is confirmed.

[0127] When the vehicle is in fast charging mode, the BMS is awakened by the fast charging ignition hard-wire signal. The BMS works in non-sleep mode and interacts with the smoke sensor in real time. If an abnormal smoke concentration alarm is detected and the battery warning system issues a thermal runaway warning, the thermal runaway alarm is confirmed.

[0128] When the vehicle is in static power-off mode, the BMS works in sleep mode. The BMS is awakened by the hard-wired signal of the smoke sensor and interacts with the smoke sensor in real time. If an abnormal smoke concentration alarm is detected and the battery warning system issues a thermal runaway warning, the thermal runaway alarm is confirmed.

[0129] like Figure 2As shown, the electric vehicle battery management system BMS, battery warning system, smoke sensor and vehicle controller VCU exchange information through the vehicle CAN and TCP / IP network, and complete the battery thermal runaway alarm in the vehicle sleep state and the vehicle non-sleep state respectively. The battery warning module is a special battery warning module on the enterprise data platform. It is based on the high temperature / temperature difference characteristics, pressure characteristics, power characteristics and overvoltage / pressure difference characteristics collected by the temperature sensor, current sensor, voltage sensor and pressure sensor that monitor the entire life cycle of the battery. After comprehensive processing by the battery mechanism model (internal resistance model, self-discharge model, internal short circuit model, temperature rise model, etc.) and the AI warning model (battery high sensitivity characteristics and optimal algorithm model, etc.), combined with the smoke concentration characteristics collected by the smoke sensor, it realizes early warning and rapid alarm of electric vehicle battery thermal runaway. See the attached Figure 3 ;

[0130] When the vehicle is in a non-dormant state (driving, AC charging, DC charging), the smoke sensor works in a continuous working mode and sends identification messages to the BMS to confirm the necessary information of the smoke sensor, battery warning system and BMS. The thermal runaway alarm signal is issued at least 30 minutes before the thermal runaway occurs. The BMS completes the vehicle high voltage disconnection and the vehicle sound and light alarm by sending identification messages between the vehicle controller VCU and BMS. See the attached Figure 4 ;

[0131] When the vehicle is in sleep (parked) state, the smoke sensor works in low power mode and sends identification messages to the BMS to confirm the necessary information of the smoke sensor, battery warning system and BMS. A thermal runaway alarm signal is issued at least 30 minutes before the thermal runaway occurs. After confirming the thermal runaway fault, the BMS wakes up the VCU through the network. The VCU and BMS send identification messages to each other to complete the vehicle high voltage disconnection and vehicle sound and light alarm. See attached Figure 5 .

[0132] The voltage difference indicates the difference between the highest cell voltage and the lowest cell voltage collected by the battery collection unit;

[0133] The temperature difference represents the difference between the highest cell temperature and the lowest cell temperature collected by the battery collection unit;

[0134] Temperature rise indicates an abnormal value where the cell temperature collected by the battery collection unit increases abnormally and rapidly;

[0135] High temperature means that the single cell temperature collected by the battery collection unit exceeds the upper limit of the battery's safe operating temperature;

[0136] Overvoltage means that the single cell voltage collected by the battery collection unit exceeds the lower limit of the single cell voltage.

[0137] Based on the high temperature / pressure differential characteristics, pressure characteristics, power characteristics and overvoltage / pressure differential characteristics collected by temperature sensors, current sensors, voltage sensors and pressure sensors that monitor the entire life cycle of the battery, after comprehensive processing by battery mechanism models (internal resistance model, self-discharge model, internal short circuit model, temperature rise model, etc.) and AI early warning models (battery high sensitivity characteristics and optimal algorithm model, etc.), combined with the smoke concentration characteristics collected by the smoke sensor, early warning and rapid alarm of thermal runaway of electric vehicle batteries are achieved. See attached. Figure 3 :

[0138] After the vehicle rolls off the production line, the BMS uses the temperature, current, voltage, and pressure sensors within the battery system to monitor high-temperature / temperature differential characteristics, pressure characteristics, power characteristics, and overvoltage / pressure differential characteristics in real time. The data is then uploaded to the onboard terminal TBOX via the vehicle's CAN. The onboard terminal TBOX then uploads the data to the enterprise data platform via the TCP / IP network.

[0139] As a dedicated battery warning module on the enterprise data platform, the battery early warning system receives and stores data on high temperature / temperature difference characteristics, pressure characteristics, power characteristics, and overvoltage / pressure difference characteristics throughout the battery's life cycle, forming a battery life cycle database.

[0140] Based on the battery life cycle database, a battery early warning system is formed through comprehensive processing of data cleaning, battery mechanism models (internal resistance model, self-discharge model, internal short circuit model, temperature rise model, etc.), AI early warning models (battery high sensitivity characteristics and optimal algorithm model, etc.), model training and model tuning. Test data is added to determine whether the early warning system meets the target. If it does not meet the requirements, the model optimization continues. If it does, a battery system early warning is issued.

[0141] The battery warning system starts to issue battery failure warnings, including thermal runaway warnings. The battery warning system transmits the thermal runaway warnings to TBOX via the TCP / IP network, and TBOX sends them to the BMS via the vehicle CAN.

[0142] The electric vehicle's BMS, battery warning system, smoke sensor, and vehicle controller (VCU) exchange information through the vehicle's CAN and TCP / IP networks, providing battery thermal runaway alarms in both the vehicle's dormant and non-dormant states. At least 30 minutes before thermal runaway occurs, the vehicle's audible and visual alarms warn passengers to leave the vehicle immediately.

[0143] When the vehicle is in a non-sleep state (driving mode, slow charging mode, fast charging mode), the BMS works in a non-sleep mode, and the BMS, battery warning system and smoke sensor exchange real-time information. If the smoke sensor does not alarm, after receiving the thermal runaway warning information monitored by the battery warning system, the BMS will actively limit the charge / discharge power to half and request to start cooling. If the smoke sensor alarms and the battery warning system reports a thermal runaway warning, the thermal runaway warning system confirms the thermal runaway alarm and reports the thermal runaway alarm to the vehicle controller VCU through the vehicle CAN. The VCU will sound and light the vehicle to remind passengers to leave the vehicle immediately.

[0144] When the vehicle is in sleep mode (idle and powered off), the BMS works in non-sleep mode. The smoke sensor detects abnormal smoke concentration and immediately wakes up the BMS. The BMS, battery early warning system and smoke sensor exchange information in real time. The thermal runaway alarm system monitors and confirms the thermal runaway alarm in real time, and reports the thermal runaway alarm to the vehicle controller VCU through the vehicle CAN. The VCU issues a vehicle-wide sound and light alarm to remind passengers to stay away from the vehicle immediately.

[0145] When the vehicle is not in sleep mode (driving mode, slow charging mode, fast charging mode), the smoke sensor works in continuous working mode and sends identification messages to the BMS to confirm the necessary information of the smoke sensor, battery warning system and BMS. A thermal runaway alarm signal is issued 30 minutes before the thermal runaway occurs. The vehicle controller VCU and BMS send identification messages to each other to complete the vehicle high voltage disconnection and vehicle sound and light alarm. See attached Figure 4 :

[0146] When the vehicle is in non-sleep mode (driving mode, slow charging mode, fast charging mode), the BMS and VCU are both working in non-sleep mode. If the BMS sends a high level signal "Request from BMS" to the smoke sensor, the smoke sensor will work in continuous working mode.

[0147] The smoke sensor monitoring period is 1S and the PWM wave frequency it emits is the threshold F1Hz and the duty cycle is the threshold D1%;

[0148] The BMS reads the PWM frequency and duty cycle of the smoke sensor and continuously determines whether the PWM wave frequency is at the threshold value F1Hz. If not, it further determines whether the PWM wave frequency is at the threshold value F2Hz. If so, it reports a fault in the smoke sensor itself based on the PWM wave duty cycle.

[0149] The smoke sensor detects that the smoke concentration exceeds the threshold C1μg / m 3 , the PWM wave frequency becomes the threshold F1Hz and the duty cycle is the threshold D2%;

[0150] The BMS receives the PWM wave frequency sent by the smoke sensor as the threshold F1Hz, and further determines that the PWM wave duty cycle is the threshold D2%. If not, it reports the smoke sensor fault based on the PWM wave duty cycle. If so, it further integrates the thermal runaway warning information from the battery warning system, determines the thermal runaway alarm and reports the thermal runaway alarm to the vehicle controller VCU through the vehicle CAN, and actively disconnects the high voltage to stop the vehicle from charging / discharging.

[0151] Upon receiving the BMS thermal runaway alarm signal, the VCU issues a high-voltage disconnect command for the entire vehicle and uses sound and light alarms to remind passengers to leave the vehicle immediately.

[0152] When the vehicle is in sleep mode (power off), the smoke sensor works in low power mode and sends identification messages to the BMS to confirm the necessary information of the smoke sensor, battery warning system and BMS. A thermal runaway alarm signal is issued 30 minutes before the thermal runaway occurs. After confirming the thermal runaway fault, the BMS wakes up the VCU through the network. The VCU and BMS send identification messages to each other to complete the high voltage disconnection of the vehicle and the sound and light alarm of the vehicle. See the attached Figure 4 :

[0153] When the vehicle is in sleep mode (power off), the BMS and VCU are both in sleep mode, and the smoke sensor is in low power mode.

[0154] The smoke sensor monitoring cycle is 12S and does not provide PWM wave;

[0155] The smoke sensor detects that the smoke concentration exceeds the C threshold of 1μg / m 3 , the smoke sensor sends a Wake_up signal to the BMS as high level;

[0156] The BMS reads the wake-up signal Wake_up from the smoke sensor and continuously checks whether the wake-up signal Wake_up is at a high level. If not, it determines that the BMS is in the initialization stage and the smoke sensor has not issued a wake-up signal Wake_up. If so, the BMS sends a signal Request from BMS at a high level to the smoke sensor and wakes up the vehicle controller VCU through the network.

[0157] When the smoke sensor receives the Request from BMS, which is high, the frequency of the PWM wave it sends becomes the threshold F1 Hz and the duty cycle becomes the threshold D2%.

[0158] The BMS receives the PWM wave frequency sent by the smoke sensor as the threshold F1Hz, and further determines that the PWM wave duty cycle is the threshold D2%. If not, the smoke sensor itself is reported to have a fault based on the PWM wave duty cycle, and the Request from BMS is sent as a high level. When the smoke sensor receives the Request from BMS as a high level, it works in low power mode and sends a Wake_up as a low level. When the BMS receives the Wake_up as a low level, it enters sleep mode. If so, the thermal runaway warning information from the battery warning system is further integrated to determine the thermal runaway alarm and report the thermal runaway alarm to the vehicle controller VCU through the vehicle CAN.

[0159] The VCU receives the BMS thermal runaway alarm signal and issues an audible and visual alarm throughout the vehicle to remind passengers to leave the vehicle immediately.

[0160] Embodiment 2: This embodiment provides a battery management system BMS;

[0161] A battery management system BMS, comprising:

[0162] The first transmission module is configured to: transmit smoke concentration data to a vehicle control unit (VCU); enable the VCU to analyze the smoke concentration data and determine whether to issue a first thermal runaway warning instruction; and receive analysis results from the VCU;

[0163] The second transmission module is configured to transmit data collected by the battery warning system in different operating modes to the cloud server via the vehicle-mounted terminal T-BOX (Telematics Box); so that the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle-mounted terminal and the trained thermal runaway warning model; and receives the analysis results of the cloud server via the vehicle-mounted terminal;

[0164] The output module is configured to determine whether to issue a thermal runaway alarm according to the two analysis results.

[0165] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0166] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0167] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0168] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0169] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then it is determined that the internal resistance of this single cell is abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0170] Furthermore, the self-discharge model takes the battery state of charge (SOC) and battery cell voltage as input values. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extracting data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process. 2) Single charging pressure difference calculation: calculating the denoised pressure difference mean ΔV of the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively. 3) Based on a time series query, the pressure difference of the corresponding charging SOC interval segments that meet the set time interval requirements is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated. 4) If the set number of continuous charging times does not trigger the current model, it is determined to be normal, otherwise it is determined to be abnormal self-discharge of the battery system. If the set number of continuous charging segments does not meet the calculation requirements, no output is given. The output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0171] Furthermore, the temperature rise model has input values of the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the charge / discharge conditions and the data of 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0172] Furthermore, the trained thermal runaway warning model; the training process includes:

[0173] Build a convolutional neural network;

[0174] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0175] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0176] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0177] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0178] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0179] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0180] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0181] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0182] The implementation details of each step in the second embodiment are consistent with those in the first embodiment.

[0183] Example 3: This embodiment provides an electric vehicle battery thermal runaway alarm method;

[0184] An electric vehicle battery thermal runaway alarm method, applied to a vehicle controller, includes:

[0185] Obtain smoke concentration data;

[0186] Analyze smoke concentration data to determine whether to issue the first thermal runaway warning instruction;

[0187] Send the first analysis result to the battery management system BMS;

[0188] The auxiliary battery management system combines the first analysis result and the second analysis result to determine whether to issue a thermal runaway alarm;

[0189] Among them, the second analysis result is sent to the battery management system BMS by the cloud server through the vehicle terminal; the second analysis result is determined by the cloud server whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; the data uploaded by the vehicle terminal is data collected by different sensors in different working modes collected by the battery warning system.

[0190] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0191] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0192] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0193] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0194] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then it is determined that the internal resistance of this single cell is abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0195] Furthermore, the self-discharge model takes the battery state of charge (SOC) and battery cell voltage as input values. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extracting data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process. 2) Single charging pressure difference calculation: calculating the denoised pressure difference mean ΔV of the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively. 3) Based on a time series query, the pressure difference of the corresponding charging SOC interval segments that meet the set time interval requirements is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated. 4) If the set number of continuous charging times does not trigger the current model, it is determined to be normal, otherwise it is determined to be abnormal self-discharge of the battery system. If the set number of continuous charging segments does not meet the calculation requirements, no output is given. The output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0196] Furthermore, the temperature rise model has input values of the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the charge / discharge conditions and the data of 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0197] Furthermore, the trained thermal runaway warning model; the training process includes:

[0198] Build a convolutional neural network;

[0199] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0200] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0201] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0202] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0203] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0204] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0205] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0206] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0207] The implementation details of each step in the third embodiment are consistent with those in the first embodiment.

[0208] Embodiment 4: This embodiment provides a vehicle controller;

[0209] Vehicle controller, including:

[0210] A first acquisition module is configured to: acquire smoke concentration data;

[0211] A first analysis module is configured to: analyze the smoke concentration data to determine whether to issue a first thermal runaway warning instruction;

[0212] A first sending module is configured to: send the first analysis result to the battery management system BMS;

[0213] a first determination module configured to: determine whether to issue a thermal runaway alarm by combining the first analysis result and the second analysis result with the auxiliary battery management system;

[0214] Among them, the second analysis result is sent to the battery management system BMS by the cloud server through the vehicle terminal; the second analysis result is determined by the cloud server whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; the data uploaded by the vehicle terminal is data collected by different sensors in different working modes collected by the battery warning system.

[0215] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0216] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0217] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0218] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0219] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then it is determined that the internal resistance of this single cell is abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0220] Furthermore, the self-discharge model takes the battery state of charge (SOC) and battery cell voltage as input values. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extracting data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process. 2) Single charging pressure difference calculation: calculating the denoised pressure difference mean ΔV of the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively. 3) Based on a time series query, the pressure difference of the corresponding charging SOC interval segments that meet the set time interval requirements is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated. 4) If the set number of continuous charging times does not trigger the current model, it is determined to be normal, otherwise it is determined to be abnormal self-discharge of the battery system. If the set number of continuous charging segments does not meet the calculation requirements, no output is given. The output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0221] Furthermore, the temperature rise model has input values of the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the charge / discharge conditions and the data of 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0222] Furthermore, the trained thermal runaway warning model; the training process includes:

[0223] Build a convolutional neural network;

[0224] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0225] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0226] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0227] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0228] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0229] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0230] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0231] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0232] The implementation details of each step in the fourth embodiment are consistent with those in the first embodiment.

[0233] Example 5. This embodiment provides an electric vehicle battery thermal runaway alarm method;

[0234] An electric vehicle battery thermal runaway alarm method, applied to a cloud server, comprising:

[0235] Acquire data in different working modes and modalities;

[0236] Determining a second analysis result based on data from different operating modes and modalities and a trained thermal runaway warning model; the second analysis result indicates whether to issue a second thermal runaway warning instruction;

[0237] Sending the determined second analysis result to the battery management system BMS via the vehicle terminal;

[0238] The auxiliary battery management system combines the first analysis result and the second analysis result to determine whether to issue a thermal runaway alarm;

[0239] Among them, data of different working modes and different modalities are collected by different sensors and uploaded to the cloud server in sequence through the battery management system BMS and the on-board terminal; the first analysis result is obtained by analyzing the smoke concentration data by the vehicle controller VCU; the first analysis result refers to whether the first thermal runaway warning instruction is issued; the smoke concentration data is collected by the smoke sensor and uploaded to the vehicle controller VCU through the battery management system BMS.

[0240] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0241] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0242] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0243] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0244] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then it is determined that the internal resistance of this single cell is abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0245] Furthermore, the self-discharge model takes the battery state of charge (SOC) and battery cell voltage as input values. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extracting data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process. 2) Single charging pressure difference calculation: calculating the denoised pressure difference mean ΔV of the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively. 3) Based on a time series query, the pressure difference of the corresponding charging SOC interval segments that meet the set time interval requirements is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated. 4) If the set number of continuous charging times does not trigger the current model, it is determined to be normal, otherwise it is determined to be abnormal self-discharge of the battery system. If the set number of continuous charging segments does not meet the calculation requirements, no output is given. The output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0246] Furthermore, the temperature rise model has input values of the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the charge / discharge conditions and the data of 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0247] Furthermore, the trained thermal runaway warning model; the training process includes:

[0248] Build a convolutional neural network;

[0249] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0250] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0251] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0252] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0253] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0254] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0255] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0256] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0257] The implementation details of each step in the fifth embodiment are consistent with those in the first embodiment.

[0258] Embodiment 6: This embodiment provides a cloud server;

[0259] A cloud server, comprising:

[0260] A second acquisition module is configured to: acquire data in different working modes and different modalities;

[0261] a second analysis module configured to determine a second analysis result based on data from different operating modes and modalities and a trained thermal runaway warning model; the second analysis result indicating whether to issue a second thermal runaway warning instruction;

[0262] A second sending module is configured to send the determined second analysis result to the battery management system BMS via the vehicle terminal;

[0263] a second determination module configured to: determine whether to issue a thermal runaway alarm by combining the first analysis result and the second analysis result with the auxiliary battery management system;

[0264] Among them, data of different working modes and different modalities are collected by different sensors and uploaded to the cloud server in sequence through the battery management system BMS and the on-board terminal; the first analysis result is obtained by analyzing the smoke concentration data by the vehicle controller VCU; the first analysis result refers to whether the first thermal runaway warning instruction is issued; the smoke concentration data is collected by the smoke sensor and uploaded to the vehicle controller VCU through the battery management system BMS.

[0265] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0266] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0267] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0268] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0269] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then it is determined that the internal resistance of this single cell is abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0270] Furthermore, the self-discharge model takes the battery state of charge (SOC) and battery cell voltage as input values. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extracting data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process. 2) Single charging pressure difference calculation: calculating the denoised pressure difference mean ΔV of the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively. 3) Based on a time series query, the pressure difference of the corresponding charging SOC interval segments that meet the set time interval requirements is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated. 4) If the set number of continuous charging times does not trigger the current model, it is determined to be normal, otherwise it is determined to be abnormal self-discharge of the battery system. If the set number of continuous charging segments does not meet the calculation requirements, no output is given. The output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0271] Furthermore, the temperature rise model has input values of the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the charge / discharge conditions and the data of 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0272] Furthermore, the trained thermal runaway warning model; the training process includes:

[0273] Build a convolutional neural network;

[0274] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0275] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0276] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0277] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0278] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0279] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0280] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0281] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0282] The implementation details of each step in the sixth embodiment are consistent with those in the first embodiment.

[0283] Embodiment 7: This embodiment also provides an electric vehicle battery thermal runaway alarm system;

[0284] An electric vehicle battery thermal runaway alarm system includes: a battery management system BMS, a vehicle controller VCU and a cloud server;

[0285] The battery management system (BMS) transmits smoke concentration data to the vehicle control unit (VCU). The VCU analyzes the smoke concentration data and determines whether to issue the first thermal runaway warning instruction.

[0286] The battery management system (BMS) transmits data collected by the battery warning system regarding different operating modes to the vehicle-mounted terminal (T-BOX). The vehicle-mounted terminal then transmits the data to the cloud server. The cloud server then determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle-mounted terminal and the trained thermal runaway warning model.

[0287] The battery management system BMS determines whether to issue a thermal runaway alarm based on the two runaway warning instructions.

[0288] Furthermore, the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including:

[0289] The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal;

[0290] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model, and a temperature rise model;

[0291] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

[0292] Furthermore, the input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage; the working principle of the internal resistance model is to process data under discharge conditions and with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change when the current frame voltage is subtracted from the previous frame voltage when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then it is determined that the internal resistance of this single cell is abnormal; the output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

[0293] Furthermore, the self-discharge model takes the battery state of charge (SOC) and battery cell voltage as input values. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will lead to abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extracting data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process. 2) Single charging pressure difference calculation: calculating the denoised pressure difference mean ΔV of the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively. 3) Based on a time series query, the pressure difference of the corresponding charging SOC interval segments that meet the set time interval requirements is calculated, and the daily pressure difference change rate of the corresponding time interval is calculated. 4) If the set number of continuous charging times does not trigger the current model, it is determined to be normal, otherwise it is determined to be abnormal self-discharge of the battery system. If the set number of continuous charging segments does not meet the calculation requirements, no output is given. The output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

[0294] Furthermore, the temperature rise model has input values of the battery cell temperature, the battery state of charge (SOC) and the battery clock signal; the working principle of the temperature rise model is: data processing is performed on the charge / discharge conditions and the data of 10%≤soc≤100%, and if the battery cell temperature rise rate is 2°C / S for three consecutive times, the temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, the temperature rise anomaly coefficient c is obtained; the temperature rise anomaly coefficients a, b and c are comprehensively judged, and the battery cell code and temperature rise anomaly level of the battery cell with temperature rise anomaly are output; the output value of the temperature rise model is the battery cell code and the temperature rise anomaly score.

[0295] Furthermore, the trained thermal runaway warning model; the training process includes:

[0296] Build a convolutional neural network;

[0297] Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage;

[0298] The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model;

[0299] The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network to train the network. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped to obtain the trained convolutional neural network, that is, the trained thermal runaway warning model.

[0300] Furthermore, the step of determining whether to issue a thermal runaway alarm based on the two analysis results specifically includes:

[0301] If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued;

[0302] If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued;

[0303] If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated.

[0304] If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

[0305] The implementation details of each step in the seventh embodiment are consistent with those in the first embodiment.

[0306] Embodiment 8: This embodiment provides an electronic device;

[0307] An electronic device, comprising:

[0308] a memory for non-transitory storage of computer-readable instructions; and

[0309] a processor for executing said computer-readable instructions,

[0310] When the computer-readable instructions are executed by the processor, the method described in the above-mentioned embodiment one, three or five is executed.

[0311] The implementation details of each step in the eighth embodiment are consistent with those in the first embodiment.

[0312] Embodiment 9: This embodiment further provides a storage medium that non-temporarily stores computer-readable instructions, wherein when the non-temporary computer-readable instructions are executed by a computer, the instructions of the method described in embodiment 1, 3 or 5 are executed.

[0313] The implementation details of each step in the ninth embodiment are consistent with those in the first embodiment.

[0314] Embodiment 10: This embodiment further provides a computer program product, including a computer program, which is used to implement the method described in the above embodiment 1, 3 or 5 when running on one or more processors.

[0315] The implementation details of each step in the tenth embodiment are consistent with those in the first embodiment.

[0316] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.

Claims

1. A method for alarming thermal runaway of an electric vehicle battery, characterized in that: Applied to battery management system BMS, including: Transmitting the smoke concentration data to the vehicle controller VCU; enabling the vehicle controller VCU to analyze the smoke concentration data and determine whether to issue a first thermal runaway warning instruction; and receiving the analysis result of the vehicle controller VCU; The data collected by the battery warning system in different working modes is transmitted to the cloud server through the vehicle terminal T-BOX. The cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model. The analysis results of the cloud server are received through the vehicle terminal. Based on the two analysis results, determine whether to issue a thermal runaway alarm; The trained thermal runaway warning model; the training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. The determination of whether to issue a thermal runaway alarm is based on the two analysis results; specifically includes: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

2. The electric vehicle battery thermal runaway alarm method according to claim 1, characterized in that: The cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model; specifically including: The cloud server transmits the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage uploaded by the vehicle terminal; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, input into the trained thermal runaway warning model, and output the classification result of whether thermal runaway occurs.

3. The electric vehicle battery thermal runaway alarm method according to claim 2, wherein: The input values of the internal resistance model are the bus current, the battery state of charge (SOC) and the battery cell voltage. The working principle of the internal resistance model is to process data under discharge conditions with 30% ≤ SOC ≤ 80%. If the code of the single cell with the largest voltage change (the current frame voltage minus the previous frame voltage) when the bus current is maximum is the same as the code of the single cell with the largest voltage change when the bus current is minimum, then the internal resistance of this single cell is determined to be abnormal. The output value of the internal resistance model is the single cell code and the internal resistance abnormality score.

4. The electric vehicle battery thermal runaway alarm method according to claim 2, wherein: The input values of the self-discharge model are the battery state of charge (SOC) and the battery cell voltage. The working principle of the self-discharge model is that abnormal self-discharge of a single cell will cause abnormal pressure difference at the system level, and the self-discharge characteristics are identified by the stability of the pressure difference in the time dimension. 1) Data source: extract the data segments corresponding to SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% during each continuous charging process; 2) Single charge voltage difference calculation: Calculate the denoised voltage difference mean ΔV for the SOC ≥ 98%, 50% ≤ SOC ≤ 60%, and 15% ≤ SOC ≤ 20% data segments respectively; 3) Based on the time series query, the voltage difference of the corresponding charging SOC interval segment that meets the set time interval requirements is calculated, and the daily voltage difference change rate of the corresponding time interval is calculated; 4) If the set number of continuous charging segments does not trigger the current model, it is determined to be normal; otherwise, it is determined to be a battery system self-discharge abnormality; if the set number of continuous charging segments does not meet the calculation requirements, no output is given; the output value of the self-discharge model is the battery system self-discharge abnormality and the self-discharge abnormality score.

5. The electric vehicle battery thermal runaway alarm method according to claim 2, wherein: The temperature rise model, whose input values are the battery cell temperature, the battery state of charge (SOC), and the battery clock signal, operates as follows: data is processed for charge / discharge conditions with 10% ≤ SOC ≤ 100%, and if the battery cell temperature rises at a rate of 2°C / s for three consecutive times, a temperature rise anomaly coefficient a is obtained; if the difference between the maximum temperature of the battery cell and the minimum temperature of the battery cell rises continuously and exceeds a certain threshold, a temperature rise anomaly coefficient b is obtained; if the maximum temperature of the battery cell rises continuously and exceeds a certain threshold, a temperature rise anomaly coefficient c is obtained; Comprehensively judge the temperature rise abnormality coefficients a, b, and c, and output the single cell code and temperature rise abnormality level of the temperature rise abnormality; the output value of the temperature rise model is the single cell code and the temperature rise abnormality score.

6. A battery management system (BMS), comprising: The first transmission module is configured to transmit smoke concentration data to the vehicle controller VCU; The vehicle controller VCU analyzes the smoke concentration data to determine whether to issue a first thermal runaway warning instruction; and receives the analysis result of the vehicle controller VCU; The second transmission module is configured to transmit data from different operating modes collected by the battery warning system to the cloud server via the vehicle-mounted terminal T-BOX; so that the cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle-mounted terminal and the trained thermal runaway warning model; and receives the analysis results of the cloud server via the vehicle-mounted terminal; an output module configured to: determine whether to issue a thermal runaway alarm based on the two analysis results; The trained thermal runaway warning model; The training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. The determination of whether to issue a thermal runaway alarm is based on the two analysis results; specifically includes: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

7. A method for alarming thermal runaway of an electric vehicle battery, characterized in that: Applied to vehicle controllers, including: Obtain smoke concentration data; Analyze smoke concentration data to determine whether to issue the first thermal runaway warning instruction; Send the first analysis result to the battery management system BMS; The auxiliary battery management system combines the first analysis result and the second analysis result to determine whether to issue a thermal runaway alarm; The second analysis result is sent by the cloud server to the battery management system (BMS) via the vehicle terminal. The cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model. The data uploaded by the vehicle terminal is data collected by different sensors in different operating modes of the battery warning system. The trained thermal runaway warning model; the training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. Determine whether to issue a thermal runaway alarm based on the two analysis results; specifically: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

8. A vehicle controller, characterized by comprising: A first acquisition module is configured to: acquire smoke concentration data; a first analysis module configured to: analyze the smoke concentration data to determine whether to issue a first thermal runaway warning instruction; A first sending module is configured to: send the first analysis result to the battery management system BMS; a first determination module configured to: determine whether to issue a thermal runaway alarm by combining the first analysis result and the second analysis result with the auxiliary battery management system; The second analysis result is sent by the cloud server to the battery management system (BMS) via the vehicle terminal. The cloud server determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model. The data uploaded by the vehicle terminal is data collected by different sensors in different operating modes of the battery warning system. The trained thermal runaway warning model; the training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. Determine whether to issue a thermal runaway alarm based on the two analysis results; specifically: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

9. A method for alarming thermal runaway of an electric vehicle battery, characterized in that: Applicable to cloud servers, including: Acquire data in different working modes and modalities; Determining a second analysis result based on data from different operating modes and modalities and a trained thermal runaway warning model; the second analysis result indicates whether to issue a second thermal runaway warning instruction; Sending the determined second analysis result to the battery management system BMS via the vehicle terminal; The auxiliary battery management system combines the first analysis result and the second analysis result to determine whether to issue a thermal runaway alarm; Data from different operating modes and modalities are collected by different sensors and uploaded to the cloud server in sequence through the battery management system (BMS) and the vehicle terminal. The first analysis result is obtained by analyzing the smoke concentration data by the vehicle control unit (VCU). The first analysis result indicates whether the first thermal runaway warning instruction is issued. The smoke concentration data is collected by the smoke sensor and uploaded to the vehicle control unit (VCU) through the battery management system (BMS). The trained thermal runaway warning model; the training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. Determine whether to issue a thermal runaway alarm based on the two analysis results; specifically: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

10. A cloud server, characterized in that: include: A second acquisition module is configured to: acquire data in different working modes and different modalities; A second analysis module is configured to: determine a second analysis result based on data of different operating modes and different modalities and a trained thermal runaway warning model; The second analysis result refers to whether to issue a second thermal runaway warning instruction; A second sending module is configured to: send the determined second analysis result to the battery management system BMS via the vehicle terminal; a second determination module configured to: determine whether to issue a thermal runaway alarm by combining the first analysis result and the second analysis result with the auxiliary battery management system; Data from different operating modes and modalities are collected by different sensors and uploaded to the cloud server in sequence through the battery management system (BMS) and the vehicle terminal. The first analysis result is obtained by analyzing the smoke concentration data by the vehicle control unit (VCU). The first analysis result indicates whether the first thermal runaway warning instruction is issued. The smoke concentration data is collected by the smoke sensor and uploaded to the vehicle control unit (VCU) through the battery management system (BMS). The trained thermal runaway warning model; the training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. Determine whether to issue a thermal runaway alarm based on the two analysis results; specifically: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

11. An electric vehicle battery thermal runaway alarm system, characterized in that: include: Battery management system BMS, vehicle controller VCU and cloud server; The battery management system BMS transmits smoke concentration data to the vehicle controller VCU; The vehicle controller VCU analyzes the smoke concentration data and determines whether to issue the first thermal runaway warning instruction; The battery management system (BMS) transmits the data of different working modes collected by the battery early warning system to the vehicle terminal T-BOX; The vehicle terminal transmits the data to the cloud server, which determines whether to issue a second thermal runaway warning instruction based on the data uploaded by the vehicle terminal and the trained thermal runaway warning model. The battery management system BMS determines whether to issue a thermal runaway alarm based on the two runaway warning instructions; The trained thermal runaway warning model; The training process includes: Build a convolutional neural network; Constructing a training set; the training set is the historical battery monitoring data of electric vehicles of the same model and batch after they have left the production line, with a known label indicating whether they have experienced thermal runaway; the historical battery monitoring data includes: the temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current, and the bus voltage; The temperature of the battery cells inside the battery pack, the pressure inside the battery pack, the bus current and the bus voltage of the battery pack in the training set are preprocessed through the battery mechanism model to obtain the internal resistance abnormality score, the self-discharge abnormality score, the internal short circuit abnormality score, and the temperature rise abnormality score; wherein the battery mechanism model includes: an internal resistance model, a self-discharge model, an internal short circuit model and a temperature rise model; The internal resistance abnormality score, self-discharge abnormality score, internal short circuit abnormality score, and temperature rise abnormality score in the preprocessing results are weighted and accumulated, and input into the convolutional neural network for network training. When the network loss function value no longer decreases or the set number of iterations is reached, the training is stopped, and the trained convolutional neural network is obtained, that is, the trained thermal runaway warning model is obtained. Determine whether to issue a thermal runaway alarm based on the two analysis results; specifically: If both runaway warning instructions are in warning mode, a thermal runaway alarm is issued; If one of the two runaway warning instructions is in warning mode and the other is in non-warning mode, no thermal runaway alarm will be issued; If the smoke sensor does not sound an alarm, but the battery warning system sounds an alarm, the charging and discharging power will be halved and the battery pack cooling mode will be activated. If both of the two runaway warning instructions are in non-warning mode, it is determined that no thermal runaway alarm is issued.

12. An electronic device, comprising: a memory for non-transitory storage of computer-readable instructions; as well as a processor for executing said computer-readable instructions, When the computer-readable instructions are executed by the processor, the method of any one of claims 1-5, 7 or 9 is performed.

13. A storage medium, characterized by non-transitory storage of computer-readable instructions, wherein: When the non-transitory computer-readable instructions are executed by a computer, the instructions of the method of any one of claims 1-5, 7 or 9 are executed.

14. A computer program product, characterized in that The invention comprises a computer program for implementing the method of any one of claims 1 to 5, 7 or 9 when the computer program is run on one or more processors.

Images