A thermal runaway early warning system and method for charging a vehicle battery

Abstract

The application provides a charging vehicle battery thermal runaway early warning system and method, and belongs to the field of electric vehicles.The system comprises a data acquisition module, which is used for sensing whether a vehicle is driving into a parking space in real time, and collecting sound information and gas information when a vehicle is driving into a parking space; a local analysis module, which is used for judging whether the gas released by the charging vehicle is abnormal according to the gas information, and alarming if so; filtering the sound information to obtain target sound information, preliminarily comparing the target sound information with abnormal sound information, and alarming when the target sound signal is abnormal; and a cloud analysis module, which is used for further comparing the target sound information by using a deep learning model, alarming when the target sound information is abnormal, and cutting off the charging power supply.The application can automatically give early warning for the thermal runaway of the battery of the charging vehicle, and improves the response speed and the safety of the charging station.

Description

Technical Field

[0001] This invention relates to the field of electric vehicles, and in particular to a battery thermal runaway early warning system and method for charging vehicles. Background Technology

[0002] With the increasing popularity of electric vehicles, the safety issue of thermal runaway in lithium batteries has gradually attracted attention. Thermal runaway refers to the inability to dissipate heat generated inside the battery during charging, discharging, or long-term use, causing the battery temperature to rise rapidly and ultimately leading to an explosion or fire. In recent years, there have been numerous fires and accidents caused by battery thermal runaway worldwide. These incidents seriously threaten the safety and reliability of electric vehicles and have become a bottleneck in the development of electric vehicles.

[0003] The process of thermal runaway evolving into a fire and explosion in lithium-ion batteries of energy storage power stations can generally be divided into four stages: ① The battery releases heat under abuse conditions, producing flammable and toxic gases; ② The heat and flammable gases create significant pressure within the sealed space of the battery casing, causing the safety valve to open and release gas; ③ The high-temperature gas release passes through the safety valve, forming a jet fire or a large amount of high-temperature flammable and toxic gas mixture; ④ The high-temperature gas mixture accumulates in the single prefabricated storage structure, eventually igniting upon encountering an ignition source. Therefore, to prevent fire and explosion accidents in power batteries, it is necessary and crucial to propose prevention and control measures based on the thermal runaway evolution process.

[0004] Currently, thermal runaway safety status monitoring and early warning technologies mainly involve real-time monitoring of battery parameters such as temperature, current, and voltage, as well as the analysis and processing of this data to predict the battery's safety status. Commonly used technologies include infrared thermal imagers, thermistors, miniature thermocouples, and surface-mount temperature sensors, which can monitor battery temperature and other parameters in real time. However, these battery thermal runaway monitoring technologies are primarily installed on the battery module, and charging stations cannot obtain vehicle battery status information. Therefore, it is necessary to install external vehicle battery thermal runaway monitoring equipment at charging stations.

[0005] Traditional charging station monitoring technologies mostly use smoke sensors, infrared temperature detectors, flame detectors, and video surveillance to provide fire early warnings for thermal runaway of the power batteries of charging vehicles. The principle is to measure the surface temperature of the vehicle and battery and monitor the smoke after the fire to trigger an alarm. This monitoring technology has high reliability, but its early warning capability is poor and the response is delayed, making it impossible to provide an early warning within 300 seconds.

[0006] Based on the above problems, there is an urgent need for a new early warning method to enable early warning of battery thermal runaway at charging stations and improve response speed. Summary of the Invention

[0007] The purpose of this invention is to provide a battery thermal runaway early warning system and method for charging vehicles, which can automatically provide early warning of battery thermal runaway at charging stations and improve response speed.

[0008] To achieve the above objectives, the present invention provides the following solution:

[0009] A battery thermal runaway early warning system for charging vehicles, comprising:

[0010] The data acquisition module is installed on the ground of the parking space in the charging station to sense in real time whether a vehicle enters the parking space and to collect environmental data when a vehicle enters the parking space; the environmental data includes sound information and gas information.

[0011] The local analysis module, connected to the data acquisition module, is used to determine whether the gas released by the charging vehicle is abnormal based on the gas information. If so, a first warning signal is generated. The sound information is filtered to obtain target sound information, and the target sound information is compared with the abnormal sound information. If the degree of match between the target sound information and the abnormal sound information is greater than a first threshold, a first warning signal is generated.

[0012] A cloud-based analysis module, connected to the local analysis module, is used to determine whether the target sound information is an abnormal sound when the degree of matching between the target sound information and the abnormal sound information is greater than a first threshold. If the matching degree is greater than a first threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If so, a second warning signal and a power-off command are generated; otherwise, a power-off command is generated. When the degree of matching between the target sound information and the abnormal sound information is less than or equal to the first threshold and greater than a second threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If the matching degree is greater than a first threshold, a first warning signal and a power-off command are generated; otherwise, a warning cancellation command is generated. The second threshold is less than the first threshold.

[0013] The cloud analysis module is also used to generate battery information for the corresponding vehicle based on the environmental data, store the battery information for the corresponding vehicle in the database, and predict the battery life based on the battery information for the corresponding vehicle stored in the database.

[0014] An alarm module is installed on the charging pile and connected to the local analysis module and the cloud analysis module respectively. It is used to trigger an alarm based on the first warning signal or the second warning signal, and to stop the alarm based on the cancellation warning command.

[0015] The power-off execution module is connected to both the charging pile and the cloud analysis module, and is used to cut off the charging power of the charging pile according to the power cut-off command, thereby stopping the charging of the vehicle.

[0016] Optionally, the data acquisition module includes: an upper cover, a lower cover, a geomagnetic sensor, multiple microphone sensors, and multiple gas sensors;

[0017] The lower surface of the upper cover is fixed to the upper surface of the lower cover; the upper cover has a detection hole; the upper cover is arc-shaped.

[0018] The geomagnetic sensor, each microphone sensor, and each gas sensor are all installed on the lower surface of the upper cover; the geomagnetic sensor is used to detect in real time whether a vehicle has entered the parking space; the microphone sensor is used to collect sound information when a vehicle enters the parking space; and the gas sensor is used to collect gas information when a vehicle enters the parking space.

[0019] Optionally, the number of microphone sensors is three; the three microphone sensors are arranged in an equilateral triangle on the lower surface of the upper cover; the number of gas sensors is two; the two gas sensors are arranged diagonally on the lower surface of the upper cover.

[0020] Optionally, the number of microphone sensors is four; three microphone sensors are arranged in an equilateral triangle on the lower surface of the upper cover, and one microphone sensor is located at the center of the upper cover; the number of gas sensors is three; the three gas sensors are respectively located at the midpoint of each side of the equilateral triangle.

[0021] Optionally, the detection hole is covered with a hydrophobic and breathable membrane.

[0022] Optionally, the data acquisition module further includes:

[0023] A temperature sensor is installed on the lower surface of the upper cover to detect the ambient temperature in real time.

[0024] The local analysis module is also connected to the temperature sensor, and the local analysis module is also used to determine whether the ambient temperature exceeds the temperature threshold. If so, a third warning signal is generated.

[0025] The alarm module is also used to issue an alarm based on the third warning signal.

[0026] Optionally, the local analysis module is a control circuit board; the control circuit board is disposed on the lower surface of the upper cover.

[0027] Optionally, the local analysis module includes:

[0028] The gas judgment submodule is connected to the data acquisition module and is used to determine whether the gas concentration in the gas information is greater than a set gas concentration threshold and whether the time for which the gas concentration is greater than the gas concentration threshold is greater than a set time threshold. If so, a first warning signal is generated.

[0029] The sound filtering submodule, connected to the data acquisition module, is used to determine the direction and location of the sound information using a direction-of-arrival estimation method, and to filter the sound information based on the direction and location of the sound information to obtain the target sound information;

[0030] The sound comparison submodule, connected to the sound filtering submodule, is used to compare the target sound information with the abnormal sound information. If the match between the target sound information and the abnormal sound information is greater than a first threshold, a first warning signal is generated. The abnormal sound information includes the sound of a battery pack rupture and the sound of a safety valve rupture.

[0031] Optionally, the alarm module is a magnetic audible and visual alarm.

[0032] To achieve the above objectives, the present invention also provides the following solution:

[0033] A method for early warning of thermal runaway of a charging vehicle battery includes:

[0034] The system can detect in real time whether a vehicle has entered a parking space. If a vehicle does enter a parking space, environmental data is collected. The environmental data includes sound information and gas information.

[0035] Based on the gas information, determine whether the gas released by the charging vehicle is abnormal. If so, generate a first warning signal and issue a level one alarm.

[0036] The sound information is filtered to obtain the target sound information;

[0037] The target sound information is compared with the abnormal sound information locally. If the degree of match between the target sound information and the abnormal sound information is greater than a first threshold, a level one alarm is triggered, and the target sound information is sent to the cloud server.

[0038] When the degree of match between the target sound information and the abnormal sound information is greater than a first threshold, the cloud server uses a deep learning model to determine whether the target sound information is an abnormal sound. If it is, a second-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle; otherwise, a first-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle.

[0039] When the degree of match between the target sound information and the abnormal sound information is less than or equal to the first threshold and greater than the second threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If it is, a level one alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle; otherwise, the alarm is stopped; the second threshold is less than the first threshold.

[0040] The cloud server generates battery information for the corresponding vehicle based on the environmental data, stores the battery information for the corresponding vehicle in the database, and predicts the battery life based on the battery information for the corresponding vehicle stored in the database.

[0041] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0042] This invention detects sound and gas information only when a vehicle enters the parking space, reducing the workload of the data acquisition module. By combining local and cloud analysis modules to analyze sound and gas information, it can issue an alarm when the sound or gas emitted by the charging vehicle is abnormal and promptly cut off the charging power of the charging pile. It can automatically provide early warning of thermal runaway of the charging vehicle battery, improving response speed and the safety of the charging station. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 A block diagram of a battery thermal runaway early warning system for charging vehicles provided by the present invention;

[0045] Figure 2 This is a schematic diagram showing the installation location of the data acquisition module;

[0046] Figure 3 A schematic diagram of the structure of the early warning system for thermal runaway of a charging vehicle battery provided by the present invention;

[0047] Figure 4 This is a schematic diagram of a first embodiment of the microphone sensor and the gas sensor;

[0048] Figure 5 This is a schematic diagram of a second embodiment of the microphone sensor and the gas sensor;

[0049] Figure 6 This is a block diagram of the working block of the charging vehicle battery thermal runaway early warning system provided by the present invention;

[0050] Figure 7 A flowchart of the charging vehicle battery thermal runaway early warning system provided by the present invention;

[0051] Figure 8 An architecture diagram of the early warning system for thermal runaway of a charging vehicle battery provided by the present invention;

[0052] Figure 9 A flowchart of the electric vehicle battery thermal runaway early warning method provided by the present invention.

[0053] Symbol explanation:

[0054] 1-Data acquisition module, 2-Local analysis module, 3-Cloud analysis module, 4-Alarm module, 5-Power failure execution module, 6-Expansion bolt, 7-Concrete floor, 8-Upper cover, 9-Lower cover, 10-Solar circuit board, 11-Gas sensor, 12-Microphone sensor, 13-Lithium battery, 14-Temperature sensor. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] During a charging anomaly, in addition to the heat released by internal decomposition reactions, a reactive gas is also produced, and the electrolyte evaporates due to the increased temperature, leading to a rapid increase in internal battery pressure. When the pressure reaches a critical value, the safety valve will open to release flammable gases, electrolyte droplets, and solid particles. The purpose of this invention is to provide a battery thermal runaway early warning system and method for charging vehicles. For charging stations, it employs gas sensing technology and voiceprint recognition early warning technology to detect, intelligently judge, and intervene in the early stages of thermal runaway of power batteries, such as rupture or the generation of special gases. This allows for a predictable and controllable solution to safety hazards caused by battery thermal runaway, providing early warning.

[0057] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0058] Before a rechargeable battery catches fire, abnormal parameters indicating thermal runaway can be monitored. Some of these parameters, when appropriately combined, can effectively reflect the early development trend of thermal runaway, enabling more sophisticated early warning systems. A detailed analysis was conducted on the types, quantities, and influencing factors of gases released during thermal runaway in power batteries of different capacities. The study revealed that CO2, CO, H2, C2H4, CH4, C2H, and C3H6 are the seven most common gases involved in lithium-ion battery thermal runaway. The total amount of gas released during thermal runaway is closely related to the battery's capacity, with an average release of approximately 2L of gas per Ah of capacity. Furthermore, before the gas is released into the external environment, high pressure forms inside the battery pack. If the gas breaches the battery safety valve, causing it to rupture, an abnormal sound will be heard.

[0059] This invention provides a thermal runaway early warning system for charging vehicle batteries. By fusing acoustic array sensors and gas array sensors, it can monitor the special sounds and special gas releases that occur in advance when the vehicle battery system is about to experience thermal runaway, thereby determining whether the battery system is in a thermal runaway state and taking timely measures to prevent the battery system from experiencing thermal runaway.

[0060] like Figure 1 As shown, the charging vehicle battery thermal runaway early warning system provided by the present invention includes: a data acquisition module 1, a local analysis module 2, a cloud analysis module 3, an alarm module 4, and a power-off execution module 5.

[0061] The data acquisition module 1 is installed on the ground of the parking space in the charging station to sense in real time whether a vehicle enters the parking space and to collect environmental data when a vehicle enters the parking space; the environmental data includes sound information and gas information.

[0062] Data acquisition module 1, also known as the parking space sniffing sensor, is a sensing device installed on the ground in the parking space of a charging vehicle. It includes an "electronic ear" composed of multiple microphone sensor arrays, an "electronic nose" composed of multiple gas sensor arrays, and a geomagnetic sensor. For example... Figure 2 As shown, the data acquisition device is mounted on the concrete ground 7 at the center of the parking space via expansion bolts 6.

[0063] Specifically, such as Figure 3 As shown, the data acquisition module 1 includes: an upper cover 8, a lower cover 9, a geomagnetic sensor (not shown in the figure), multiple microphone sensors 12, and multiple gas sensors 11. The number of microphone sensors 12 is at least 3, and the number of gas sensors 11 is at least 2.

[0064] The lower surface of the upper cover 8 is fixed to the upper surface of the lower cover 9. Specifically, the upper cover 8 and the lower cover 9 are fastened with two hexagonal socket head cap screws. To ensure the airtightness of the device, a sealing ring (silicone gasket) is embedded in the lower cover 9. To ensure the pressure resistance of the device, the upper cover 8 and the lower cover 9 are made of Teflon material. The excess space of the upper cover 8 and the lower cover 9 is potted with adhesive. The upper cover 8 and the lower cover 9 constitute the outer shell of the data acquisition module 1. The upper cover 8 and the lower cover 9 have high pressure resistance and a high protection level. The high pressure resistance ensures that even if a vehicle tire runs over it, the internal sensors will not be damaged, and the high protection ensures that rain, snow, and other harsh weather conditions will not affect its working performance.

[0065] Considering that the thermal runaway warning system for the charging vehicle battery is installed in the parking space, the top cover is arc-shaped to reduce the pressure exerted by the vehicle on the system. The geomagnetic sensor, microphone sensor, and gas sensor are mounted using a flexible PCB design to enhance the system's pressure resistance (16t) and damage resistance.

[0066] This invention divides the entire device into two parts, which facilitates later equipment operation and maintenance, including battery replacement and troubleshooting, simplifies the later operation and maintenance process, and reduces the difficulty of the production process.

[0067] To ensure the normal operation of the sensors while maintaining waterproof performance, the upper cover 8 has detection holes. These holes are covered with a hydrophobic and breathable membrane. This ensures both the normal operation of the sensors and the airtightness of the equipment.

[0068] The geomagnetic sensor, each microphone sensor 12, and each gas sensor 11 are all disposed on the lower surface of the upper cover 8. The geomagnetic sensor is used to detect in real time whether a vehicle has entered the parking space. The microphone sensor 12 is used to collect sound information when a vehicle enters the parking space. The gas sensor 11 is used to collect gas information when a vehicle enters the parking space.

[0069] The geomagnetic sensor uses the GMRL series multi-mode composite detection module. It is primarily used to detect the presence of vehicles in lanes and parking spaces. Utilizing millimeter-wave radar and geomagnetic joint detection technology, the device operates with dual precision detection, resulting in low power consumption. To further reduce power consumption and extend the device's lifespan, a low-power mode is activated when the geomagnetic sensor is not in use.

[0070] Gas sensor 11 transmits gas information to the control board via a Universal Asynchronous Receiver / Transmitter (UART). Microphone sensor 12 transmits sound information to the control board via an Integrated Interface of Sound (IIS) interface.

[0071] Gas sensor 11 uses the ZES21-CS battery leak detection module, which utilizes electrochemical theory to detect gases volatilized from the electrolyte and gases produced during battery combustion. It incorporates a built-in temperature sensor for temperature compensation. It features high sensitivity, high resolution, low power consumption, and long service life, meeting the device's low power consumption requirements.

[0072] The microphone sensor 12 uses the SPH0645 module, employs IIS communication, and outputs digital signals digitally. It does not require an external audio codec, can be directly interconnected with the processor, features an ultra-small package design, supports dual microphones, multiple mode adjustments, and ultra-low power consumption, which greatly reduces the power consumption of the device.

[0073] As a specific implementation method, such as Figure 4 As shown, there are three microphone sensors 12. These three microphone sensors 12 are arranged in an equilateral triangle on the lower surface of the upper cover 8, detecting ambient sound from three angles and collecting sound information from the battery at the bottom of the car and the surrounding environment. There are two gas sensors 11 (detecting CO and CO2 respectively). These two gas sensors 11 are arranged diagonally on the lower surface of the upper cover 8. That is, two gas sensors 11 are distributed along a 180° diagonal, forming a gas detection array used to collect toxic gases emitted due to damage to internal circuit components caused by battery malfunction.

[0074] As another specific implementation method, such as Figure 5 As shown, there are four microphone sensors 12. Three microphone sensors 12 are arranged in an equilateral triangle on the lower surface of the upper cover 8, and one microphone sensor 12 is located at the center of the upper cover 8. There are three gas sensors 11. The three gas sensors 11 are respectively located at the midpoint of each side of the equilateral triangle. The four microphone sensors 12 can better locate the voiceprint position and analyze the voiceprint data, greatly reducing interference from external factors. The three gas sensors 11 are arranged to detect toxic and harmful gases emitted by the battery in the environment from three angles, ensuring detection accuracy. The microphone sensors 12 and gas sensors 11 are arranged alternately, with one gas sensor 11 placed next to each microphone sensor 12. This invention uses an array to detect sound and gas, optimizing the detection error of a single sensor.

[0075] Furthermore, such as Figure 3 As shown, the data acquisition module 1 also includes a temperature sensor 14. The temperature sensor 14 is disposed on the lower surface of the upper cover 8 and is used to detect the ambient temperature in real time. In this embodiment, the temperature sensor 14 is disposed on the control circuit board. In the three-level early warning of battery thermal runaway, before a fire occurs, the ambient temperature of the vehicle suddenly increases. In addition to gas detection and sound detection, the temperature growth trend can also be used to help predict battery thermal runaway.

[0076] The data acquisition module 1 is installed in the center of the parking space floor and enters sleep mode when no vehicle is parked in the space. In sleep mode, the microphone sensor 12 and gas sensor 11 are inactive; only the geomagnetic sensor is active. The sleep mode retains the remote wireless upgrade function. This invention's data acquisition module is low-cost, highly integrated, and eliminates the need for multiple sensor installations, reducing system construction difficulty and minimizing system resource requirements and configuration.

[0077] The data acquisition module 1 can be powered by a battery or by DC power.

[0078] When using battery power, the first method to extend the device's operating time is to install a solar circuit board 10 on the top of the upper cover 8 to convert solar energy into electrical energy. The second method is to use three 18650 lithium batteries 13, increasing battery capacity, ensuring longer operating time, and allowing for 1000 recharge cycles to ensure normal operation. The 18650 lithium batteries 13 are located on the lower cover 9.

[0079] When using DC power, power is supplied through an external DC wide-range power supply, eliminating the need to consider power consumption issues, improving the performance of the data acquisition module, ensuring connection stability, and avoiding the hassle of battery replacement during equipment maintenance.

[0080] like Figure 6 and Figure 7 As shown, in order to better understand the technical solution of the present invention, the working process of the battery thermal runaway early warning system for charging vehicles is described in detail below:

[0081] (1) The geomagnetic sensor detects whether a vehicle has entered the parking space. If a vehicle is detected, it is assumed that the vehicle needs to be charged next. At this time, the microphone sensor 12 and the gas sensor 11 are activated by the geomagnetic sensor to enter the working mode.

[0082] (2) When the working mode is started, the microphone sensor 12 and the gas sensor 11 perform a self-test. During the self-test, the microphone sensor 12 and the gas sensor 11 are checked to see if they are working properly. This process lasts for about 5 minutes.

[0083] (3) Zero-point self-calibration process. After 5 minutes, if the self-test is successful and the preheating is complete, a zero-point voltage comparison is performed. That is, the zero-point voltage of the gas sensor 11 is compared with the previously stored zero-point voltage of the gas sensor 11. If the comparison exceeds a certain threshold, a zero-point self-calibration is performed, and the current zero-point voltage is overwritten to avoid false alarms due to the difference in zero-point voltage.

[0084] (4) Formal Monitoring State. After self-calibration, the device enters formal environmental monitoring state. The microphone sensor 12 and gas sensor 11 collect sound and gas information of the environment around the parking space in real time. Typically, when the battery is about to experience thermal runaway, abnormal sound information will precede abnormal gas information. Abnormal sound information mainly includes the sound of the battery safety valve breaking, abnormal charging sounds during the charging process, and the sound of the battery continuously releasing gas after breaking; abnormal gas information mainly includes abnormally high concentrations of CO gas, abnormally high concentrations of CO2 gas, abnormally high concentrations of VOCs gas, and abnormally high concentrations of H2 gas.

[0085] (5) Comparison and Analysis Process. The comparison and analysis process is divided into two types: one is analysis performed on the device side (i.e., the local analysis module), and the other is analysis performed on the server side (i.e., the cloud analysis module). First, a preliminary local analysis is performed, while simultaneously backing up data to the cloud. When the local analysis is in an abnormal state, an alert is activated, and cloud algorithm analysis is performed. If the anomaly persists, the charging power is automatically cut off, and management personnel are contacted for investigation. Through phased and iterative high- and low-order algorithm analysis, the accuracy of detection is ensured, and the false alarm rate is reduced.

[0086] The local analysis module 2 is connected to the data acquisition module 1. The local analysis module 2 is used to determine whether the gas released by the charging vehicle is abnormal based on the gas information. If so, a first warning signal is generated. The sound information is filtered to obtain target sound information, and the target sound information is compared with the abnormal sound information. If the degree of matching between the target sound information and the abnormal sound information is greater than a first threshold, a first warning signal is generated.

[0087] The first warning signal generated by the local analysis module 2 when determining gas or sound abnormalities is the same, both serving as a Level 1 warning. That is, a Level 1 warning is triggered when there is a gas abnormality and the voiceprint comparison matches by 50%, or when the voiceprint reaches a fixed value or when there is a gas abnormality.

[0088] Furthermore, the local analysis module 2 is also connected to the temperature sensor 14. The local analysis module 2 is also used to determine whether the ambient temperature exceeds the temperature threshold. If so, a third warning signal is generated.

[0089] In this embodiment, the local analysis module 2 is a control circuit board. The control circuit board is located on the lower surface of the upper cover 8. The control circuit board collects, parses, and performs algorithm analysis on the data. The control circuit board is an ESP32S2 series microcontroller with an external 2MB pseudo-static random access memory (PSRAM) to store local algorithms, providing an additional layer of early warning analysis and reducing the false alarm rate. Furthermore, the ESP32S2 integrates Wi-Fi communication, supporting large data transmission volumes at a rate of up to 150Mbps, ensuring the real-time performance of the voiceprint cloud analysis.

[0090] Specifically, the local analysis module 2 includes: a gas judgment submodule, a sound filtering submodule, and a sound comparison submodule.

[0091] The gas detection submodule is connected to the data acquisition module 1. The gas detection submodule is used to determine whether the gas released by the charging vehicle is abnormal based on the gas information. If so, a first warning signal is generated. Specifically, the gas detection submodule determines whether the gas concentration in the gas information is greater than a set gas concentration threshold, and whether the time for which the gas concentration is greater than the gas concentration threshold is greater than a set time threshold. If so, a first warning signal is generated. The gas concentration includes the concentration of carbon monoxide, hydrogen, etc.

[0092] The sound filtering submodule is connected to the data acquisition module 1. The sound filtering submodule is used to determine the direction and position of the sound information using the direction of arrival estimation method, and to filter the sound information according to the direction and position of the sound information to obtain the target sound information.

[0093] The sound comparison submodule is connected to the sound filtering submodule. The sound comparison submodule compares the target sound information with the abnormal sound information. If the match between the target sound information and the abnormal sound information is greater than a first threshold, a first warning signal is generated. The abnormal sound information includes the sound of a battery pack rupture and the sound of a safety valve rupture.

[0094] In this embodiment, the data acquisition module 1 and the local analysis module 2 together constitute a local device. The local device is connected to the cloud analysis module 3 via a wireless transmission module. The wireless transmission module is a WIFI or 4G wireless transmission module.

[0095] The local analysis module compares the target sound information with the abnormal sound information, including: determining whether the degree of similarity between the target sound information and the abnormal sound information is greater than a first threshold, and determining whether the degree of similarity between the target sound information and the abnormal sound information is less than or equal to the first threshold and greater than a second threshold.

[0096] The local analysis module leverages the inherent characteristics of the sensors and integrates algorithmic processing to reduce device power consumption. By comparing gas detection and synchronous acoustic anomaly algorithms, the two triggering methods prevent the device's sensors from malfunctioning and missing the optimal warning period.

[0097] The cloud analysis module 3 is connected to the local analysis module 2. The cloud analysis module 3 is used to determine whether the target sound information is abnormal when the degree of matching between the target sound information and the abnormal sound information is greater than a first threshold. If so, a second warning signal and a power-off command are generated; otherwise, a power-off command is generated. When the degree of matching between the target sound information and the abnormal sound information is less than or equal to the first threshold and greater than a second threshold, the cloud analysis module 3 uses a deep learning model to determine whether the target sound information is abnormal. If so, a first warning signal and a power-off command are generated; otherwise, a warning cancellation command is generated. The second threshold is less than the first threshold.

[0098] The first threshold represents the slope of the increase in gas concentration collected by the gas sensor within a unit time (1 second), and the local voiceprint analysis similarity reaches a fixed value.

[0099] The second threshold represents the voiceprint similarity of the battery analyzed in the cloud. It is used to retrieve the vehicle's past charging battery status analysis and perform multi-dimensional analysis to improve the warning rate and reduce false alarms.

[0100] In this embodiment, the cloud analysis module 3 is a cloud server. The cloud server includes a database server, a backend server, and a frontend server.

[0101] The local analysis module of this invention is triggered by gas and acoustic anomalies, and uses local acoustic signatures to detect the battery's health status. The cloud analysis module is also used to generate battery information for the corresponding vehicle based on the environmental data, store the battery information for the corresponding vehicle in a database, and predict the battery's lifespan based on the battery information stored in the database. By analyzing acoustic signatures in the cloud and retrieving the vehicle's past battery health status as auxiliary constraints, the battery's lifespan can be predicted in advance, providing early warning.

[0102] This means that the battery information of the charging vehicle is uploaded and recorded each time a problem is detected. If a problem is found, the battery information of the vehicle in the past is retrieved from the database for longitudinal analysis, and combined with real-time voiceprint information for horizontal analysis. Through diversified detection, the accuracy of battery warning is improved.

[0103] To enhance the intuitiveness of rechargeable battery thermal runaway monitoring, the data acquisition module, local analysis module, and cloud analysis module are also connected to a 3D digital twin platform. This platform displays the charging station's on-site operating status. The invention integrates the entire 3D digital twin platform, providing a real-time 1:1 display of the charging station's on-site operating status on a large screen, significantly improving the intuitiveness of charging station monitoring.

[0104] To better understand the technical solution of this invention, the working process of the local analysis module 2 and the cloud analysis module 1 is described in detail below:

[0105] Gas information is analyzed only in the local analysis module. If an anomaly is detected in a certain type of gas, a Level 1 warning is issued; if multiple types of gases are detected, a Level 2 warning is issued. Specifically, for gas information, if the local analysis module determines that a certain type of gas is abnormal, it generates a first warning signal for a Level 1 warning; if it determines that multiple types of gases are abnormal, it generates a second warning signal for a Level 2 warning.

[0106] The sound information collected by the microphone sensor is first processed through signal processing such as filtering, and then undergoes simple analysis by the local analysis module. The first step is to analyze the location of the abnormal sound source. Since the data acquisition module is installed in the middle of the parking space, under the vehicle, the abnormal sound source should originate from the vehicle chassis, i.e., above the data acquisition module. If the sound source originates from other directions, such as the side, it can be preliminarily determined to be other environmental noise. Next, it is compared with a pre-stored fuzzy voiceprint database. If the match is greater than a first threshold, a first warning signal is generated for a level one alert, and the collected sound information is simultaneously sent to the cloud analysis module for a second comparison and analysis.

[0107] If the cloud analysis module confirms no error, it generates a second warning signal, escalating to a level two warning. If the cloud analysis module cannot determine the cause, it maintains the level one warning. If the match is below the first threshold but above the second threshold, the collected sound information is sent to the cloud analysis module for secondary comparison and analysis. If the cloud analysis module determines the anomaly is correct, it directly issues a level one warning based on the first warning signal. If the cloud analysis module determines the anomaly is incorrect, it cancels the level one warning but continues to monitor other sensor data. If the match is below the second threshold, no alarm is triggered, but the collected sound information is still sent to the cloud analysis module, which does not perform analysis but simply stores it on the server. In other words, after a level one warning is triggered, the process of initiating a level two warning through cloud-based multi-dimensional analysis begins.

[0108] If both the sound and gas sensors detect an anomaly simultaneously, a level-two warning signal will be generated and a level-two warning will be issued.

[0109] The alarm module is installed on the charging pile and is connected to the local analysis module and the cloud analysis module respectively. The alarm module is used to trigger an alarm based on the first warning signal or the second warning signal, and to stop the alarm based on the cancellation warning command.

[0110] Furthermore, the alarm module is also used to issue an alarm based on the third warning signal.

[0111] In this embodiment, the alarm module is a magnetic audible and visual alarm with wireless IoT functionality. The alarm module can also be installed near charging stations or parking spaces. The magnetic design avoids the need for wiring. The alarm module is placed above the charging station, in a location visible to the naked eye. It is internally powered by a battery, but can also be externally powered, and communicates with a cloud-based analytics module via Wi-Fi.

[0112] When battery thermal runaway occurs, the local analysis module and the cloud analysis module will send alarm commands to their respective alarm devices and the charging station's vehicle thermal runaway early warning and control platform, alerting the charging vehicle at that parking space to an anomaly. The frequency and decibel level of the alarm sound will increase accordingly with the warning level.

[0113] The power-off execution module is connected to both the charging pile and the cloud analysis module. The power-off execution module is used to cut off the charging power of the charging pile according to the power cut-off command, thereby stopping the charging of the vehicle.

[0114] Once an alarm is triggered, the power-off module (usually an automatic shut-off valve, etc.) cuts off the charging power, stopping the vehicle from charging. On-site personnel prepare firefighting measures.

[0115] In industrial and emergency scenarios, to improve alarm response speed while reducing equipment power consumption and extending equipment lifespan, multi-level models are often applied to multi-dimensional sensor fusion monitoring. This invention employs a four-level model for battery thermal runaway early warning and monitoring. By monitoring and judging battery thermal runaway in stages, it aims to achieve timely response, reduced power consumption, and improved accuracy. The first and second level models correspond to the local analysis module, while the third and fourth level models correspond to the cloud analysis module.

[0116] The first-level model (low-power sound location model) identifies the sound source through a combination of multiple microphone sensors. By performing time and phase difference analysis on the sound signals collected by different microphone sensors, the direction and location of the sound source can be determined. The location features are obtained through the relative position information between the microphone sensors and can be used to accurately locate the sound source. This invention uses a Direction of Arrival (DOA) estimation method to calculate the sound source's location information. First, the collected signals from multiple microphone sensors are processed in frames, and an autocorrelation function is used to estimate the time difference of sound source emission, thereby calculating the angle of arrival of the sound source to each microphone in the array.

[0117] The second-level model (low-power acoustic sensing model) is a low-power local device-side analysis model based on Fourier transform and filter design. This model is primarily used to detect low-frequency and characteristic sound signals generated by the battery pack, such as noise from battery pack rupture or safety valve failure. This model requires minimal processing and computation, effectively reducing power consumption, while simultaneously monitoring the basic state of the vehicle's battery pack. The low-power acoustic sensing model employs Fourier transform and filter design techniques to detect low-frequency noise signals emitted by the battery pack. After detecting low-frequency and characteristic sound signals, the second-level model sends the data to the server for secondary comparison. If the comparison is successful, a second-level warning is issued.

[0118] The third-level model is an advanced acoustic sensing model based on deep learning. By processing high-frequency noise and other abnormal signals generated by the battery pack, it can accurately determine whether the battery pack has experienced abnormal conditions such as thermal runaway. This advanced acoustic model improves monitoring accuracy by transmitting data to the server for processing and analysis. This level of model uses a deep learning model to process the high-frequency noise signals emitted by the battery pack to determine the battery's state and thermal runaway status. The third-level model mainly uses deep learning models such as convolutional neural networks and long short-term memory networks to achieve high-precision detection of battery thermal runaway status. The formula for the third-level model is as follows: s_H(t) = f_H(X); where s_H(t) represents the signal output by the third-level model; X represents the input audio signal; and f_H() represents the judgment result obtained after processing the input signal through deep learning models such as convolutional neural networks and long short-term memory networks.

[0119] The fourth-level model is a battery thermal runaway detection algorithm based on multi-sensor fusion. It integrates data from multiple sensors and uses methods such as weight allocation to obtain more accurate and reliable judgment results, thereby determining whether the battery pack has experienced major anomalies such as thermal runaway. The fourth-level model can achieve efficient and accurate battery pack thermal runaway management during the charging process in various application scenarios, improving the safety and reliability of vehicle charging.

[0120] This invention combines measurement data from sound sensors and different gas sensors (and may also combine other sensors such as temperature sensors), and uses machine learning algorithms for data processing and analysis, thereby realizing multi-sensor fusion early warning judgment of battery thermal runaway.

[0121] The fourth-level model performs data fusion processing on data collected from multiple sensors, and the specific implementation process is as follows:

[0122] (1) Data preprocessing: The data collected by each sensor is processed, including data cleaning, wavelet transform, etc., to obtain multi-scale wavelet coefficients.

[0123] (2) Feature extraction and selection: Calculate the mean and standard deviation of the multi-scale wavelet coefficients, and use methods such as correlation coefficient and chi-square test to select and screen features to obtain the final feature vector.

[0124] (3) Classification modeling: The feature vector is used as input, and the Support Vector Machine (SVM) and Random Forest (RF) algorithms are used for modeling and prediction to obtain the classification results of the two classifiers.

[0125] (4) Fusion Method: The classification results of SVM and RF are fused using a weighted average method to obtain the final judgment result. The weighted average method can be expressed as the following formula:

[0126]

[0127] Where y represents the final prediction result, n is the number of classifiers, and w i p represents the weight of the i-th classifier. i This represents the prediction result of the i-th classifier.

[0128] In summary, the four-level model can optimize equipment operating efficiency, improve the timeliness, accuracy and reliability of monitoring, and achieve advanced early warning and monitoring of thermal runaway of charging vehicles.

[0129] In addition, users can also view the voiceprint distribution and the surrounding sound heatmap through the display platform. For example... Figure 8 The diagram shown is an overall architecture diagram of the battery thermal runaway early warning system for charging vehicles according to the present invention.

[0130] This invention reduces overall power consumption and extends lifespan through algorithm analysis at different stages and flexible device operating modes. On the hardware side, this is mainly reflected in the geomagnetic sensor above the control circuit board, which only activates the operating mode when a vehicle is detected in the parking space, remaining in sleep mode most of the time to reduce battery waste. On the software side, the automatic selection between local and cloud-based voiceprint algorithm analysis is key. Different algorithm analysis modes are adopted based on different states, reducing the workload of the central processing unit and the high energy consumption caused by large amounts of data transmission during communication.

[0131] like Figure 9 As shown, the present invention also provides a method for early warning of thermal runaway of a charging vehicle battery, comprising:

[0132] Step 100: Real-time detection of whether a vehicle has entered the parking space. If a vehicle has entered the parking space, environmental data is collected. The environmental data includes sound information and gas information.

[0133] Step 200: Determine whether the gas released by the charging vehicle is abnormal based on the gas information. If so, generate a first warning signal and issue a level one alarm.

[0134] Step 300: Filter the sound information to obtain the target sound information.

[0135] Step 400: Compare the target sound information with the abnormal sound information locally. If the match between the target sound information and the abnormal sound information is greater than a first threshold, a level one alarm is triggered, and the target sound information is sent to the cloud server.

[0136] Step 500: When the degree of match between the target sound information and the abnormal sound information is greater than the first threshold, the cloud server uses a deep learning model to determine whether the target sound information is an abnormal sound. If so, a second-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle; otherwise, a first-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle.

[0137] Step 600: When the degree of match between the target sound information and the abnormal sound information is less than or equal to a first threshold and greater than a second threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If so, a level one alarm is triggered, and the charging power supply of the charging pile is cut off to stop charging the vehicle; otherwise, the alarm is stopped. The second threshold is less than the first threshold.

[0138] Step 700: Generate the corresponding vehicle's battery information based on the environmental data through the cloud server, store the corresponding vehicle's battery information in the database, and predict the battery's lifespan based on the corresponding vehicle's battery information stored in the database.

[0139] This invention is based on a combination of multiple gas sensors and a multi-microphone array. By combining and analyzing the collected gas and sound information, it can detect and warn of thermal runaway of the power battery during vehicle charging. This predictive alarm improves the response speed of electric vehicle power battery thermal runaway and the safety of charging stations.

[0140] In summary, the present invention has the following advantages:

[0141] 1. Compared to traditional temperature, smoke, or single gas sensing technologies, it can detect abnormal signals of fires in charging vehicles more quickly, earlier, and more accurately, which helps to provide early warning and handle fire risks.

[0142] 2. Sound sensing and gas sensing technologies can detect ultrasonic waves and minute gas changes, and are not easily affected by environmental factors such as light, humidity, and temperature, thus exhibiting high sensitivity.

[0143] 3. The equipment does not require a lot of space, is easy to install, and does not require a lot of manpower and resources for maintenance, making it very convenient to implement.

[0144] 4. It can automatically identify signals of fires in charging vehicles, enabling automated detection. Once a signal of a fire in a charging vehicle is detected, it can automatically issue an early warning without manual intervention, thus improving early warning efficiency.

[0145] 5. By conducting big data analysis on sound and gas data from multiple charging stations, it is possible to predict the trend and potential hazards of fires, thereby improving the safety of charging stations.

[0146] 6. By combining artificial intelligence technology, deep learning and optimization can be performed on sound and gas data. With the accumulation of data, recognition accuracy and precision can be continuously improved. Based on internet technology, remote monitoring and management of charging vehicles at charging stations nationwide can be achieved, including early warning of battery thermal runaway fires.

[0147] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0148] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A battery thermal runaway early warning system for charging vehicles, characterized in that, The battery thermal runaway early warning system for the charging vehicle includes: The data acquisition module is installed on the ground of the parking space in the charging station to sense in real time whether a vehicle enters the parking space and to collect environmental data when a vehicle enters the parking space; the environmental data includes sound information and gas information. The local analysis module, connected to the data acquisition module, is used to determine whether the gas released by the charging vehicle is abnormal based on the gas information. If so, a first warning signal is generated. The sound information is filtered to obtain target sound information, and the target sound information is compared with the abnormal sound information. If the degree of match between the target sound information and the abnormal sound information is greater than a first threshold, a first warning signal is generated. A cloud-based analysis module, connected to the local analysis module, is used to determine whether the target sound information is an abnormal sound when the degree of matching between the target sound information and the abnormal sound information is greater than a first threshold. If the matching degree is greater than a first threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If so, a second warning signal and a power-off command are generated; otherwise, a power-off command is generated. When the degree of matching between the target sound information and the abnormal sound information is less than or equal to the first threshold and greater than a second threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If the matching degree is greater than a first threshold, a first warning signal and a power-off command are generated; otherwise, a warning cancellation command is generated. The second threshold is less than the first threshold. The cloud analysis module is also used to generate battery information for the corresponding vehicle based on the environmental data, store the battery information for the corresponding vehicle in the database, and predict the battery life based on the battery information for the corresponding vehicle stored in the database. An alarm module is installed on the charging pile and connected to the local analysis module and the cloud analysis module respectively. It is used to trigger an alarm based on the first warning signal or the second warning signal, and to stop the alarm based on the cancellation warning command. The power-off execution module is connected to both the charging pile and the cloud analysis module, and is used to cut off the charging power of the charging pile according to the power cut-off command, thereby stopping the charging of the vehicle.

2. The battery thermal runaway early warning system for charging vehicles according to claim 1, characterized in that, The data acquisition module includes: an upper cover, a lower cover, a geomagnetic sensor, multiple microphone sensors, and multiple gas sensors; The lower surface of the upper cover is fixed to the upper surface of the lower cover; the upper cover has a detection hole; the upper cover is arc-shaped. The geomagnetic sensor, each microphone sensor, and each gas sensor are all installed on the lower surface of the upper cover; the geomagnetic sensor is used to detect in real time whether a vehicle has entered the parking space; the microphone sensor is used to collect sound information when a vehicle enters the parking space; and the gas sensor is used to collect gas information when a vehicle enters the parking space.

3. The battery thermal runaway early warning system for charging vehicles according to claim 2, characterized in that, The number of microphone sensors is 3; the 3 microphone sensors are arranged in an equilateral triangle on the lower surface of the upper cover; the number of gas sensors is 2; the 2 gas sensors are arranged diagonally on the lower surface of the upper cover.

4. The battery thermal runaway early warning system for charging vehicles according to claim 2, characterized in that, The number of microphone sensors is 4; of which, 3 microphone sensors are arranged in an equilateral triangle on the lower surface of the upper cover, and 1 microphone sensor is located at the center of the upper cover; the number of gas sensors is 3; the 3 gas sensors are respectively located at the midpoint of each side of the equilateral triangle.

5. The battery thermal runaway early warning system for charging vehicles according to claim 2, characterized in that, The detection hole is covered with a hydrophobic and breathable membrane.

6. The battery thermal runaway early warning system for charging vehicles according to claim 2, characterized in that, The data acquisition module also includes: A temperature sensor is installed on the lower surface of the upper cover to detect the ambient temperature in real time. The local analysis module is also connected to the temperature sensor, and the local analysis module is also used to determine whether the ambient temperature exceeds the temperature threshold. If so, a third warning signal is generated. The alarm module is also used to issue an alarm based on the third warning signal.

7. The battery thermal runaway early warning system for charging vehicles according to claim 2, characterized in that, The local analysis module is a control circuit board; the control circuit board is disposed on the lower surface of the upper cover.

8. The battery thermal runaway early warning system for charging vehicles according to claim 1, characterized in that, The local analysis module includes: The gas judgment submodule is connected to the data acquisition module and is used to determine whether the gas concentration in the gas information is greater than a set gas concentration threshold and whether the time for which the gas concentration is greater than the gas concentration threshold is greater than a set time threshold. If so, a first warning signal is generated. The sound filtering submodule, connected to the data acquisition module, is used to determine the direction and location of the sound information using a direction-of-arrival estimation method, and to filter the sound information based on the direction and location of the sound information to obtain the target sound information; The sound comparison submodule, connected to the sound filtering submodule, is used to compare the target sound information with the abnormal sound information. If the match between the target sound information and the abnormal sound information is greater than a first threshold, a first warning signal is generated. The abnormal sound information includes the sound of a battery pack rupture and the sound of a safety valve rupture.

9. The battery thermal runaway early warning system for charging vehicles according to claim 1, characterized in that, The alarm module is a magnetic audible and visual alarm.

10. A method for early warning of thermal runaway in a charging vehicle battery, characterized in that, The method for early warning of thermal runaway of charging vehicle batteries includes: The system can detect in real time whether a vehicle has entered a parking space. If a vehicle does enter a parking space, environmental data is collected. The environmental data includes sound information and gas information. Based on the gas information, determine whether the gas released by the charging vehicle is abnormal. If so, generate a first warning signal and issue a level one alarm. The sound information is filtered to obtain the target sound information; The target sound information is compared with the abnormal sound information locally. If the degree of match between the target sound information and the abnormal sound information is greater than a first threshold, a level one alarm is triggered, and the target sound information is sent to the cloud server. When the degree of match between the target sound information and the abnormal sound information is greater than a first threshold, the cloud server uses a deep learning model to determine whether the target sound information is an abnormal sound. If it is, a second-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle; otherwise, a first-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle. When the degree of agreement between the target sound information and the abnormal sound information is less than or equal to the first threshold and greater than the second threshold, a deep learning model is used to determine whether the target sound information is an abnormal sound. If so, a first-level alarm is triggered, and the charging power of the charging pile is cut off to stop charging the vehicle. Otherwise, stop the alarm; the second threshold is less than the first threshold. The cloud server generates battery information for the corresponding vehicle based on the environmental data, stores the battery information for the corresponding vehicle in the database, and predicts the battery life based on the battery information for the corresponding vehicle stored in the database.

Images