Multi-data fusion lithium battery thermal runaway detector and detection method

Through the lithium battery thermal runaway detector with multi-data fusion, using deep learning models and data analysis of multiple sensors, the problem of low accuracy of thermal runaway detection in the existing technology is solved, and more accurate and early warnings are achieved. Reduce false alarm rate.

CN119087263APending Publication Date: 2024-12-06ANHUI ELECTRIC POWER DESIGN INST CEEC
View PDF 0 Cites 4 Cited by

Patent Information

Application Number
CN202411188768.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-28
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

The detection accuracy of existing lithium battery thermal runaway detectors is low, and single parameter detection and simple threshold determination cannot fully reflect the battery health status and potential risks, and there are problems such as slow response speed and high false alarm rate.

Method used

The lithium battery thermal runaway detector with multi-data fusion is adopted, including a hollow shell, a microprocessor and a variety of sensors. Data is collected through dual-band dual-optical smoke sensors and other sensors, and a comprehensive analysis is performed using a deep learning-based thermal runaway detection model. Get the detection data in the sliding window in real time for thermal runaway detection.

Benefits of technology

It improves the accuracy and anti-interference ability of thermal runaway detection of lithium batteries, achieves earlier warnings and reduces false alarms, and enhances monitoring of the health status of lithium batteries.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119087263A_ABST
    Figure CN119087263A_ABST
Patent Text Reader

Abstract

The invention relates to a multi-data fusion lithium battery thermal runaway detector and detection method.The lithium battery thermal runaway detector comprises a hollow shell, a microprocessor and multiple sensors, the microprocessor and the multiple sensors are installed in the shell, and one sensor is a dual-waveband dual-light-path smoke sensor; the dual-band dual-optical-path smoke sensor comprises two detection light sources, a first full-band photodiode and a second full-band photodiode, the two detection light sources are arranged in parallel and are a 940nm red light emitting diode and a 450nm blue light emitting diode respectively, and the included angle between the first full-band photodiode and the detection light source is 135 degrees. The included angle between the second full-wave band photodiode and the detection light source is 66.5 degrees; and the microprocessor is configured to perform thermal runaway detection on the lithium battery through a thermal runaway detection model based on deep learning based on various detection data acquired by the various sensors. The problem that an existing lithium battery thermal runaway detector is low in detection accuracy is solved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of lithium batteries, and in particular to a multi-data fusion lithium battery thermal runaway detector and detection method. Background Art

[0002] Currently, lithium battery monitoring technologies on the market mainly focus on the detection of a single parameter, and only use simple thresholds to determine the warning level, such as temperature monitoring, voltage monitoring, single gas concentration, etc. Although these single parameter monitoring methods and simple threshold algorithms can prevent thermal runaway to a certain extent, due to the complexity of thermal runaway of lithium batteries and the interference of the surrounding environment, relying solely on a single parameter cannot fully reflect the health status and potential risks of the battery, and there are shortcomings such as slow response speed and high false alarm rate. Summary of the invention

[0003] The present invention provides a multi-data fusion lithium battery thermal runaway detector to solve the problem of low detection accuracy of existing lithium battery thermal runaway detectors.

[0004] In a first aspect, the present invention provides a multi-data fusion lithium battery thermal runaway detector, comprising a hollow shell, a microprocessor and a plurality of sensors installed inside the shell, wherein one of the sensors is a dual-band dual-optical path smoke sensor;

[0005] The dual-band dual-light path smoke sensor comprises two detection light sources and a first full-band photodiode and a second full-band photodiode. The two detection light sources are arranged in parallel and are respectively a 940nm red light emitting diode and a 450nm blue light emitting diode. The angle between the first full-band photodiode and the detection light source is 135°, and the angle between the second full-band photodiode and the detection light source is 66.5°. The two detection light sources are configured to light up alternately.

[0006] The microprocessor is configured to perform thermal runaway detection on the lithium battery based on the various detection data collected by the various sensors and through a thermal runaway detection model based on deep learning.

[0007] In a second aspect, the present invention provides a multi-data fusion lithium battery thermal runaway detection method, which is applied to a thermal runaway detector, wherein the thermal runaway detector includes a hollow shell and a microprocessor and multiple sensors installed inside the shell;

[0008] The thermal runaway detection method comprises:

[0009] Acquire multiple detection data collected by the multiple sensors in real time;

[0010] Using the multiple detection data in the real-time sliding window as current input of a thermal runaway detection model based on machine learning;

[0011] The current thermal runaway detection result of the lithium battery is obtained through the thermal runaway detection model.

[0012] Compared with the detection of a single parameter in the prior art and determining the warning level only by a simple threshold, the present invention performs thermal runaway detection by fusing the detection data of multiple sensors through a thermal runaway detection model based on deep learning. First, the integration of multiple detection data for thermal runaway detection has better anti-interference capability than thermal runaway detection using a single detection data. Secondly, the use of a thermal runaway detection model based on deep learning for thermal runaway detection also has better anti-interference capability than simple threshold judgment. In summary, the multi-data fusion lithium battery thermal runaway detector provided by the present invention has a strong anti-interference capability, and can thus achieve more accurate lithium battery thermal runaway detection, solving the problem of low detection accuracy of existing lithium battery thermal runaway detectors.

[0013] Details of one or more embodiments of the present application are set forth in the following drawings and description to make other features, objects, and advantages of the present application more readily apparent. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 is a schematic diagram of a housing of a lithium battery thermal runaway detector in some embodiments of the present invention;

[0015] Figure 2 is a side view of a dual-band dual-optical path smoke sensor in some embodiments of the present invention;

[0016] Figure 3 is a top view of a dual-band dual-optical path smoke sensor in some embodiments of the present invention;

[0017] Figure 4 is a front schematic diagram of a circuit board in some embodiments of the present invention;

[0018] Figure 5 is a schematic diagram of the back side of a circuit board in some embodiments of the present invention;

[0019] Figure 6 is a system diagram of a dual-band dual-optical path smoke sensor in some embodiments of the present invention;

[0020] Figure 7 is a schematic diagram of a process of intercepting detection data with a real-time sliding window in some embodiments of the present invention;

[0021] Figure 8 Schematic diagram of the structure of a thermal runaway detection model in some embodiments of the present invention. DETAILED DESCRIPTION

[0022] In order to more clearly understand the purpose, technical solutions and advantages of the present application, the present application is described and illustrated below in conjunction with the accompanying drawings and embodiments.

[0023] Unless otherwise defined, the technical terms or scientific terms involved in this application shall have the general meaning understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the", "these" and the like in this application do not represent quantitative restrictions, and they can be singular or plural. The terms "include", "comprise", "have" and any variants thereof involved in this application are intended to cover non-exclusive inclusions; for example, a process, method and system, product or device comprising a series of steps or modules (units) is not limited to the listed steps or modules (units), but may include unlisted steps or modules (units), or may include other steps or modules (units) inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether directly or indirectly. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that there can be three relationships. For example, "A and / or B" can mean: A exists alone, A and B exist at the same time, and B exists alone. Usually, the character " / " indicates that the objects associated with each other are in an "or" relationship. The terms "first", "second", "third", etc. involved in this application are only used to distinguish similar objects and do not represent a specific ordering of the objects.

[0024] Reference Figures 1 to 6 In the present invention, a multi-data fusion lithium battery thermal runaway detector is provided, which includes a hollow shell 100 and a microprocessor 700 and multiple sensors installed inside the shell 100.

[0025] The above-mentioned lithium battery thermal runaway detector integrates multiple sensors, which is a composite detector. Different information data are detected and collected by multiple sensors respectively, and these information data are transmitted to the microprocessor. The microprocessor fuses and analyzes these information data through a thermal runaway detection model based on deep learning, and gives the detection result of thermal runaway. The detection result of thermal runaway includes whether the lithium battery has thermal runaway, whether the lithium battery is about to have thermal runaway, and whether the lithium battery has the risk of thermal runaway, etc., and then a thermal runaway warning can be given, which can be determined according to the specific needs of thermal runaway monitoring. The thermal runaway detection model is a deep learning model, which needs to be trained in advance. Compared with the detection of a single parameter in the prior art and only determining the warning level by a simple threshold, the present invention uses a thermal runaway detection model based on deep learning to fuse the detection data of multiple sensors for thermal runaway detection. First, the thermal runaway detection is performed by integrating multiple detection data, which has better anti-interference ability than the thermal runaway detection of a single detection data. Secondly, the thermal runaway detection is performed by using a thermal runaway detection model based on deep learning, which also has better anti-interference ability than simple threshold judgment. In summary, the multi-data fusion lithium battery thermal runaway detector provided by the present invention has strong anti-interference ability, and can thus achieve more accurate lithium battery thermal runaway detection, solving the problem of low detection accuracy of existing lithium battery thermal runaway detectors.

[0026] Among them, the outer shell 100 adopts a hollow structure, which is conducive to the diffusion of gas and smoke into the interior of the detector, reducing the detection delay time and improving the detection response speed.

[0027] In some embodiments, the multiple sensors include a dual-band dual-optical path smoke sensor 200, a negative temperature coefficient sensor 300, a carbon monoxide sensor 400, a hydrogen sensor 500, and a volatile organic compound sensor 600. Through the above sensors, the lithium battery thermal runaway detector can respectively detect environmental parameters such as smoke, temperature, nitric oxide content, hydrogen content, etc. that can be used for lithium battery thermal runaway warning.

[0028] Optionally, the negative temperature coefficient sensor 300 uses a thermistor as a detection element; the model of the carbon monoxide sensor 400 is ZE16B; the model of the hydrogen sensor 500 is ZE07; and the model of the volatile organic compound sensor 600 is ZP16. The dual-band dual-optical path smoke sensor 200, the carbon monoxide sensor 400, the hydrogen sensor 500, and the volatile organic compound sensor 600 communicate with the microprocessor through UART; the negative temperature coefficient sensor 300 outputs a voltage analog quantity, and the microprocessor 700 obtains temperature data by collecting and calculating the voltage analog quantity.

[0029] Reference Figure 2 and 3, wherein the preferred dual-band dual-light path smoke sensor 200 includes two detection light sources and a first full-band photodiode and a second full-band photodiode, the two detection light sources are arranged in parallel and are respectively a 940nm red light emitting diode and a 450nm blue light emitting diode, the angle between the first full-band photodiode and the detection light source is 135°, and the angle between the second full-band photodiode and the detection light source is 66.5°. In this embodiment, a new dual-band dual-light path smoke sensor is provided, in which a full-band photodiode and the two detection light sources have an angle of 135° to form a forward scattering type smoke sensor, and another full-band photodiode and the two detection light sources have an angle of 66.5° to form a backward scattering type smoke sensor.

[0030] Furthermore, the first full-band photodiode is located on a side of the second full-band photodiode away from the two detection light sources; the first full-band photodiode is equidistant from the two detection light sources; and the second full-band photodiode is equidistant from the two detection light sources.

[0031] First, since the angle between the first full-band photodiode and the detection light source is larger than that between the second full-band photodiode and the detection light source, setting the second full-band photodiode between the detection light source and the first full-band photodiode can optimize the spatial layout of the three and occupy less space.

[0032] Furthermore, whether it is the first full-band photodiode or the second full-band photodiode, the distances to the red light emitting diode and the blue light emitting diode are equal, thereby facilitating comparative analysis of the red light data and the blue light data generated by the full-band photodiode.

[0033] At the same time, there are both forward scattering smoke sensors and backscattering smoke sensors in the dual-band dual-optical path smoke sensor 200. Therefore, the dual-band dual-optical path smoke sensor can simultaneously obtain forward scattered red and blue light data and backscattered red and blue light data.

[0034] Based on Mie scattering theory, the size of particles is distinguished by parameters such as the forward and backward scattering ratio and the red-blue light scattering ratio. According to theoretical analysis, for large dust particles, the scattering intensity of red light and blue light is basically the same, and the ratio is close to 1. For small dust particles, the scattering intensity of blue light is stronger than that of red light. Then, through the information data collected by the dual-band dual-optical path smoke sensor, large dust particles can be distinguished, thereby reducing the interference of large dust particles and improving the accuracy of smoke judgment.

[0035] It should be noted that the microprocessor 700 and a variety of sensors can be integrated on the same circuit board, which is installed inside the housing 100. In order to realize the communication between the thermal runaway detector and the host computer, the thermal runaway detector also includes a CAN communication chip 800, which is used to realize the communication between the microprocessor 700 and the host computer. The CAN communication chip 800 is also integrated on the circuit board to send the data generated by the microprocessor 700 or the original data collected by the sensor to the host computer. In order to meet some basic usage requirements, the circuit board should also integrate components such as a power module 910, a reset button 920, a crystal oscillator 930, an indicator light 940 and a terminal block 950.

[0036] As mentioned above, the specific structure of the multi-data fusion lithium battery thermal runaway detector in some embodiments is introduced. Different environmental information is collected mainly through a variety of sensors, so that the microprocessor can integrate more data for thermal runaway analysis. Among them, a new dual-band dual smoke sensor is provided, which can improve the detection accuracy of smoke compared with traditional smoke sensors. In addition, the present invention also optimizes the data processing of the microprocessor, thereby further improving the detection accuracy of thermal runaway.

[0037] In some embodiments, the thermal runaway detection step of the lithium battery includes:

[0038] Step S110, acquiring various detection data collected by various sensors in real time.

[0039] Step S120, using the various detection data in the real-time sliding window as current input of a thermal runaway detection model based on machine learning.

[0040] Step S130, obtaining the current thermal runaway detection result of the lithium battery through the thermal runaway detection model.

[0041] The time end point of the real-time sliding window is the current moment, and its position is continuously updated with the change of the current moment. Therefore, this embodiment packages a variety of detection data within the current moment and a certain moment before it as the current input of the model.

[0042] By executing the above steps, the microprocessor can realize real-time thermal runaway detection. First, the detection data is analyzed by the thermal runaway detection model based on machine learning, which has a better data analysis effect than the existing simple threshold judgment. At the same time, in one input of the thermal runaway detection model, it not only includes a variety of detection data, but also includes a variety of detection data at multiple consecutive moments, so that the thermal runaway detection model can not only analyze the relationship between different data for thermal runaway judgment, but also analyze the change trend of various data for thermal runaway judgment. Therefore, using a variety of detection data in the real-time sliding window as the current input of the thermal runaway detection model can further improve the thermal runaway detection results of the thermal runaway detection model compared to using the detection data at a single moment or a single type of detection data as the current input of the thermal runaway detection model.

[0043] Reference Figure 7 , wherein optionally, the multiple sensors are configured to: send detection data to the microprocessor once every 2 seconds; and / or, the time length of the real-time sliding window is 120 seconds. At this time, each input of the thermal runaway detection model includes the detection data of the most recent 60 moments.

[0044] Reference Figure 8 Specifically, the thermal runaway detection model includes a preprocessing layer, multiple linear layers (Linear), a fully connected layer (Fully Connection) and a Softmax layer connected in sequence, and the number of linear layers is less than 10. At the same time, the output of the linear layer is residually connected to the output. The preprocessing layer is mainly used to implement normalization and filtering operations (Normalze&Filter), and each linear layer has an activation function Relu. The thermal runaway detection model only uses a few layers of linear layers and residual structures, and requires few computing resources. It can be run on an ordinary MCU (such as stm32f103), which greatly reduces the hardware cost while improving the accuracy. Preferably, the number of linear layers is between 3-5. When the linear layers are between 3-5 layers, the accuracy and structure can be better balanced, that is, the model structure can be simplified as much as possible while ensuring high accuracy.

[0045] As described above, at least one embodiment of each part of the multi-data fusion lithium battery thermal runaway detector provided by the present invention has been described separately. It should be noted that the various parts of the multi-data fusion lithium battery thermal runaway detector can be combined using different embodiments to form different embodiments of the multi-data fusion lithium battery thermal runaway detector.

[0046] For example, in a specific overall embodiment, a multi-data fusion lithium battery thermal runaway detector includes:

[0047] Dual-band dual-optical path smoke sensor, NTC sensor, CO sensor, H2 sensor, VOC sensor, circuit board, indicator light, shell. Dual-band dual-optical path smoke sensor is newly proposed by the present invention, using 940nm red light emitting diode and 450nm blue light emitting diode to light up alternately as two detection light sources, two full-band photodiodes to convert smoke scattering signals into voltage, one forward scattering at an angle of 135 degrees with the emission light source, and the other backward scattering at an angle of 66.5 degrees with the emission light source. NTC sensor uses NTC thermistor sensor module to output analog quantity; CO sensor uses ZE16B; H2 sensor uses ZE07; VOC sensor uses ZP16; communication chip uses TJA1050T / CM CAN communication chip.

[0048] Among them, NTC temperature sensor: uses NTC thermistor module to output analog quantity. CO sensor: uses electrochemical principle, has high sensitivity and fast response speed, and communicates with the chip via UART. H2 sensor: uses electrochemical principle and communicates with the chip via UART. VOC sensor: uses semiconductor gas sensor and communicates with the chip via UART. The current status of the lithium battery is judged by using the neural network algorithm through the smoke, gas concentration and temperature information captured by the detector.

[0049] Among them, the dual-band dual-optical path smoke sensor, CO sensor, H2 sensor and VOC sensor all communicate with the MCU through UART (Universal Asynchronous Receiver / Transmitter); the NTC sensor outputs an analog voltage, and the MCU collects and calculates the temperature value through analog quantity.

[0050] The detector collects red and blue light forward and backward scattering voltage, temperature, and gas concentrations of carbon monoxide, hydrogen, volatile organic compounds, etc., and uses a sliding window neural network algorithm to extract all data every 1 second and the previous 120 seconds after the start time 2 minutes later, forming a matrix input thermal runaway detection model, and classifying the current state into three categories. The detection precision and accuracy are high, and very little MCU memory is occupied to realize the operation of low-cost neural network algorithms on the detector.

[0051] Specifically, since the detector is designed to feed back a set of parameters to the host computer every 2 seconds, and each sensor is designed to feed back a set of data to the MCU every 2 seconds, a sliding window of y×60 (y represents the number of sensors, and 60 represents the number of moments) is used to read the data at this moment and the previous 120 seconds, that is, a y×60 matrix is ​​taken out every second using the sliding window, and then these matrices are stacked as parameters for machine learning or judgment; this method can extract the connection between time changes and feature parameters and the connection between each feature parameter. The amount of information extracted far exceeds that of a simple threshold algorithm, avoiding false alarms caused by using only threshold composite judgment; and the machine learning model used in this method only uses a few linear layers and residual structures, requires few computing resources, and can be run on an ordinary MCU (such as stm32f103), while improving accuracy, greatly reducing hardware costs.

[0052] In addition, based on the Mie scattering theory, the size of particles is distinguished by parameters such as the forward and backward scattering ratio and the red-blue light scattering ratio. According to theoretical analysis, for large dust particles, the scattering intensity of red and blue light is basically the same, and the ratio is close to 1. For small dust particles, the scattering intensity of blue light is stronger than that of red light. This method can reduce the interference of large dust particles.

[0053] In this embodiment, the working principle of the lithium battery thermal runaway detector is:

[0054] Because this detector is a diffusion type, the detector shell adopts a hollow structure, which is conducive to the diffusion of gas and smoke into the interior of the detector, reducing the detection delay time and improving the detection response speed.

[0055] Except for the NTC sensor, the communication between each sensor and the main detector chip (i.e., the microprocessor or the circuit board with the microprocessor integrated) adopts the UART communication mode. The NTC sensor calculates the temperature by collecting analog quantity. The main detector chip communicates with the host computer through the CAN bus.

[0056] 120s of data are collected through a sliding window as learning samples (two-dimensional tensor) to input into the neural network for training, instead of using multiple data at a certain moment as parameters. The reason is to learn how each parameter changes over time and enhance the detector's anti-interference ability.

[0057] The input two-dimensional tensor is normalized and filtered, then straightened and trained using multiple linear layers and fully connected layers. The entire network has few layers and parameters, and is easily ported into the stm32f103 chip.

[0058] The working process of the detector is as follows: before thermal runaway, the lithium battery cell releases a small amount of gas, which diffuses into the detector. The gas sensor detects a trace concentration of characteristic gas. The forward and backward scattering ratio and the red and blue light scattering ratio of the smoke sensor fluctuate. The NTC sensor fluctuates. The detector records various data and inputs the data at this moment and the data 120 seconds before as parameters into the trained machine learning algorithm, and derives the status classification and makes corresponding warnings.

[0059] In this embodiment, the beneficial effects of the lithium battery thermal runaway detector are:

[0060] 1. Very early warning of thermal runaway of lithium batteries: The existing technology uses a combination of smoke detectors and temperature detectors to detect, which cannot provide early warning of thermal runaway of lithium batteries. The present invention partially uses gas sensors, which can detect abnormalities in lithium batteries at the earliest stage.

[0061] 2. Reduce false alarm rate and resist interference: This is mainly reflected in three aspects. The first is the use of smoke, temperature and gas composite detection. Compared with a single parameter, the false alarm rate will be greatly reduced; the second is the lightweight machine learning algorithm used. Compared with the traditional threshold algorithm, it comprehensively judges the trend changes of parameters over a period of time and eliminates the false alarms caused by interference with the parameters; the third is that the smoke sensor used in the present invention is a dual-light source and dual-optical path smoke sensor. Compared with the traditional single-light source and single-optical path sensor, it can distinguish between smoke and interference sources to a certain extent.

[0062] The present invention also provides a multi-data fusion lithium battery thermal runaway detection method, which is applied to a thermal runaway detector. The thermal runaway detector includes a hollow shell and a microprocessor and multiple sensors installed inside the shell.

[0063] Thermal runaway detection methods include:

[0064] Step S110, acquiring various detection data collected by various sensors in real time.

[0065] Step S120, using the various detection data in the real-time sliding window as current input of a thermal runaway detection model based on machine learning.

[0066] Step S130, obtaining the current thermal runaway detection result of the lithium battery through the thermal runaway detection model.

[0067] The time end point of the real-time sliding window is the current moment, and its position is continuously updated with the change of the current moment. Therefore, this embodiment packages a variety of detection data within the current moment and a certain moment before it as the current input of the model.

[0068] By executing the above steps, the microprocessor can realize real-time thermal runaway detection. First, the detection data is analyzed by the thermal runaway detection model based on machine learning, which has a better data analysis effect than the existing simple threshold judgment. At the same time, in one input of the thermal runaway detection model, it not only includes a variety of detection data, but also includes a variety of detection data at multiple consecutive moments, so that the thermal runaway detection model can not only analyze the relationship between different data for thermal runaway judgment, but also analyze the change trend of various data for thermal runaway judgment. Therefore, using a variety of detection data in the real-time sliding window as the current input of the thermal runaway detection model can further improve the thermal runaway detection results of the thermal runaway detection model compared to using the detection data at a single moment or a single type of detection data as the current input of the thermal runaway detection model.

[0069] Optionally, the multiple sensors are configured to: send detection data to the microprocessor once every 2 seconds; and / or the time length of the real-time sliding window is 120 seconds. In this case, each input of the thermal runaway detection model includes the detection data of the most recent 60 moments.

[0070] Specifically, the thermal runaway detection model includes a preprocessing layer, multiple linear layers (Linear), a fully connected layer (Fully Connection) and a Softmax layer connected in sequence, and the number of linear layers is less than 10. At the same time, the output of the linear layer is residually connected to the output. The preprocessing layer is mainly used to implement normalization and filtering operations (Normalze&Filter), and each linear layer has an activation function Relu. The thermal runaway detection model only uses a few linear layers and residual structures, and requires few computing resources. It can be run on an ordinary MCU (such as stm32f103), which greatly reduces the hardware cost while improving the accuracy. Preferably, the number of linear layers is between 3-5. When the linear layers are between 3-5 layers, the accuracy and structure can be better balanced, that is, the model structure can be simplified as much as possible while ensuring high accuracy.

[0071] It should be understood that the specific embodiments described herein are only used to explain the application, rather than to limit it. Based on the embodiments provided in this application, all other embodiments obtained by ordinary technicians in this field without creative work are within the protection scope of this application.

[0072] Obviously, the drawings are only some examples or embodiments of the present application. For ordinary technicians in the field, the present application can also be applied to other similar situations based on these drawings without creative work. In addition, it is understandable that although the work done in this development process may be complicated and lengthy, for ordinary technicians in the field, certain changes in design, manufacturing or production based on the technical content disclosed in this application are only conventional technical means and should not be regarded as insufficient content disclosed in this application.

Claims

1. A multi-data fusion lithium battery thermal runaway detector, characterized in that: It comprises a hollow shell, a microprocessor and a plurality of sensors installed inside the shell, wherein one of the sensors is a dual-band dual-optical path smoke sensor; The dual-band dual-light path smoke sensor comprises two detection light sources and a first full-band photodiode and a second full-band photodiode. The two detection light sources are arranged in parallel and are respectively a 940nm red light emitting diode and a 450nm blue light emitting diode. The angle between the first full-band photodiode and the detection light source is 135°, and the angle between the second full-band photodiode and the detection light source is 66.5°. The two detection light sources are configured to light up alternately. The microprocessor is configured to: based on the various detection data collected by the various sensors, perform thermal runaway detection on the lithium battery through a thermal runaway detection model based on deep learning.

2. The multi-data fusion lithium battery thermal runaway detector according to claim 1, characterized in that: The thermal runaway detection step of the lithium battery comprises: Acquire multiple detection data collected by the multiple sensors in real time; Using the multiple detection data in the real-time sliding window as current input of a thermal runaway detection model based on machine learning; The current thermal runaway detection result of the lithium battery is obtained through the thermal runaway detection model.

3. The multi-data fusion lithium battery thermal runaway detector according to claim 2 is characterized in that: The plurality of sensors are configured to: send detection data to the microprocessor once every 2 seconds; And / or, the time length of the real-time sliding window is 120 seconds.

4. The multi-data fusion lithium battery thermal runaway detector according to claim 2, characterized in that: The thermal runaway detection model includes a preprocessing layer, multiple linear layers, a fully connected layer and a Softmax layer connected in sequence, and the number of the linear layers is less than 10.

5. The multi-data fusion lithium battery thermal runaway detector according to claim 2, characterized in that: The number of the linear layers is between 3 and 5.

6. The multi-data fusion lithium battery thermal runaway detector according to claim 1, characterized in that: The first full-band photodiode is located at a side of the second full-band photodiode away from the two detection light sources; The first full-band photodiode is equidistant from the two detection light sources; The second full-band photodiode is equidistant from the two detection light sources.

7. The multi-data fusion lithium battery thermal runaway detector according to any one of claims 1 to 5, characterized in that: The various sensors also include negative temperature coefficient sensors, carbon monoxide sensors, hydrogen sensors, and volatile organic compound sensors.

8. The multi-data fusion lithium battery thermal runaway detector according to claim 7, characterized in that: The negative temperature coefficient sensor uses a thermistor as a detection element; The model of the carbon monoxide sensor is ZE16B; The model of the hydrogen sensor is ZE07; The model of the volatile organic compound sensor is ZP16.

9. The multi-data fusion lithium battery thermal runaway detector according to claim 8, characterized in that: The dual-band dual-optical path smoke sensor, the carbon monoxide sensor, the hydrogen sensor and the volatile organic compound sensor communicate with the microprocessor via UART; The negative temperature coefficient sensor outputs a voltage analog quantity, and the microprocessor obtains temperature data by collecting and calculating the voltage analog quantity.

10. A multi-data fusion lithium battery thermal runaway detection method, characterized in that: Applied to a thermal runaway detector, the thermal runaway detector comprises a hollow shell and a microprocessor and a plurality of sensors installed inside the shell; The thermal runaway detection method comprises: Acquire multiple detection data collected by the multiple sensors in real time; Using the multiple detection data in the real-time sliding window as current input of a thermal runaway detection model based on machine learning; The current thermal runaway detection result of the lithium battery is obtained through the thermal runaway detection model.

Citation Information

Cited By

  • Battery box thermal runaway detection method and system

    CN120473591A

  • Battery box thermal runaway detection method and system

    CN120473591B

  • Lithium battery thermal runaway detection system based on multi-sensor fusion

    CN120629993A

  • Anti-interference method, device and circuit for dual-band photoelectric smoke sensor of energy storage power station

    CN122618740A