Lithium battery early warning system based on multi-modal data fusion

Abstract

The invention relates to a lithium battery intelligent early warning system based on multi-modal data fusion, and relates to the technical field of battery safety monitoring, the field of multi-modal sensor fusion and the like. A lithium battery intelligent early warning system based on multi-modal data fusion comprises a shell, a multi-modal sensing module, an embedded microcontroller module, an alarm buzzer and a cloud decision platform, the shell is made of a high-temperature-resistant material, the top of the shell is provided with an air inlet, an air inlet fan and a touch display screen, and the side face of the shell is provided with an exhaust port; the multi-mode sensing module and the embedded microcontroller module are integrated on a circuit; the multi-mode sensing module comprises a power supply circuit, a gas sensing array, a thermodynamic monitoring unit, a mechanical deformation detection unit, an acoustic characteristic acquisition unit and an environment temperature and humidity sensing unit; the embedded microcontroller module is configured with a Wi-Fi communication interface and a serial communication capability; and the cloud decision platform deploys an intelligent diagnosis system based on an improved KNN-GRU hybrid model. The system provided by the invention realizes the omnibearing monitoring of the safety state of the lithium battery, and is particularly suitable for the real-time safety monitoring of the power battery pack of the electric vehicle.

Description

Technical Field

[0001] The present invention relates to the technical fields of battery safety monitoring, multimodal sensor fusion, etc. Background Art

[0002] Lithium batteries have become the core power source in the fields of new energy storage and electric vehicles due to their advantages such as high energy density and long cycle life. However, the complex electrochemical reactions inside and external abuse conditions (such as overcharging, high temperature, mechanical shock) are likely to trigger a thermal runaway chain reaction, leading to safety accidents such as fire and explosion, seriously threatening life and property safety. Existing safety management technologies mainly focus on the monitoring of battery monomer parameters, such as abnormal judgment through external characteristic signals such as voltage, temperature, and pressure. However, such solutions do not solve the problem of multi-battery collaborative analysis. In addition, existing early warning strategies mostly rely on a single physical quantity, ignoring characteristic signals such as gas composition and internal stress evolution. Therefore, there is an urgent need to develop a lithium battery safety monitoring and early warning system with both early warning capabilities and group risk prediction functions. Through multi-dimensional signal fusion, the early warning accuracy can be improved, and the bottleneck of monomer monitoring can be broken through. Summary of the Invention

[0003] The purpose of the present invention is to provide an intelligent early warning system that is conducive to high-precision and comprehensive analysis of the state of lithium batteries, mainly solving the problems of single means of existing lithium battery safety monitoring and lagging early warning response. Through the collaborative work of multi-dimensional perception and intelligent decision-making, the system realizes real-time monitoring and hierarchical early warning of the development process of lithium battery thermal runaway.

[0004] To achieve the above purpose, the technical solution adopted by the present invention is as follows: A lithium battery intelligent early warning system based on multi-modal data fusion, including a high-temperature resistant housing, a multi-modal sensing module, an embedded microcontroller module, an alarm buzzer, and a cloud decision-making platform. The top of the housing is provided with an air inlet, an air intake fan, and a touch display screen. The display screen is used to display the evaluation results, and the exhaust port is arranged on the side. Each functional module realizes hardware interconnection through circuit integration;

[0005] Furthermore, the multimodal sensing module includes a power supply circuit and five sensing units: a gas sensing array composed of 14 semiconductor gas sensors, which can detect thermal runaway characteristic gases such as C1-C4 alkanes, hydrogen, carbon monoxide, ammonia, acetone, sulfur dioxide, and electrolyte decomposition products; the thermodynamic monitoring unit uses a thermocouple sensor with a measurement range of -20°C to 200°C to monitor the temperature of the gas around the lithium battery; the mechanical deformation detection unit is configured with XY-axis double strain sensors to detect the surface mechanical deformation caused by internal stress changes before the thermal runaway of the lithium battery; the acoustic feature acquisition unit is equipped with a sound sensor to collect the acoustic feature signals released before the thermal runaway of the lithium battery; the ambient temperature and humidity sensing unit is equipped with a temperature and humidity sensor to collect the temperature and humidity data of the gas around the lithium battery. Each sensor transmits 0-5V analog-to-digital signals through serial communication;

[0006] Furthermore, the embedded microcontroller module is configured with a Wi-Fi communication interface and serial communication capabilities. This module collects multimodal sensing data at a period of 100 ms, and uploads the synchronously obtained 14+1+2+1+2 = 20-dimensional eigenvalue vector to the cloud decision platform in real time through Wi-Fi.

[0007] Furthermore, the improved KNN-GRU hybrid model deployed in the cloud performs feature matching and decision analysis based on a pre-trained database, and outputs three types of evaluation results: the type and concentration of characteristic gases, the thermal runaway development rate (°C / min), and the safety status (normal / warning / danger). When it is determined to be in the "warning" or "danger" state, the embedded microcontroller drives the alarm buzzer to sound an alarm, and at the same time pushes the diagnostic results to the mobile terminal through Wi-Fi.

[0008] Based on the above technical solutions, compared with the prior art, the advantages of the present invention are as follows: Through the fusion of multi-physical field information and the matching of the KNN-GRU hybrid model, the accuracy and timeliness of the early warning of lithium battery thermal runaway are significantly improved, providing a solution for battery safety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] Figure 1 It is a schematic diagram of the device structure of the present invention

[0010] Figure 2 It is a front schematic diagram of the circuit of the multimodal sensing module - embedded microcontroller module of the present invention

[0011] Figure 3 It is a back schematic diagram of the circuit of the multimodal sensing module - embedded microcontroller module of the present invention DETAILED DESCRIPTION OF THE INVENTION

[0012] The present invention will be further described below in conjunction with the drawings and embodiments:

[0013] As Figures 1 to 3 shown, the intelligent early warning system for lithium batteries based on multi-modal data fusion in this embodiment includes: a high-temperature resistant housing 1, which is provided with an air inlet 11 at the top and an air intake fan 12 is installed, a touch display screen 13 is embedded, and an exhaust port 14 is arranged on the side.

[0014] Inside the housing 1, a multi-modal sensing module 2 and an embedded microcontroller module 3 are integrated. The multi-modal sensing module 2 is connected to the embedded microcontroller module 3 through serial communication, and includes:

[0015] (1) A power supply circuit 21, which uses a 5V DC power supply input;

[0016] (2) A gas sensing array, which is composed of 14 semiconductor gas sensors 22 arranged in a matrix. The semiconductor sensitive layer of each sensor 22 uses different metal oxide materials, and the detection objects cover characteristic gases such as C1-C4 alkanes, hydrogen, and carbon monoxide. An internal heating control circuit 221 maintains the temperature of the sensitive layer at 200-400°C, and a supporting signal conditioning circuit 222 outputs an analog voltage of 0-5V;

[0017] (3) A thermocouple sensor 23, with a measuring range of -20°C to 200°C;

[0018] (4) A mechanical deformation detection unit, which includes two foil strain sensors, and is pasted on the battery surface through flexible wires passing through the air inlet 11;

[0019] (5) A sound sensor 24 and a temperature and humidity sensor 25.

[0020] The embedded microcontroller module 3 is configured with a Wi-Fi communication interface and serial communication capabilities. This module synchronously collects 20 analog-digital signals output by each sensor at a period of 100ms, and forms an eigenvalue vector to be uploaded to the cloud in real time through Wi-Fi.

[0021] The cloud decision-making platform runs an improved KNN-GRU hybrid model, which has stored multiple groups of multi-modal data during the thermal runaway process of lithium batteries during the training stage. When receiving the eigenvalue vector, it matches the historical sample with the highest similarity through the KNN algorithm, and outputs three types of evaluation results: the concentration of characteristic gases, the thermal runaway rate (unit: °C / min), and the safety status.

[0022] An alarm buzzer is installed inside the housing 1. When receiving a "warning" or "danger" instruction, it is driven by the embedded microcontroller module 3 to emit an alarm sound. At the same time, the evaluation results are synchronously displayed through the touch display screen 13 and the mobile application.

[0023] Example: Implementation method of an electric vehicle power battery pack

[0024] This embodiment is applied to the safety monitoring of an electric vehicle power battery pack. The outer shell 1 of the warning system is fixed to the inner side of the battery box, and the air inlet 11 faces the gap between the battery modules. Two foil strain sensors are pasted along the diagonal direction of the battery shell to detect the deformation in the X / Y axial directions respectively.

[0025] After the system is started, the embedded microcontroller module 3 performs the following operations at a cycle of 100 ms:

[0026] 1. Synchronously read the analog voltages of 14 gas sensors 22 through serial communication;

[0027] 2. Read the temperature value of the thermocouple sensor 23 to monitor the temperature change on the battery surface;

[0028] 3. Read the deformation voltage signals of the XY-axis strain sensors to detect the battery expansion state;

[0029] 4. Read the acoustic characteristics of the sound sensor 24 and the temperature and humidity values of the temperature and humidity sensor 25;

[0030] 5. Upload the 20-dimensional feature vector to the cloud decision-making platform through Wi-Fi.

[0031] The cloud improved KNN-GRU model matches the historical thermal runaway data in real time. When it is determined to be in the "warning" or "dangerous" state, it triggers the alarm buzzer to sound an alarm, the touch display screen 13 displays the thermal runaway rate, and an emergency warning is pushed through the mobile APP.

[0032] The above is only the specific implementation manner of this application, enabling those skilled in the art to understand or implement this application. Without departing from the principle of the present invention, several improvements and replacements can be made to the specific implementation details such as the selection of the semiconductor sensitive layer material. Various modifications to these embodiments are obvious to those skilled in the art. The general principles defined herein can be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, this application will not be limited to these embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features claimed in the text.

Claims

1. A lithium battery intelligent early warning system based on multimodal data fusion, characterized in that: It includes a housing, a multimodal sensor module, an embedded microcontroller module, an alarm buzzer and a cloud decision-making platform. The housing is made of high-temperature resistant materials, with an air inlet, an air inlet fan and a touch display screen on the top, an exhaust port on the side, and a display screen that can display the evaluation results. The multimodal sensing module and the embedded microcontroller module are integrated on a circuit; The multimodal sensing module includes a power supply circuit, a gas sensing array, a thermodynamic monitoring unit, a mechanical deformation detection unit, an acoustic feature collection unit and an environmental temperature and humidity sensing unit; The embedded microcontroller module is configured with a Wi-Fi communication interface and serial communication capabilities; The cloud decision platform deploys an intelligent diagnosis system based on an improved KNN-GRU hybrid model; The alarm buzzer sounds an alarm when the evaluation result is "warning" or "danger".

2. According to claim 1, a lithium battery intelligent early warning system based on multimodal data fusion is characterized in that: The gas sensor array comprises an array of 14 semiconductor gas sensors, and the detection objects of the 14 semiconductor gas sensors cover the characteristic gas group of thermal runaway of lithium batteries, including C1-C4 alkanes, hydrogen, carbon monoxide, ammonia, acetone, sulfur dioxide and electrolyte decomposition products; each semiconductor gas sensor outputs an analog voltage, a total of 14 analog voltages, which are transmitted to the embedded microcontroller module through serial communication; each gas sensor comprises: (1) a semiconductor sensitive layer, whose surface resistance value changes exponentially with the concentration of the target gas, and the materials used for the semiconductor sensitive layers of the 14 semiconductor gas sensors are different; (2) Heating control circuit to maintain the operating temperature of the sensitive layer in the range of 200-400°C; (3) Signal conditioning circuit, which converts the resistance change of the sensitive layer into a 0-5V analog digital signal.

3. According to claim 1, a lithium battery intelligent early warning system based on multimodal data fusion is characterized in that: The thermodynamic monitoring unit includes a thermocouple sensor with a measurement range of -20°C to 200°C, and the output analog voltage is transmitted to the embedded microcontroller module via serial communication. The mechanical deformation detection unit includes two strain sensors, which are connected to the multimodal sensing module through flexible wires, pass through the exhaust port of the shell, are distributed according to the XY axes of the plane rectangular coordinate system, and are attached to the surface of the battery to be tested; output two analog voltages and transmit them to the embedded microcontroller module through serial communication. The acoustic feature acquisition unit includes a sound sensor, and the output analog voltage is transmitted to the embedded microcontroller module via serial communication. The environmental temperature and humidity sensing unit comprises a temperature and humidity sensor with a temperature measurement range of 0°C to 50°C and a relative humidity measurement range of 20% to 90%. The output analog voltage is transmitted to the embedded microcontroller module via serial communication.

4. The lithium battery intelligent early warning system based on multimodal data fusion according to claim 1 is characterized in that: The embedded microcontroller module reads the analog digital signal of the multimodal sensing module every 100 ms. A group of analog digital signals read at the same time constitutes a 14+1+2+1+2=20-dimensional eigenvalue vector, which is sent to the cloud decision platform in real time via Wi-Fi communication.

5. The lithium battery intelligent early warning system based on multimodal data fusion according to claim 1 is characterized in that: The cloud-based decision-making platform deploys an intelligent diagnosis system based on an improved KNN-GRU hybrid model, which outputs the following evaluation results: characteristic gas type and concentration value, thermal runaway development rate prediction value (unit: °C / min), and safety status (normal / warning / dangerous); Wi-Fi communication sends the evaluation results to the embedded microcontroller module and the mobile application.

6. The lithium battery intelligent early warning system based on multimodal data fusion according to claim 5 is characterized in that: The improved KNN-GRU hybrid model has learned and stored a large amount of data in the simulation training stage; the eigenvalue vector is input to the model, and the model uses the KNN algorithm to match samples similar to the input eigenvalue vector from the trained data and outputs the corresponding evaluation results.

Images