Multi-modal technology fused regional fire extinguishing monitoring system

Abstract

The invention discloses a regional fire extinguishing monitoring system based on multi-modal technology fusion. The regional fire extinguishing monitoring system comprises a multi-sensor acquisition system, an AI video image recognition system and an upper computer centralized monitoring system, the multi-sensor acquisition system consists of a flame sensor, a smoke sensor, a temperature sensor, a CO sensor and an intrinsic safety type controller and is used for acquiring environmental parameters; the AI video image recognition system adopts a mining infrared intrinsic safety camera to collect video images, performs preprocessing and edge extraction on the images based on an OpenCV computer vision library, and performs flame recognition by using a YOLOv5s algorithm; and the upper computer centralized monitoring system is used for simultaneously receiving and analyzing the environmental parameters of the multi-sensor acquisition system and the flame recognition result of the AI video image recognition system, judging the fire state and issuing a fire extinguishing instruction when the fire is confirmed. According to the invention, the fire can be effectively identified, analyzed and warned, and the fire can be timely prevented from happening and further spreading.

Description

Technical Field

[0001] The present invention belongs to the technical field of regional fire extinguishing monitoring, and in particular relates to a regional fire extinguishing monitoring system integrating multimodal technologies. Background Art

[0002] Fire monitoring equipment for underground substations in coal mines is an important guarantee for ensuring safe production in coal mines. Underground substations in coal mines are complex and have the characteristics of being highly concealed and difficult to detect in the early stages of a fire. Once a fire accident occurs, the underground power supply will be completely paralyzed, generating a large amount of CO and toxic and harmful gases, which greatly threatens the lives of coal mine workers.

[0003] The controller of the automatic powder spray fire extinguishing device for mining areas is connected to the intrinsically safe pressure transmitter, intrinsically safe solenoid valve, sound and light alarm, and intrinsically safe flame sensor for mining. It can realize active and rapid powder spray fire extinguishing and explosion suppression, and is the main security equipment for fire prevention and extinguishing in underground coal mine substations. For a certain period of time, it achieved good fire extinguishing and explosion suppression effects. However, in recent years, with the gradual acceleration of the pace of intelligent upgrading and transformation of the coal mining industry, the automatic powder spray fire extinguishing device for mining areas cannot meet the development needs of intelligent, information-based, and unmanned coal mines. The main problems are as follows:

[0004] (1) Only smoke sensors and flame sensors are used for data collection, and the data collected is limited and cannot meet the current needs of coal mines;

[0005] (2) Sensor failures cannot be diagnosed autonomously, and inspection personnel and the centralized control center cannot be notified;

[0006] (3) The collection area is small and the detection method is single, which makes it easy to make false detections and missed detections.

[0007] (4) Inspection personnel need to be arranged regularly to check the regional fire extinguishing equipment. The equipment maintenance process is cumbersome and the dispatch room cannot monitor the on-site situation in real time.

[0008] Therefore, in order to solve the above problems, it is necessary to develop a regional fire extinguishing monitoring system that integrates multimodal technologies. Summary of the Invention

[0009] The present invention addresses these issues and overcomes the shortcomings of existing technologies by providing a regional fire extinguishing monitoring system that integrates multimodal technologies. This system utilizes AI video image recognition technology and multi-sensor technology to achieve intelligent interconnection and remote monitoring of underground coal mine fires.

[0010] To achieve the above objectives, the present invention adopts the following technical solutions.

[0011] The present invention provides a regional fire extinguishing monitoring system integrating multimodal technologies, comprising: a multi-sensor acquisition system, an AI video image recognition system, and a host computer centralized monitoring system;

[0012] The multi-sensor acquisition system consists of a flame sensor, a smoke sensor, a temperature sensor, a CO sensor and an intrinsically safe controller, and is deployed in the underground substation and the belt conveyor head of the coal mine to collect environmental parameters;

[0013] The AI ​​video image recognition system uses a mining infrared intrinsically safe camera to capture video images, preprocesses and extracts edges based on the OpenCV computer vision library, and uses the YOLOv5s algorithm for flame recognition.

[0014] The host computer centralized monitoring system is used to simultaneously receive and analyze the environmental parameters of the multi-sensor acquisition system and the flame recognition results of the AI ​​video image recognition system, determine the fire status, and issue a fire extinguishing instruction when a fire is confirmed.

[0015] As a preferred solution of the present invention, the image preprocessing of the AI ​​video image recognition system includes: bitmap conversion to YCrCb image, grayscale, binarization and image enhancement operations;

[0016] The recognition process of the AI ​​video image recognition system includes: labeling flame / non-flame samples using the labelimg tool, training the YOLOv5s model, and fusing multi-scale feature maps using a feature pyramid structure.

[0017] As another preferred solution of the present invention, the YOLOv5s algorithm has autonomous learning capability and improves the robustness of flame scene recognition by continuously increasing training samples.

[0018] As another preferred embodiment of the present invention, the multi-sensor acquisition system further includes:

[0019] Signal isolation module, used for optical coupling isolation of analog signals, digital signals and switching signals;

[0020] Multi-communication interface module, used to support RS485, network port and optical port communication methods.

[0021] As another preferred solution of the present invention, the intrinsically safe controller adopts an STM32 chip, and is respectively connected to the flame sensor, smoke sensor, temperature sensor and CO sensor through the signal isolation module.

[0022] As another preferred solution of the present invention, the signal isolation module includes three groups of independent circuits: an analog isolation circuit, a digital isolation circuit and a switch isolation circuit, all of which adopt an optocoupler isolation design.

[0023] As another preferred embodiment of the present invention, the host computer centralized monitoring system includes:

[0024] Sensor parameter monitoring unit for real-time display of flame, smoke, temperature and CO data;

[0025] AI video image monitoring unit, used to display infrared video streams and flame recognition results in real time.

[0026] As another preferred solution of the present invention, when the host computer centralized monitoring system confirms a fire, it synchronously triggers the sound and light alarm and sends a start instruction of the mining intrinsically safe solenoid valve to the intrinsically safe controller.

[0027] As another preferred embodiment of the present invention, the intrinsically safe controller is linked with a mining intrinsically safe solenoid valve, an audible and visual alarm, and a pressure transmitter to perform fire extinguishing operations.

[0028] Compared with the prior art, the present invention has the following beneficial effects:

[0029] By analyzing the advantages and disadvantages of traditional sensor fire extinguishing and infrared video image recognition, this paper designs an efficient and intelligent regional fire extinguishing monitoring system that integrates multimodal technologies. The present invention combines multi-sensor technology with AI video image recognition technology to effectively identify, analyze and warn of fires, and promptly prevent the occurrence and further spread of fires. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 This is a schematic structural diagram of a regional fire extinguishing monitoring system integrating multi-modal technologies according to the present invention;

[0031] Figure 2 This is a communication architecture diagram for communicating between the multi-sensor acquisition system of the present invention and the host computer centralized monitoring system through multiple communication interface modules;

[0032] Figure 3 A digital signal isolation circuit diagram of the signal isolation module of the present invention;

[0033] Figure 4 This is a circuit diagram of analog signal isolation of the signal isolation module of the present invention;

[0034] Figure 5 This is a circuit diagram of a switching signal isolation module of the present invention;

[0035] Figure 6 A flowchart for building an AI video image recognition model. DETAILED DESCRIPTION

[0036] In order to make the technical problems, technical solutions and beneficial effects solved by the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0037] Combine Figures 1 to 6 As shown, an embodiment of the present invention provides a regional fire extinguishing monitoring system with multimodal technology integration, including: a multi-sensor acquisition system, an AI video image recognition system and a host computer centralized monitoring system; the multi-sensor acquisition system is composed of a flame sensor, a smoke sensor, a temperature sensor, a CO sensor and an intrinsically safe controller, and is arranged in the underground substation and belt conveyor head of the coal mine to collect environmental parameters; the AI ​​video image recognition system uses a mining infrared intrinsically safe camera to collect video images, preprocesses and extracts edges of the images based on the OpenCV computer vision library, and uses the YOLOv5s algorithm for flame recognition; the host computer centralized monitoring system is used to simultaneously receive and analyze the environmental parameters of the multi-sensor acquisition system and the flame recognition results of the AI ​​video image recognition system, judge the fire status, and issue a fire extinguishing command when the fire is confirmed.

[0038] Specifically, the image preprocessing of the AI ​​video image recognition system includes: bitmap conversion to YCrCb image, grayscale conversion, binarization and image enhancement operations; the recognition process of the AI ​​video image recognition system includes: labeling flame / no-flame samples through the labelimg tool, training the YOLOv5s model, and fusing multi-scale feature maps through the feature pyramid structure; the YOLOv5s algorithm has autonomous learning capabilities and improves the robustness of flame scene recognition by continuously increasing training samples.

[0039] Specifically, the multi-sensor acquisition system further includes: a signal isolation module for performing optical coupling isolation on analog signals, digital signals and switch signals; and a multi-communication interface module for supporting RS485, network port and optical port communication modes.

[0040] Specifically, the intrinsically safe controller adopts an STM32 chip and is respectively connected to a flame sensor, a smoke sensor, a temperature sensor and a CO sensor through the signal isolation module.

[0041] Specifically, the signal isolation module includes three groups of independent circuits: an analog isolation circuit, a digital isolation circuit, and a switch isolation circuit, all of which adopt an optocoupler isolation design.

[0042] Specifically, the host computer centralized monitoring system includes: a sensor parameter monitoring unit for real-time display of flame, smoke, temperature and CO data; an AI video image monitoring unit for real-time display of infrared video streams and flame recognition results.

[0043] Specifically, when the host computer centralized monitoring system confirms a fire, it synchronously triggers the sound and light alarm and sends a start instruction of the mining intrinsically safe solenoid valve to the intrinsically safe controller.

[0044] Specifically, the intrinsically safe controller is linked with the intrinsically safe solenoid valve, the sound and light alarm and the pressure transmitter for mining to perform fire extinguishing operations.

[0045] The multi-sensor acquisition system of the present invention uses the STM32 chip as the control core and collects and monitors the on-site environment through flame sensors, smoke sensors, temperature sensors, and CO sensors; the new technologies used in multi-sensor acquisition are as follows:

[0046] (1) Traditional regional fire extinguishing device sensors are mostly smoke and flame sensors. When smoldering and fire occur in coal mines, the collected data is limited and cannot reflect the on-site environmental conditions in real time. The design of the present invention uses four parameters of flame sensor, smoke sensor, temperature sensor, and CO sensor as on-site environmental data collection, making the on-site environmental data collection more comprehensive.

[0047] (2) Multiple communication modes are available. The system's communication modes include RS485 communication, optical port communication, and network port communication, which facilitates communication and transmission of on-site equipment.

[0048] (3) It has analog, digital and switch signal isolation. The multi-sensor acquisition system isolates the front-end and back-end power supply of the signal, and the signal part is isolated by optical coupler.

[0049] The AI ​​video image recognition system of the present invention collects video images by a mine-used infrared intrinsically safe camera to provide video data for fire image recognition. The image recognition algorithm is constructed based on the OpenCV computer vision library and Yolov5s. First, the video image is collected by the mine-used infrared intrinsically safe camera, and the collected image is preprocessed by using the OpenCV database to retrieve the collected image, including bitmap conversion to YCrCb image, grayscale, binarization, image enhancement operations, and then edge extraction of the image is performed. Finally, the image is recognized using the Yolov5s algorithm. Image recognition process: First, the flame and non-flame images are classified and labeled by the labelimg tool, and then the image is loaded for model training. The Yolov5s algorithm effectively fuses feature maps of different scales by fusing feature information layer by layer from the bottom of the pyramid upward or the top layer downward. Moreover, as the number of samples increases, the Yolov5s algorithm will encounter many different flame scenes during the learning process, which will further improve the Yolov5s algorithm's ability to recognize flame images and gradually reach a stable state, thereby improving the robustness and accuracy of flame recognition.

[0050] The host computer centralized monitoring system of the present invention includes a sensor parameter monitoring unit for displaying sensor data and an AI video image monitoring unit for displaying video images. If a fire is detected, the AI ​​video image will first issue an early warning. Simultaneously, the host computer centralized monitoring system will send a fire extinguishing command to the intrinsically safe controller. The intrinsically safe controller will then control the intrinsically safe solenoid valve used in mining to extinguish the fire at the fire site, ensuring on-site safety.

[0051] In summary, by analyzing the advantages and disadvantages of traditional sensor fire extinguishing and infrared video image recognition, the present invention provides an efficient and intelligent regional fire extinguishing monitoring system that integrates multimodal technologies. The present invention combines multi-sensor technology with AI video image recognition technology to effectively identify, analyze and warn of fires, and promptly prevent the occurrence and further spread of fires.

[0052] It can be understood that the above specific description of the present invention is only used to illustrate the present invention and is not limited to the technical solutions described in the embodiments of the present invention. Those skilled in the art should understand that the present invention can still be modified or replaced by equivalents to achieve the same technical effects; as long as the use requirements are met, they are within the scope of protection of the present invention.

Claims

1. A regional fire extinguishing monitoring system integrating multimodal technologies, characterized by: include: Multi-sensor acquisition system, AI video image recognition system and host computer centralized monitoring system; The multi-sensor acquisition system consists of a flame sensor, a smoke sensor, a temperature sensor, a CO sensor and an intrinsically safe controller, and is deployed in the underground substation and the belt conveyor head of the coal mine to collect environmental parameters; The AI ​​video image recognition system uses a mining infrared intrinsically safe camera to capture video images, preprocesses and extracts edges based on the OpenCV computer vision library, and uses the YOLOv5s algorithm for flame recognition. The host computer centralized monitoring system is used to simultaneously receive and analyze the environmental parameters of the multi-sensor acquisition system and the flame recognition results of the AI ​​video image recognition system, determine the fire status, and issue a fire extinguishing instruction when a fire is confirmed.

2. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 1 is characterized by: The image preprocessing of the AI ​​video image recognition system includes: bitmap conversion to YCrCb image, grayscale, binarization and image enhancement operations; The recognition process of the AI ​​video image recognition system includes: labeling flame / non-flame samples using the labelimg tool, training the YOLOv5s model, and fusing multi-scale feature maps using a feature pyramid structure.

3. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 1 is characterized by: The YOLOv5s algorithm has autonomous learning capabilities and improves the robustness of flame scene recognition by continuously adding training samples.

4. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 1 is characterized by: The multi-sensor acquisition system further includes: Signal isolation module, used for optical coupling isolation of analog signals, digital signals and switching signals; Multi-communication interface module, used to support RS485, network port and optical port communication methods.

5. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 4 is characterized by: The intrinsically safe controller adopts an STM32 chip and is respectively connected to a flame sensor, a smoke sensor, a temperature sensor and a CO sensor through the signal isolation module.

6. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 4 is characterized by: The signal isolation module includes three groups of independent circuits: analog isolation circuit, digital isolation circuit and switch isolation circuit, all of which adopt optocoupler isolation design.

7. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 1 is characterized by: The host computer centralized monitoring system includes: Sensor parameter monitoring unit for real-time display of flame, smoke, temperature and CO data; AI video image monitoring unit, used to display infrared video streams and flame recognition results in real time.

8. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 1 is characterized by: When the host computer centralized monitoring system confirms a fire, it will synchronously trigger the sound and light alarm and send a start instruction of the mining intrinsically safe electromagnetic valve to the intrinsically safe controller.

9. The regional fire extinguishing monitoring system integrating multimodal technologies according to claim 1, characterized in that: The intrinsically safe controller is linked with the intrinsically safe solenoid valve, the sound and light alarm and the pressure transmitter for use in mining to perform fire extinguishing operations.

Images