White spirit village fire-fighting management system based on machine vision and composite sensor

Abstract

The invention relates to the technical field of a wine village fire-fighting system, and discloses a white spirit village fire-fighting management system based on machine vision and a composite sensor, and the system comprises a server, a wine village fire-fighting control center, an intelligent camera, a composite sensor, a linkage device, a visual evacuation guide module, a fire-fighting unmanned aerial vehicle, a mobile terminal and a computer terminal. The intelligent camera monitors fire signals such as flame and smoke in real time by using a YOLO target detection algorithm, and the composite sensor collects environmental data such as temperature, humidity and alcohol vapor concentration. According to the system, the fire risk is warned in advance through intelligent analysis, the fire resisting shutter, the fire extinguishing device and other equipment are triggered according to linkage logic, and intelligentization and quick response are achieved. In addition, the fire-fighting unmanned aerial vehicle is used for pre-disaster inspection and real-time fire feedback, and the visual evacuation guiding module guides personnel to evacuate in modes of LED floor tile display, wall projection and the like. And the mobile terminal and the computer terminal realize remote monitoring and operation, so that the fire early warning and emergency management capabilities are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of winery fire protection systems, and in particular to a white winery fire protection management system based on machine vision and composite sensors. Background Art

[0002] With the rapid development of the liquor industry, many large liquor wineries have gradually modernized, expanded their production scale, and upgraded their equipment. However, due to the highly flammable nature of liquor itself, the fire safety issues of wineries have become increasingly prominent. Traditional fire management systems mostly rely on simple temperature and smoke sensors, which cannot cope with the complex and diverse winery environment, especially in the case of high alcohol volatility, complex building structures and dense personnel. Traditional fire protection technology has many limitations.

[0003] 1. Limitations of traditional fire protection systems

[0004] Winery fire management systems have long relied on simple sensors such as temperature sensors, smoke sensors, and manual alarm devices. These systems have several significant problems:

[0005] Fire detection lag: Smoke detectors and temperature detectors usually need to wait until the fire has spread to a certain extent before they can detect the occurrence of a fire, at which point the fire may be difficult to control. This passive detection method leads to delayed fire response and increases the risk of fire losses.

[0006] Limited sensor monitoring range: Existing smoke and temperature sensors are poorly adaptable to the special environment of wineries. For example, in areas such as wine cellars and fermentation tanks, there are high humidity and high alcohol concentration environmental conditions, which will interfere with the accuracy of conventional sensors and cause false alarms or missed alarms. In addition, since winery buildings usually contain multi-story underground wine cellars, wooden building structures and a large number of scattered functional areas, it is difficult for existing sensors to fully cover all high-risk areas.

[0007] Insufficient linkage response mechanism: Traditional fire protection systems are mostly passive responses, that is, fire extinguishing equipment or alarm systems are activated only after a fire occurs, lacking intelligent, real-time early warning and active fire prevention mechanisms. In addition, in traditional fire protection systems, various components such as sensors, alarms, and fire extinguishing devices are independent of each other, lack linkage control, and cannot achieve coordinated response of multiple devices.

[0008] 2. The particularity of the winery environment

[0009] Fire management in wineries presents unique challenges due to the following aspects:

[0010] Highly flammable environment: The winery stores a large amount of highly flammable liquor, which is extremely prone to fire. Especially during the production process, the volatilization of alcohol vapor also increases the risk of fire.

[0011] Complex building structures: Many wineries have underground cellars, wooden buildings, etc., and the functional areas are scattered. These building environments are not only difficult to deploy firefighting equipment, but may also increase the speed of fire spread when a fire occurs, making it difficult to control the situation.

[0012] High humidity environment: In order to maintain the quality of liquor, a high humidity needs to be maintained in the wine cellar. This environment poses a great challenge to the reliability of conventional fire-fighting equipment and may cause some electronic equipment to malfunction or reduce sensitivity.

[0013] High density of people: Liquor wineries are usually not only production bases, but also tourism facilities, especially during the peak tourist season when there are a large number of tourists. Under such conditions, when a fire occurs, it is necessary not only to ensure the rapid response of equipment, but also to effectively evacuate people to avoid stampedes or congestion.

[0014] Chemical management risks: Wineries use a variety of chemicals such as disinfectants and solvents in the production process. Once these chemicals react with alcohol, they may cause more serious fires or explosions. Therefore, the fire protection system needs to monitor not only fires, but also the concentration of toxic gases and volatile chemicals in the air.

[0015] 3. The existing winery fire management system generally has the following deficiencies:

[0016] Single monitoring method: Most current systems rely on temperature and smoke sensors, which can only detect a single type of fire risk and cannot identify other important parameters such as alcohol vapor concentration and air humidity, resulting in incomplete and untimely fire warnings.

[0017] Lack of intelligence: Most existing systems lack intelligent analysis capabilities and are unable to make comprehensive judgments on multi-source information. Especially in complex winery environments, the systems often cannot distinguish between real fire signals and other harmless interference information (such as water vapor, dust, etc.).

[0018] Lack of linkage mechanism: Most existing systems are passive responses, triggering equipment only when the fire has spread. Modern wineries need to have an intelligent linkage response mechanism that can take appropriate preventive measures at the early stage of a fire, such as automatically opening the ventilation system, shutting off the power supply, or activating the fire extinguishing device.

[0019] Innovative solutions based on machine vision and composite sensors To solve the above problems, it is urgent to develop a comprehensive fire management system that can adapt to the special environment of the winery and realize comprehensive monitoring and intelligent analysis. The present invention proposes an intelligent fire management system for liquor winery based on machine vision and composite sensors. Summary of the invention

[0020] The purpose of the present invention is to address the deficiencies of the prior art and provide a white wine winery fire management system based on machine vision and composite sensors to solve the problems raised in the background technology.

[0021] To achieve the above object, the present invention provides the following technical solutions:

[0022] The solution of the present invention provides a white wine winery fire management system based on machine vision and composite sensors, the system includes a server, a winery fire control center, an intelligent camera, a composite sensor, a linkage device, a visual evacuation guidance module, a firefighting drone, a mobile terminal and a computer terminal.

[0023] Smart cameras use machine vision technology to collect, analyze and detect fire targets in real time, and make early warning and linkage control decisions based on data from composite sensors.

[0024] Composite sensors include flame detection sensors, temperature and humidity sensors, and electronic noses to collect environmental data;

[0025] The linkage equipment is controlled according to the sensor data, including fire shutters, fire extinguishers, sound and light alarms and other equipment. The linkage equipment is connected with smart cameras and the fire control center to trigger emergency operations according to the fire situation;

[0026] The visual evacuation guidance module includes an LED floor tile display, a wall projection system and a sound prompt system, which is used to provide real-time updates of the best evacuation route for people in the winery when a fire occurs, guiding people to quickly evacuate the dangerous area;

[0027] Firefighting drones are equipped with high-definition cameras and infrared thermal imagers, which are used for pre-disaster inspections and real-time information feedback during disasters. They fly to the fire area and transmit fire images and data to the fire control center;

[0028] The server is used to store and manage the system's operating data, including real-time alarm information, sensor data, target detection results, and equipment linkage status, and is connected to mobile terminals and computer terminals through wireless communication for users to view and operate;

[0029] The winery fire control center is responsible for monitoring the operation of the entire fire protection system, processing data from smart cameras, composite sensors and firefighting drones, and activating related linkage equipment according to preset linkage rules, while generating the best evacuation routes for different areas;

[0030] The mobile terminal is connected to the server, and the user can view and control the operating status of each module of the system in real time through the mobile terminal, and query the alarm history, sensor data, and linkage device status;

[0031] The computer terminal is used for system configuration and model training. Users upload and update target detection models and linkage logic scripts through the computer terminal, and configure relevant system parameters.

[0032] As a preferred technical solution of the present invention, the smart camera has an embedded image recognition algorithm and loads different target detection models according to different application scenarios to identify flame, smoke and fire targets.

[0033] As a preferred technical solution of the present invention, the image recognition algorithm supports multiple target detection models, including but not limited to the YOLO series models. Users can train the detection model of a specific fire target through a computer and convert it into a model format suitable for the smart camera embedded system.

[0034] As a preferred technical solution of the present invention, the YOLO target detection algorithm supports users to upload pre-trained models according to specific application scenarios, and load the models into the camera device through an Ethernet or WiFi interface.

[0035] As a preferred technical solution of the present invention, the smart camera performs inference calculations on the collected image data in real time through its internal computing unit, and generates a control signal in combination with the data collected by the composite sensor, and sends it to the server or linkage device.

[0036] As a preferred technical solution of the present invention, the composite sensor communicates with the smart camera and the fire control center through a bus-type communication method, and the communication method includes but is not limited to RS485, 232, Ethernet, TCP, LoRa and Zigbee.

[0037] As a preferred technical solution of the present invention, the firefighting drone is equipped with a high-definition camera and an infrared thermal imager, has an automatic obstacle avoidance function, automatically cruises to a designated area when a disaster occurs, and transmits fire images and thermal imaging data to a fire control center in real time.

[0038] As a preferred technical solution of the present invention, the system realizes real-time monitoring of alcohol vapor through an electronic nose. When the alcohol concentration exceeds a safety threshold, the system automatically alarms and links related equipment to start an emergency ventilation device.

[0039] As a preferred technical solution of the present invention, the smart camera works in coordination with a composite sensor. When the smart camera detects a fire risk, it reads relevant environmental data from the composite sensor through a communication unit, and comprehensively judges the fire risk by combining multi-dimensional data of temperature, humidity and gas concentration.

[0040] Compared with the prior art, the present invention has the following beneficial effects:

[0041] 1. The present invention combines the advantages of machine vision and composite sensors to achieve comprehensive monitoring of various fire risks in liquor wineries. The intelligent camera cooperates with the YOLO target detection algorithm to accurately detect fire signals such as flames and smoke. At the same time, the composite sensor collects key parameters such as environmental temperature and humidity, alcohol vapor concentration, etc. in real time to form a comprehensive monitoring network. This multi-dimensional monitoring method effectively improves the accuracy of fire warnings, and identifies potential risks in advance through intelligent analysis, significantly shortens the fire response time, and improves the system's prevention and emergency response capabilities.

[0042] 2. Through the built-in intelligent linkage control module of the system, the present invention can automatically trigger fireproof roller shutters, fire extinguishing devices, ventilation systems and other equipment according to the actual situation of the fire, and realize active prevention and control in the early warning stage. In addition, the system can comprehensively analyze the data of cameras and sensors to avoid false alarms, effectively link multiple devices to work together, and ensure that fire prevention and fire extinguishing measures are quickly in place in the early stage of the fire. This intelligent linkage mechanism greatly improves the emergency response efficiency of the system and reduces the property losses and casualties that may be caused by fire.

[0043] 3. The present invention introduces a visual evacuation guidance module, which dynamically guides the personnel in the winery to evacuate along the best evacuation route through wall projection, LED floor tile display screen and voice prompt system. In addition, the combination of mobile terminal and computer terminal enables users to monitor the system operation status anytime and anywhere, and perform remote operation and management as needed. This function not only ensures the rapid and safe evacuation of personnel, but also provides fire management personnel with efficient post-disaster emergency management means, and improves the overall fire safety management level. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 This is a system block diagram of the fire protection management system for a liquor winery based on machine vision and composite sensors proposed by the present invention;

[0045] Figure 2 This is a system diagram of the intelligent camera of the fire management system of the liquor winery based on machine vision and composite sensors proposed by the present invention;

[0046] Figure 3This is a workflow diagram of the white wine winery fire management system based on machine vision and composite sensors proposed in the present invention. DETAILED DESCRIPTION

[0047] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0048] The following is combined with Figure 1-Figure 3 With reference to the accompanying drawings and a number of embodiments, specific embodiments of the present invention are described in detail.

[0049] The embodiment of the present invention relates to a fire management system for a liquor winery based on machine vision and composite sensors, which is mainly composed of a server, a winery fire control center, an intelligent camera, a composite sensor, a linkage device, a visual evacuation guidance module, a firefighting drone, a mobile terminal and a computer terminal. The system of the present invention realizes fire detection, monitoring, emergency handling and personnel evacuation in the winery through the coordinated work of machine vision, composite sensors and linkage devices. Specific implementation process:

[0050] Step 1:

[0051] Model training (optional): Select the target object to be detected, prepare the labeled data set, import the data set into the computer deployed with the machine vision training framework, complete the corresponding parameter configuration according to the YOLO model training requirements (it belongs to the professional basic technology, and the specific configuration method is configured according to the YOLO parameter requirements), and then export the trained data, and then convert the exported data into the RKNN model (applicable to the model structure of Rockchip chips. If chip modules from other manufacturers are used, convert them into the corresponding model) for easy deployment in embedded systems. In addition, you can also directly use the model files and linkage logic scripts provided by manufacturers or third parties.

[0052] Step 2:

[0053] Deploy server: Deploy basic services on the server to support the winery fire management system, including data communication services, data storage services, equipment model configuration parameter management services (including equipment models and linkage logic scripts) and configuration update services.

[0054] Among them, the data communication service is mainly used to realize data communication between mobile terminals, computer terminals and smart camera devices; the data storage service is mainly used to save the data generated in the data communication service and the data in the device model configuration parameter management service; the device model configuration parameter management service is mainly used to manage the device model configuration parameters of different detection tasks; the configuration update service is mainly used to complete the configuration parameters (including device models and linkage logic scripts) uploaded by the computer terminal and store them in the device model configuration parameter management service, and at the same time, the configuration parameters in the smart camera device are updated through the configuration update service.

[0055] Step 3:

[0056] Write linkage logic scripts: According to application requirements, set the number and basic information of smart camera devices, sensor types and numbers, and complete the basic parameter configuration of each sensor, linkage device type and number, and complete the basic parameter configuration of each linkage device in accordance with the linkage logic script format;

[0057] Step 4:

[0058] System parameter configuration (configure system parameters when using the smart camera device for the first time or when you need to modify the device parameters): Connect the smart camera device to the computer via an Ethernet cable, and complete the configuration of the network connection parameters of the smart camera device on the computer, including server connection parameters (domain name / IP address, port number, user name and user name password, data reporting cycle time, if using WiFi to connect to the Internet, you need to configure the corresponding wireless network information, whether to enable the original video and target detection results to be saved in the storage unit, sensor data timing reporting cycle, etc.), and at the same time transfer the trained model and linkage logic script to the smart camera device.

[0059] Step 5:

[0060] Install each device: install the smart camera device, sensor, and linkage device to the corresponding location, and connect the sensor and linkage device to the smart camera device through the bus, and connect the smart camera device to the corresponding server via Ethernet / WiFi.

[0061] Step 6:

[0062] The equipment starts working according to the established parameter configuration, and its working process is as follows:

[0063] ① Initialize the system and complete the initialization of basic peripherals, including display screen, communication bus, serial port, storage unit, camera unit, buzzer, etc.;

[0064] ② Determine whether the network connection is normal. If the network connection fails, the buzzer alarm will be triggered and the network will be initialized again;

[0065] ③After the network is initialized successfully, the latest configuration is downloaded from the server. If the latest configuration exists, the latest configuration is downloaded to the storage unit;

[0066] ④ Load the target detection model and linkage logic script from the storage unit;

[0067] ⑤ Read image data from the image acquisition unit, transmit the image data to the computing unit, complete target detection, and return the detection results to the central processing unit;

[0068] ⑥ The central processing unit reads the sensor data in sequence, and determines whether the data is abnormal, whether there is a detection target, and whether it satisfies the control logic configured in the linkage device according to the linkage logic script. If it does, there is a data abnormality. When the data is abnormal, the device bus of the communication unit controls the linkage unit action, and reports the target detection results, device linkage status and sensor data to the server through the Ethernet / WiFi interface; otherwise, it determines whether the data reporting time has arrived. If it does, the sensor data is reported to the server through the Ethernet / WiFi interface of the communication unit;

[0069] ⑦ Update the display data, including real-time target detection images, sensor data, linkage device status, network connection status, time and other information.

[0070] ⑧ Determine whether data recording is enabled. If enabled, record the original data and target detection result data, and then return to step ①.

[0071] Step 7:

[0072] System detection and data viewing through mobile terminals (optional): In the mobile terminal, select different devices by device name and device number to view video data and sensor data in real time. At the same time, you can check different devices to generate statistical data according to time and alarm type.

[0073] Step 8:

[0074] Update detection target and / or linkage logic script:

[0075] Reselect the target object to be detected, prepare the labeled data set, import the data set into the computer where the machine vision training framework is deployed, and complete the corresponding parameter configuration according to the yolo model training requirements (it belongs to professional basic technology, and the specific configuration method is configured according to the yolo parameter requirements), then export the trained model, and then convert the exported data into a rknn model (applicable to the model structure of Rockchip chips. If chip modules from other manufacturers are used, convert them into the corresponding model) for easy deployment in embedded systems. In addition, you can also directly use the model files and linkage logic scripts provided by the manufacturer or a third party. Then upload the corresponding model files and linkage logic scripts to the server, and select the corresponding device to complete the update of the detection target / or linkage logic script.

[0076] Example 1: Basic deployment and operation of the system

[0077] In this embodiment, the system forms a complete fire management framework through servers, smart cameras, composite sensors, linkage equipment, etc., which is responsible for real-time monitoring of the fire situation of the entire winery.

[0078] 1. System startup and initialization

[0079] When the system starts, it will initialize all smart cameras, composite sensors and linkage devices. The server will connect smart cameras, composite sensors and mobile terminals through the network to ensure that the configuration data and parameters of all devices are updated synchronously.

[0080] Server: used to store real-time data from sensors and cameras, including temperature and humidity, smoke concentration, flame detection data, etc. The server is also responsible for saving system operation logs, alarm records, and device status.

[0081] Smart cameras: Installed in different areas of the winery, such as production workshops, wine cellars, fermentation tanks, etc., they collect image data in real time and have a built-in YOLO target detection algorithm to detect fire targets such as flames and smoke.

[0082] Composite sensors: including temperature and humidity sensors, electronic noses, flame sensors, etc., responsible for collecting data such as environmental temperature, humidity, and alcohol vapor concentration. The sensors communicate with the fire control center or smart camera through bus communication methods (such as RS485, LoRa, Zigbee).

[0083] Interconnected equipment: such as fire shutters, fire extinguishing devices, and sound and light alarms, are responsible for executing corresponding fire prevention measures based on sensor data or camera recognition results.

[0084] 2. Data collection and fire detection

[0085] The system continuously collects environmental data through smart cameras and composite sensors. The smart cameras execute the YOLO image recognition algorithm to detect flames or smoke.

[0086] Camera acquisition and processing: The image data collected by the camera in real time is processed by the camera's computing unit and loaded into the target detection model. When a fire target (such as flames, smoke) is detected, the system transmits the data to the fire control center.

[0087] Sensor collection and judgment: The composite sensor monitors temperature and humidity, alcohol concentration, and harmful gases in the air in real time. The electronic nose is used to detect alcohol vapor in the air. When the alcohol concentration exceeds the predetermined threshold, an alarm will be immediately sent to the fire control center.

[0088] 3. Intelligent linkage and emergency response

[0089] When the system detects flames or smoke through the camera, or the sensor detects abnormal environmental conditions (such as high temperature, excessive alcohol concentration), the system automatically links the equipment to initiate corresponding fire prevention measures.

[0090] Fireproof roller shutter: When the flame sensor and smart camera detect a fire signal, the fireproof roller shutter will automatically fall to isolate the fire source and prevent the fire from spreading.

[0091] Automatic fire extinguisher: When the sensor detects that the temperature exceeds the predetermined threshold, the fire extinguisher is automatically triggered to spray the fire extinguishing agent.

[0092] Sound and light alarm: The system triggers the sound and light alarm to remind people on site to evacuate as soon as possible, and sends a fire alarm to the control center and mobile terminals.

[0093] 4. Visual evacuation guidance

[0094] The visual evacuation guidance module within the system includes an LED floor tile display, a wall projection system and a sound prompt system to help personnel quickly find a safe exit in the event of a fire.

[0095] LED floor tile display: The floor tile screen in the evacuation channel displays the evacuation route in real time, helping people quickly find the escape direction.

[0096] Wall projection: The system projects evacuation path signs on the walls in corridors and rooms, dynamically updating the evacuation routes.

[0097] Voice guidance: The voice prompt system of the evacuation channel will play real-time voice prompts according to the evacuation route to inform personnel of the best evacuation route.

[0098] Example 2: Firefighting drone inspection and disaster feedback

[0099] The main functions of the firefighting drone are pre-disaster inspection and real-time information feedback when a fire occurs. This embodiment describes in detail the drone's automatic cruise, fire monitoring, and post-disaster feedback functions.

[0100] 1. Automatic cruise control

[0101] After the system is started, the drone will cruise according to the preset schedule. During unmanned inspections, the drone is equipped with a high-definition camera and infrared thermal imager to collect real-time images and temperature data of the winery environment.

[0102] Cruise path setting: Users can pre-set the drone’s cruise path through a computer terminal or mobile terminal. The path covers the entire winery, including difficult-to-access underground cellars, the top of fermentation tanks and other areas.

[0103] Data collection: The drone uses high-definition cameras and infrared thermal imagers to collect images of areas with abnormal temperatures and analyze whether there is a fire risk. When abnormal temperatures or other potential fire hazards are found, the data is transmitted to the fire control center.

[0104] 2. Disaster monitoring and feedback

[0105] When a fire occurs, the drone will fly to the fire area and transmit images and fire data to the fire control center in real time.

[0106] Real-time data transmission: The drone transmits high-definition images and infrared thermal imaging data collected by the camera to the control center in real time. Operators can view the scope and development of the fire through the display screen of the control center.

[0107] Automatic obstacle avoidance: The drone has an automatic obstacle avoidance function, which can avoid collision with the complex building structures or equipment of the winery, ensuring smooth arrival at the disaster area.

[0108] Post-disaster data storage and analysis: After the drone returns, the system stores all the data it has collected in the server for post-analysis and disaster review.

[0109] Example 3: Electronic nose sensor and gas monitoring linkage

[0110] This embodiment focuses on describing the linkage operation of the electronic nose sensor when detecting alcohol vapor and other volatile gases.

[0111] 1. Data collection of electronic nose sensors

[0112] Winery fermentation tanks and cellars usually release a lot of alcohol vapor, which can easily cause fires in high concentrations. For this reason, electronic nose sensors are deployed in these high-risk areas to monitor volatile gases in the air.

[0113] Alcohol vapor monitoring: The electronic nose can quickly detect the concentration of alcohol vapor in the air. When the concentration is detected to exceed the safety threshold, the system will trigger an alarm and automatically start the ventilation system.

[0114] Early fire warning: Compared with traditional smoke detectors, electronic noses are more sensitive to gas components and can detect trace amounts of combustion gases in the early stages of a fire, helping to sound an alarm in advance.

[0115] 2. Automatic ventilation system linkage

[0116] When the electronic nose detects that the concentration of alcohol vapor is too high, the fire control center will trigger the automatic ventilation system to quickly expel the high concentration of alcohol vapor.

[0117] Ventilation system start-up: The control center sends instructions through bus-type communication to automatically start the ventilation system to ensure that high-concentration steam is quickly discharged from the wine cellar or fermentation area.

[0118] False alarm prevention: The electronic nose sensor works together with the temperature and humidity sensor and the flame detection sensor to prevent false alarms caused by environmental errors or other harmless gases.

[0119] Example 4: Remote monitoring and management of the system based on mobile terminals

[0120] The mobile terminal provides remote monitoring and management functions for the entire fire protection system. Users can view the system operation status in real time through mobile devices and perform remote operations in emergency situations.

[0121] 1. Real-time monitoring and alarm

[0122] Users can view camera video streams, sensor data, and system operating status in real time through mobile terminals. When the system detects a fire, the mobile terminal will receive an instant alarm notification.

[0123] Live video streaming: Users can choose to view camera videos from different areas and understand the fire scene in real time.

[0124] Alarm push: When a fire or other safety hazard occurs, the system will immediately push alarm information to the user, and the user can remotely direct response measures based on the alarm details.

[0125] 2. Remote control and operation

[0126] Through mobile terminals, users can remotely control the start and stop of linked equipment.

[0127] Fire extinguisher activation: When the system detects a fire, users can remotely activate the fire extinguisher or close the fire shutter through a mobile terminal to reduce the spread of fire.

[0128] System parameter adjustment: Users can also adjust sensor threshold settings, linkage logic and other configurations in real time through mobile terminals to deal with special situations.

[0129] Example 5: System configuration and model training based on computer terminals

[0130] This example demonstrates how users can deeply configure the system through a computer terminal and train and deploy a target detection model.

[0131] 1. Model training and deployment

[0132] In specific scenarios, such as flame detection or fire target identification, users can train new target detection models through computer terminals and deploy them to smart cameras.

[0133] Model training process: Users prepare a labeled dataset of fire targets through the YOLO algorithm training framework, adjust model parameters and perform training. The trained model will be converted into a format suitable for camera embedded systems (such as RKNN model).

[0134] Model upload and update: The trained model is uploaded to the server through the computer terminal. The smart camera downloads the model from the server through the network connection and replaces the old model. The updated camera will automatically start to execute the new task.

[0135] 2. System parameter configuration

[0136] The computer terminal allows the user to configure the system's network connection, sensor communication method, linkage logic and alarm thresholds.

[0137] Network configuration: Users can set the IP address, port number, connection username and password for server connection to ensure that the device runs normally on the network.

[0138] Linkage logic script: Users can write linkage logic scripts to set the response conditions of sensor data and linkage devices, such as a flame detection sensor and a temperature sensor jointly triggering a fire extinguisher.

[0139] The contents not described in detail in this specification belong to the prior art known to professional and technical personnel in the field. Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A fire management system for liquor winery based on machine vision and composite sensors, the system includes a server, a winery fire control center, a smart camera, a composite sensor, linkage equipment, a visual evacuation guidance module, a firefighting drone, a mobile terminal and a computer terminal, and is characterized by: Smart cameras use machine vision technology to collect, analyze and detect fire targets in real time, and make early warning and linkage control decisions based on data from composite sensors. Composite sensors include flame detection sensors, temperature and humidity sensors, and electronic noses to collect environmental data; The linkage equipment is controlled according to the sensor data, including fire shutters, fire extinguishers, sound and light alarms and other equipment. The linkage equipment is connected with smart cameras and the fire control center to trigger emergency operations according to the fire situation; The visual evacuation guidance module includes an LED floor tile display, a wall projection system and a sound prompt system, which is used to provide real-time updates of the best evacuation route for people in the winery when a fire occurs, guiding people to quickly evacuate the dangerous area; Firefighting drones are equipped with high-definition cameras and infrared thermal imagers, which are used for pre-disaster inspections and real-time information feedback during disasters. They fly to the fire area and transmit fire images and data to the fire control center; The server is used to store and manage the system's operating data, including real-time alarm information, sensor data, target detection results, and equipment linkage status, and is connected to mobile terminals and computer terminals through wireless communication for users to view and operate; The winery fire control center is responsible for monitoring the operation of the entire fire protection system, processing data from smart cameras, composite sensors and firefighting drones, and activating related linkage equipment according to preset linkage rules, while generating the best evacuation routes for different areas; The mobile terminal is connected to the server, and the user can view and control the operating status of each module of the system in real time through the mobile terminal, and query the alarm history, sensor data, and linkage device status; The computer terminal is used for system configuration and model training. Users upload and update target detection models and linkage logic scripts through the computer terminal, and configure relevant system parameters.

2. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 1 is characterized in that: The smart camera has an embedded image recognition algorithm and loads different target detection models according to different application scenarios to identify flame and smoke fire targets.

3. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 1 is characterized in that: The image recognition algorithm supports multiple target detection models, including but not limited to the YOLO series models. Users can train the detection model of a specific fire target through a computer and convert it into a model format suitable for the smart camera embedded system.

4. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 3 is characterized in that: The YOLO target detection algorithm supports users to upload pre-trained models according to specific application scenarios and load the models into the camera device through Ethernet or WiFi interface.

5. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 4 is characterized in that: The smart camera performs inference calculations on the collected image data in real time through its internal computing unit, and generates a control signal in combination with the data collected by the composite sensor, and sends it to the server or linkage device.

6. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 1 is characterized in that: The composite sensor communicates with the smart camera and the fire control center via a bus communication method, and the communication method includes but is not limited to RS485, 232, Ethernet, TCP, LoRa and Zigbee.

7. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 1 is characterized in that: The firefighting drone is equipped with a high-definition camera and an infrared thermal imager, and has an automatic obstacle avoidance function. When a disaster occurs, it automatically cruises to a designated area and transmits fire images and thermal imaging data to the fire control center in real time.

8. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 1 is characterized in that: The system uses an electronic nose to monitor alcohol vapor in real time. When the alcohol concentration exceeds a safety threshold, the system automatically alarms and links related equipment to start emergency ventilation.

9. The fire protection management system for liquor winery based on machine vision and composite sensors according to claim 1 is characterized in that: The smart camera works in conjunction with the composite sensor. When the smart camera detects a fire risk, it reads relevant environmental data from the composite sensor through a communication unit, and comprehensively determines the fire risk by combining multi-dimensional data of temperature, humidity, and gas concentration.

Images