Remote fire safety intelligent monitoring method

Abstract

The application relates to the technical field of fire safety supervision, and discloses a remote fire safety intelligent supervision method, which comprises the following steps: S1: multi-dimensional perception network deployment and data collection: deploying Internet of Things terminal equipment in a supervision area, including smoke sensing / temperature sensing detectors, electrical fire monitoring devices, fire water pressure / air pressure sensors, AI intelligent cameras and fireman individual positioning equipment; collecting fire hazard data (smoke density, temperature, current / voltage abnormalities) in real time, realizing low-delay data transmission through 5G+edge computing, and fusing multi-dimensional perception and edge computing in the application: constructing a full-factor perception network, combining edge computing to realize local data preprocessing, reducing transmission delay and bandwidth occupation, and guaranteeing monitoring real-time performance; and adopting a CNN+LSTM hybrid model to realize hazard identification and false alarm identification, so that the false alarm rate is reduced to below 5%, and the pain point of high false alarm rate of a traditional system is solved.

Description

Technical Field

[0001] This application relates to the technical field of fire safety supervision, and in particular to a remote intelligent fire safety supervision method. Background Technology

[0002] With the acceleration of urbanization, fire safety supervision faces prominent problems such as "lagging monitoring, slow response, scattered resources, and high false alarm rate".

[0003] However, in practical applications, most existing traditional fire safety supervision models rely on limited sensing methods: manual inspections and independent smoke detectors fail to achieve full coverage and 24 / 7 monitoring, resulting in delayed hazard detection; false alarm rates remain high: traditional alarm systems lack intelligent identification capabilities, and cooking fumes, steam, and equipment malfunctions easily trigger false alarms. Statistics show that the false alarm rate of traditional fire alarms is as high as 70%, leading to a significant waste of rescue resources; and response strategies are crude: the lack of a risk classification mechanism means that all alarms are handled according to the same standard, resulting in over-handling of general hazards and under-response to major hazards.

[0004] For example, the patent with publication number CN114822998A proposed a smart fire remote monitoring system, but it only realized basic data collection and alarm functions and lacked intelligent risk assessment and hierarchical disposal mechanisms; the patent with publication number CN115662939A used AI for fire identification, but it did not involve multi-department collaborative dispatch and safety assurance during the rescue process.

[0005] Therefore, those skilled in the art have provided a remote intelligent fire safety monitoring method to solve the problems mentioned in the background art. Summary of the Invention

[0006] To address the shortcomings of most traditional fire safety monitoring methods mentioned in the background, such as reliance on manual inspections and independent smoke detectors (failing to achieve full coverage and 24 / 7 monitoring, resulting in delayed hazard detection), high false alarm rates (traditional alarm systems lack intelligent discrimination capabilities, easily triggering false alarms from cooking fumes, steam, equipment malfunctions, etc., with statistics showing a false alarm rate as high as 70%, leading to a significant waste of rescue resources), and inefficient response strategies (the lack of a risk classification mechanism, treating all alarms with the same standard, resulting in over-handling of general hazards and under-response to major hazards), this application provides a remote intelligent fire safety monitoring method.

[0007] This application provides a remote intelligent fire safety monitoring method using the following technical solution, which includes the following steps:

[0008] S1: Multi-dimensional perception network deployment and data collection: Deploy IoT terminal devices in the monitored area, including smoke / heat detectors, electrical fire monitoring devices, fire water pressure / air pressure sensors, AI smart cameras, and individual firefighter positioning devices; collect fire hazard data (smoke concentration, temperature, abnormal current / voltage), fire facility status data (fire extinguisher pressure, fire hydrant water level), on-site image data, and personnel positioning data in real time, and achieve low-latency data transmission through 5G+edge computing;

[0009] S2: Data Fusion Processing and Intelligent Risk Assessment: Construct a two-level analysis architecture of "edge preprocessing + cloud deep learning". Edge nodes perform noise reduction and normalization processing on the collected data. The cloud uses a CNN+LSTM hybrid model to identify fire hazards (accuracy ≥95%) and identify false alarms (false alarm rate ≤5%). Combined with the building 3D model and historical hazard data, the risk level (red / orange / yellow / blue four levels) is intelligently assessed.

[0010] S3: Generation of hierarchical and graded response strategies: Dynamically match response plans based on risk levels. Blue risks (general hazards) push rectification reminders to the responsible units; yellow risks (serious hazards) initiate remote guidance and time-limited rectification; orange risks (major hazards) dispatch the nearest mini fire station for on-site response; red risks (fire alarms) simultaneously link with the 119 command center and public security and medical departments for coordinated rescue.

[0011] S4: Full-process visualized dispatch and resource collaboration: Generate the optimal emergency response route through the remote monitoring platform (avoiding congested sections), and push fire details (burning materials, location of trapped personnel, distribution of fire-fighting facilities) to the rescue terminal; remotely control traffic lights around the fire area to open a green rescue channel, and simultaneously notify medical institutions to prepare for receiving the injured;

[0012] S5: Real-time monitoring and safety assurance of the rescue process: Real-time monitoring of rescue progress through on-site video streams and firefighters' physiological monitoring data (heart rate, blood oxygen saturation), push risk area warnings and safe evacuation routes based on thermal imaging analysis, and automatically trigger evacuation reminders when personnel safety risks occur;

[0013] S6: Closed-loop management and model iteration optimization: After the fire is handled, the handling data (response time, rescue duration, resource consumption, and hidden danger rectification rate) is automatically summarized to generate a review report; the risk assessment model and handling strategy are iteratively optimized based on machine learning algorithms to form a fully closed-loop supervision system of "perception-assessment-handling-feedback".

[0014] Preferably, the IoT terminal device in step S1 further includes:

[0015] Fire lane occupancy monitoring sensors, fire door status sensors, and emergency lighting equipment status monitors;

[0016] The terminal device supports edge computing, which can complete data preprocessing and anomaly threshold judgment locally, and only upload abnormal data and key image information, thereby reducing transmission bandwidth usage.

[0017] Preferably, the intelligent risk assessment in step S2 specifically includes:

[0018] Hazard identification: AI algorithms are used to identify 12 common fire hazards, including aging electrical wiring, unauthorized hot work, blocked fire exits, and fire doors not being closed.

[0019] Fire detection: By combining the rate of change of smoke concentration, temperature gradient, and cross-validation of video image features, false alarm scenarios such as cooking fumes and steam can be distinguished from real fires;

[0020] Risk level assessment: An assessment model is constructed based on the severity of the hazard, the scope of impact, the population density, and the fire resistance rating of the building. The assessment time is ≤3 seconds.

[0021] Preferably, the triggering condition for the hierarchical and graded handling strategy in step S3 is:

[0022] Blue risk: Single non-critical hidden danger with no immediate safety threat (such as slightly low fire extinguisher pressure or partial damage to emergency lighting).

[0023] Yellow risk: Multiple general hazards combined or critical hazards (such as minor overload of electrical wiring or partial obstruction of fire escape routes);

[0024] Orange risk: High-risk hazards with a tendency to spread (such as abnormally high local temperatures or excessive smoke concentration).

[0025] Red risk: A fire has been confirmed, with open flames, large amounts of smoke, or people trapped.

[0026] Preferably, the security mechanism in step S5 includes:

[0027] Firefighter positioning: UWB+GPS dual-mode positioning is adopted, with indoor positioning accuracy ≤1 meter and outdoor positioning accuracy ≤3 meters;

[0028] Physiological status monitoring: Real-time collection of heart rate (normal range 60-120 beats / minute) and blood oxygen saturation (normal range ≥95%) via smart wearable devices, with automatic alarm when exceeding the threshold;

[0029] Risk area warning: Based on thermal imaging data, identify high-temperature areas (≥300℃) and areas with excessive concentrations of toxic gases, and mark collapse risk points.

[0030] Preferably, the closed-loop management in step S6 further includes:

[0031] Hazard rectification tracking: The rectification status is checked in real time through IoT devices. Once the rectification is completed, the issue is automatically closed. If the issue is not rectified within the specified time, the action will be escalated.

[0032] Data sharing: Synchronize regulatory data with local fire safety regulatory departments to provide data support for law enforcement inspections and achieve "precise supervision and targeted enforcement";

[0033] Model optimization: The algorithm model is updated quarterly based on newly added hidden danger data and handling cases, which improves the risk identification accuracy by 3%-5% annually.

[0034] In summary, this application includes the following beneficial technical effects:

[0035] 1. Multi-dimensional perception + edge computing fusion: Construct a full-element perception network, combine edge computing to realize local data preprocessing, reduce transmission latency and bandwidth consumption, and ensure real-time monitoring;

[0036] 2. AI dual-level judgment mechanism: The CNN+LSTM hybrid model is used to realize the identification of hidden dangers and the identification of false alarms, reducing the false alarm rate to below 5%, thus solving the pain point of high false alarm rate of traditional systems;

[0037] 3. Red, orange, yellow, and blue risk-based differentiated response: Matching differentiated response plans based on risk levels to avoid resource waste and insufficient response, and improve the accuracy of supervision;

[0038] 4. Cross-departmental collaborative dispatch: Integrating resources and data from multiple departments to enable collaborative operations among fire, public security, and medical services, opening green rescue channels, and shortening response time;

[0039] 5. Dual protection for rescue safety: Integrating dual-mode positioning, physiological monitoring, and thermal imaging risk warning, it comprehensively protects the lives of firefighters;

[0040] 6. Closed-loop iterative optimization: Establish a "disposal-review-optimization" mechanism to continuously improve regulatory and rescue capabilities through data-driven approaches. Detailed Implementation

[0041] The described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without inventive effort are within the scope of protection of the present invention.

[0042] This application discloses a remote intelligent fire safety monitoring method, including the following steps:

[0043] S1: Multi-dimensional perception network deployment and data collection: Deploy IoT terminal devices in the monitored area, including smoke / heat detectors, electrical fire monitoring devices, fire water pressure / air pressure sensors, AI smart cameras, and individual firefighter positioning devices; collect fire hazard data (smoke concentration, temperature, abnormal current / voltage), fire facility status data (fire extinguisher pressure, fire hydrant water level), on-site image data, and personnel positioning data in real time, and achieve low-latency data transmission through 5G+edge computing;

[0044] S2: Data Fusion Processing and Intelligent Risk Assessment: Construct a two-level analysis architecture of "edge preprocessing + cloud deep learning". Edge nodes perform noise reduction and normalization processing on the collected data. The cloud uses a CNN+LSTM hybrid model to identify fire hazards (accuracy ≥95%) and identify false alarms (false alarm rate ≤5%). Combined with the building 3D model and historical hazard data, the risk level (red / orange / yellow / blue four levels) is intelligently assessed.

[0045] S3: Generation of hierarchical and graded response strategies: Dynamically match response plans based on risk levels. Blue risks (general hazards) push rectification reminders to the responsible units; yellow risks (serious hazards) initiate remote guidance and time-limited rectification; orange risks (major hazards) dispatch the nearest mini fire station for on-site response; red risks (fire alarms) simultaneously link with the 119 command center and public security and medical departments for coordinated rescue.

[0046] S4: Full-process visualized dispatch and resource collaboration: Generate the optimal emergency response route through the remote monitoring platform (avoiding congested sections), and push fire details (burning materials, location of trapped personnel, distribution of fire-fighting facilities) to the rescue terminal; remotely control traffic lights around the fire area to open a green rescue channel, and simultaneously notify medical institutions to prepare for receiving the injured;

[0047] S5: Real-time monitoring and safety assurance of the rescue process: Real-time monitoring of rescue progress through on-site video streams and firefighters' physiological monitoring data (heart rate, blood oxygen saturation), push risk area warnings and safe evacuation routes based on thermal imaging analysis, and automatically trigger evacuation reminders when personnel safety risks occur;

[0048] S6: Closed-loop management and model iteration optimization: After the fire is handled, the handling data (response time, rescue duration, resource consumption, and hidden danger rectification rate) is automatically summarized to generate a review report; the risk assessment model and handling strategy are iteratively optimized based on machine learning algorithms to form a fully closed-loop supervision system of "perception-assessment-handling-feedback".

[0049] In this application, the IoT terminal device in step S1 further includes:

[0050] Fire lane occupancy monitoring sensors, fire door status sensors, and emergency lighting equipment status monitors;

[0051] The terminal device supports edge computing, which can complete data preprocessing and anomaly threshold judgment locally, and only upload abnormal data and key image information, thereby reducing transmission bandwidth usage.

[0052] In this application, the intelligent risk assessment in step S2 specifically includes:

[0053] Hazard identification: AI algorithms are used to identify 12 common fire hazards, including aging electrical wiring, unauthorized hot work, blocked fire exits, and fire doors not being closed.

[0054] Fire detection: By combining the rate of change of smoke concentration, temperature gradient, and cross-validation of video image features, false alarm scenarios such as cooking fumes and steam can be distinguished from real fires;

[0055] Risk level assessment: An assessment model is constructed based on the severity of the hazard, the scope of impact, the population density, and the fire resistance rating of the building. The assessment time is ≤3 seconds.

[0056] In this application, the triggering condition for the hierarchical and graded handling strategy in step S3 is:

[0057] Blue risk: Single non-critical hidden danger with no immediate safety threat (such as slightly low fire extinguisher pressure or partial damage to emergency lighting).

[0058] Yellow risk: Multiple general hazards combined or critical hazards (such as minor overload of electrical wiring or partial obstruction of fire escape routes);

[0059] Orange risk: High-risk hazards with a tendency to spread (such as abnormally high local temperatures or excessive smoke concentration).

[0060] Red risk: A fire has been confirmed, with open flames, large amounts of smoke, or people trapped.

[0061] In this application, the security mechanism in step S5 includes:

[0062] Firefighter positioning: UWB+GPS dual-mode positioning is adopted, with indoor positioning accuracy ≤1 meter and outdoor positioning accuracy ≤3 meters;

[0063] Physiological status monitoring: Real-time collection of heart rate (normal range 60-120 beats / minute) and blood oxygen saturation (normal range ≥95%) via smart wearable devices, with automatic alarm when exceeding the threshold;

[0064] Risk area warning: Based on thermal imaging data, identify high-temperature areas (≥300℃) and areas with excessive concentrations of toxic gases, and mark collapse risk points.

[0065] In this application, the closed-loop management in step S6 also includes:

[0066] Hazard rectification tracking: The rectification status is checked in real time through IoT devices. Once the rectification is completed, the issue is automatically closed. If the issue is not rectified within the specified time, the action will be escalated.

[0067] Data sharing: Synchronize regulatory data with local fire safety regulatory departments to provide data support for law enforcement inspections and achieve "precise supervision and targeted enforcement";

[0068] Model optimization: The algorithm model is updated quarterly based on newly added hidden danger data and handling cases, which improves the risk identification accuracy by 3%-5% annually.

[0069] All standard parts used in this invention can be purchased from the market, and irregular parts can be customized according to the description in the specification. The specific connection methods of each part adopt conventional methods such as bolts, rivets, and welding that are mature in the prior art. The machinery, parts and equipment adopt conventional models in the prior art, and the circuit connection adopts conventional connection methods in the prior art, which will not be described in detail here.

[0070] Finally, the following points should be noted: First, in the description of this application, it should be noted that, unless otherwise specified and limited, the terms "installation", "connection", and "linkage" should be interpreted broadly, and can refer to mechanical or electrical connections, or internal connections between two components, or direct connections. "Up", "down", "left", "right", etc., are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may change.

Claims

1. A remote intelligent fire safety monitoring method, characterized in that, Includes the following steps: S1: Multi-dimensional perception network deployment and data collection: Deploy IoT terminal devices in the monitored area, including smoke / heat detectors, electrical fire monitoring devices, fire water pressure / air pressure sensors, AI smart cameras, and individual firefighter positioning devices; collect fire hazard data (smoke concentration, temperature, abnormal current / voltage), fire facility status data (fire extinguisher pressure, fire hydrant water level), on-site image data, and personnel positioning data in real time, and achieve low-latency data transmission through 5G+edge computing; S2: Data Fusion Processing and Intelligent Risk Assessment: Construct a two-level analysis architecture of "edge preprocessing + cloud deep learning". Edge nodes perform noise reduction and normalization processing on the collected data. The cloud uses a CNN+LSTM hybrid model to identify fire hazards (accuracy ≥95%) and identify false alarms (false alarm rate ≤5%). Combined with the building 3D model and historical hazard data, the risk level (red / orange / yellow / blue four levels) is intelligently assessed. S3: Generation of hierarchical and graded response strategies: Dynamically match response plans based on risk levels. Blue risks (general hazards) push rectification reminders to the responsible units; yellow risks (serious hazards) initiate remote guidance and time-limited rectification; orange risks (major hazards) dispatch the nearest mini fire station for on-site response; red risks (fire alarms) simultaneously link with the 119 command center and public security and medical departments for coordinated rescue. S4: Full-process visualized dispatch and resource collaboration: Generate the optimal emergency response route through the remote monitoring platform (avoiding congested sections), and push fire details (burning materials, location of trapped personnel, distribution of fire-fighting facilities) to the rescue terminal; remotely control traffic lights around the fire area to open a green rescue channel, and simultaneously notify medical institutions to prepare for receiving the injured; S5: Real-time monitoring and safety assurance of the rescue process: Real-time monitoring of rescue progress through on-site video streams and firefighters' physiological monitoring data (heart rate, blood oxygen saturation), push risk area warnings and safe evacuation routes based on thermal imaging analysis, and automatically trigger evacuation reminders when personnel safety risks occur; S6: Closed-loop management and model iteration optimization: After the fire is handled, the handling data (response time, rescue duration, resource consumption, and hidden danger rectification rate) is automatically summarized to generate a review report; the risk assessment model and handling strategy are iteratively optimized based on machine learning algorithms to form a fully closed-loop supervision system of "perception-assessment-handling-feedback".

2. The method according to claim 1, characterized in that, The IoT terminal device mentioned in step S1 also includes: Fire lane occupancy monitoring sensors, fire door status sensors, and emergency lighting equipment status monitors; The terminal device supports edge computing, which can complete data preprocessing and anomaly threshold judgment locally, and only upload abnormal data and key image information, thereby reducing transmission bandwidth usage.

3. The method according to claim 1, characterized in that, The intelligent risk assessment described in step S2 specifically includes: Hazard identification: AI algorithms are used to identify 12 common fire hazards, including aging electrical wiring, unauthorized hot work, blocked fire exits, and fire doors not being closed. Fire detection: By combining the rate of change of smoke concentration, temperature gradient, and cross-validation of video image features, false alarm scenarios such as cooking fumes and steam can be distinguished from real fires; Risk level assessment: An assessment model is constructed based on the severity of the hazard, the scope of impact, the population density, and the fire resistance rating of the building. The assessment time is ≤3 seconds.

4. The method according to claim 1, characterized in that, The triggering condition for the hierarchical and graded handling strategy described in step S3 is: Blue risk: Single non-critical hidden danger with no immediate safety threat (such as slightly low fire extinguisher pressure or partial damage to emergency lighting). Yellow risk: Multiple general hazards combined or critical hazards (such as minor overload of electrical wiring or partial obstruction of fire escape routes); Orange risk: High-risk hazards with a tendency to spread (such as abnormally high local temperatures or excessive smoke concentration). Red risk: A fire has been confirmed, with open flames, large amounts of smoke, or people trapped.

5. The method according to claim 1, characterized in that, The security mechanism described in step S5 includes: Firefighter positioning: UWB+GPS dual-mode positioning is adopted, with indoor positioning accuracy ≤1 meter and outdoor positioning accuracy ≤3 meters; Physiological status monitoring: Real-time collection of heart rate (normal range 60-120 beats / minute) and blood oxygen saturation (normal range ≥95%) via smart wearable devices, with automatic alarm when exceeding the threshold; Risk area warning: Based on thermal imaging data, identify high-temperature areas (≥300℃) and areas with excessive concentrations of toxic gases, and mark collapse risk points.

6. The method according to claim 1, characterized in that, The closed-loop management described in step S6 also includes: Hazard rectification tracking: The rectification status is checked in real time through IoT devices. Once the rectification is completed, the issue is automatically closed. If the issue is not rectified within the specified time, the action will be escalated. Data sharing: Synchronize regulatory data with local fire safety regulatory departments to provide data support for law enforcement inspections and achieve "precise supervision and targeted enforcement"; Model optimization: The algorithm model is updated quarterly based on newly added hidden danger data and handling cases, which improves the risk identification accuracy by 3%-5% annually.