Smoke sensor fault monitoring system based on forest fire monitoring

Abstract

The invention relates to the technical field of fault monitoring, in particular to a smoke sensor fault monitoring system based on forest fire monitoring, which comprises distributed monitoring units and constructs a fire monitoring network, the fire monitoring network comprises a front-end monitoring node and a data processing platform, and the front-end monitoring node comprises a smoke sensor, a temperature sensor and a humidity sensor; the infrared monitoring unit is used for acquiring an infrared image in a fire monitoring area; the fault detection unit is used for marking abnormal monitoring points; and the abnormity analysis unit determines a corresponding real-time influence range based on the position, the wind direction and the wind speed of the abnormal monitoring point and a preset induction range, records the real-time influence range as a monitoring area of the abnormal monitoring point, and checks whether the sensor has a fault based on the abnormal temperature value in the infrared image of the monitoring area. According to the invention, targeted verification is carried out on the abnormal front-end monitoring node, whether the abnormality is from a forest fire or a sensor fault is judged, and then active identification and exploration of the fault are realized.

Description

Technical Field

[0001] This invention relates to the field of fault monitoring technology, and more specifically to a smoke sensor fault monitoring system based on forest fire monitoring. Background Technology

[0002] In modern forest fire prevention systems, smoke sensors, as core components of the front-end sensing network, play an irreplaceable role in early warning. Fire monitoring systems, through the scientific deployment of intelligent monitoring nodes integrating smoke detection, environmental information collection, wireless communication, and solar power at key locations in forest areas, construct a widely covered, all-weather IoT sensing layer.

[0003] Early signs of forest fires are usually smoldering without open flames. The main characteristic of this type of combustion is the production of large amounts of smoke without obvious flames. Therefore, smoke sensors can be used to detect forest fires in their early stages. However, it is assumed that the failure rate of smoke sensors in the field will increase with the increase of interference sources, which can easily lead to false alarms. Therefore, there is an urgent need for a fault monitoring system to verify the anomalies reported by smoke sensors and distinguish between smoke sensor failures and actual forest fires. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the existing technology, the present invention provides a smoke sensor fault monitoring system based on forest fire monitoring, which can effectively solve the problem of difficulty in monitoring faults of smoke sensors installed in the field in the existing technology.

[0005] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a smoke sensor fault monitoring system based on forest fire monitoring, comprising at least: Distributed monitoring units are used to construct a fire monitoring network, which includes a cloud monitoring center, multiple front-end monitoring nodes, and multiple data processing platforms. The front-end monitoring nodes include smoke sensors, temperature sensors, and humidity sensors. Based on the number of data processing platforms, the fire monitoring area is divided into multiple unit monitoring areas, and multiple front-end monitoring nodes within a unit monitoring area are controlled by the data processing platform. The infrared monitoring unit monitors the thermal imaging of the fire monitoring area and collects thermal imaging images of the monitoring areas of each unit within the fire monitoring area, which are recorded as unit infrared images. The fault detection unit marks abnormal monitoring points based on monitoring data uploaded by multiple front-end monitoring nodes. The uploading of data by the front-end monitoring nodes is divided into periodic uploading and conditional uploading. The anomaly analysis unit determines the corresponding real-time impact range based on the location, wind direction, wind speed, and preset sensing range of the anomaly monitoring point and records it as the monitoring area of ​​the anomaly monitoring point. It extracts the infrared image corresponding to the monitoring area and records it as the target infrared image. Based on the abnormal temperature value in the target infrared image, it checks whether the sensor is faulty.

[0006] Furthermore, a smoke index threshold and a smoke change threshold are set for the smoke sensor, a temperature index threshold and a temperature change threshold are set for the temperature sensor, and a humidity index threshold and a humidity change threshold are set for the humidity sensor. The conditions for conditional upload include: Condition 1: The smoke index detected by the smoke sensor is greater than the smoke index threshold or the change in the smoke index per unit time is greater than the smoke change threshold. Condition 2: The temperature index captured by the temperature sensor is greater than the temperature index threshold or the change in temperature index per unit time is greater than the temperature change threshold; Condition 3: The humidity index captured by the humidity sensor is greater than the humidity index threshold or the change in humidity index per unit time is greater than the humidity change threshold. When any of the above conditions are met, conditional upload will be performed.

[0007] Furthermore, when the uploaded data comes from conditional uploads, the corresponding monitoring point is marked as an abnormal monitoring point. When the uploaded data comes from periodic uploads, the average value of each monitoring index is calculated. When the difference between any monitoring index of any monitoring point and the corresponding average value is greater than the preset deviation threshold, it is recorded as an abnormal monitoring point.

[0008] Furthermore, the process for determining the real-time impact range is as follows: A digital elevation model (DEM) is constructed within the fire monitoring area. Vegetation cover within the fire monitoring area is identified and combined with the DEM to obtain a vegetation cover model. The fire monitoring area is divided into multiple coverage areas with different vegetation cover types. An initial range is drawn with a preset sensing range as the radius. Multiple points evenly distributed on the circumference of the initial range are recorded as initial boundary points. Based on the vegetation cover between the abnormal monitoring points and the initial boundary points, combined with wind speed and wind direction, the positions of multiple initial boundary points are corrected to obtain discrete boundary points. Multiple discrete boundary points are connected sequentially to obtain the real-time influence range.

[0009] Furthermore, the initial boundary point position correction process is as follows: The connecting line segment between the initial boundary point and the anomaly monitoring point is called the initial line segment. The straight line where the initial line segment is located is called the reference straight line. A rectangular coordinate system is constructed with the anomaly monitoring point as the origin and called the reference coordinate system. The vertical axis of the reference coordinate system extends along the wind direction. A base combustion rate is assigned to each vegetation cover type. The initial line segment is divided into multiple unit initial line segments in different vegetation cover areas. Based on the base combustion rate and combined with wind direction and wind speed, the conduction speed corresponding to the unit initial line segment is analyzed. The conduction distance is divided by the conduction speed to obtain the conduction time of the unit initial line segment. The time difference between the current time and the last upload time of the anomaly monitoring point is recorded as the blind zone time. The sum of multiple transmission times corresponding to the initial line segment is recorded as the influence duration and compared with the blind zone time. When the blind zone time is less than or equal to the influence duration, a corresponding discrete boundary point is determined between the initial boundary point and the anomaly monitoring point. When the blind zone time is greater than the influence duration, a corresponding discrete boundary point is determined on the side of the initial boundary point away from the anomaly monitoring point. The discrete boundary point satisfies the following condition: Condition A: It lies on the initial line segment; Condition B: The influence time corresponding to the discrete boundary point is equal to the blind zone time.

[0010] Furthermore, the sensing range is equal to the maximum value of the blind zone time multiplied by the base combustion rate.

[0011] Furthermore, the conduction velocity analysis process is as follows: The conduction velocity is equal to the vector sum of the wind direction velocity and the natural conduction velocity, with the wind direction velocity along the vertical axis of the reference coordinate system and the natural conduction velocity along the horizontal axis of the reference coordinate system. Wind speed is equal to the natural conduction speed multiplied by the wind force influence coefficient. The natural conduction speed is equal to the base combustion rate of the corresponding vegetation cover route. When the fire spreads along the initial line segment to the abnormal monitoring point in a headwind manner, the wind force influence coefficient is equal to 1 minus the wind speed influence coefficient. When the fire spreads along the initial line segment to the abnormal monitoring point in a tailwind manner, the wind force influence coefficient is equal to 1 plus the wind speed influence coefficient. The wind speed influence coefficient is equal to the wind speed multiplied by a preset proportional coefficient.

[0012] Furthermore, the vegetation cover identification process is as follows: A drone equipped with a camera performs a comprehensive scan of the fire monitoring area, collecting aerial images of the area. These images are then divided into multiple unit identification regions. Based on image recognition algorithms, the coverage percentage of each vegetation type is determined. The vegetation type with the largest coverage percentage within a unit identification region is used as the labeled coverage type for that region. Adjacent unit identification regions with the same labeled coverage type are merged to obtain multiple shrub coverage regions, forest coverage regions, and grassland coverage regions.

[0013] Furthermore, the drone's flight altitude control process is as follows: The camera parameters of the drone are obtained, including sensor width, pixel width and lens focal length. A sharpness index range and a flight altitude calculation formula are preset to calculate the flight altitude range, so that the drone's altitude during flight is within the flight altitude range.

[0014] Furthermore, the formula for calculating flight altitude is: Flight altitude = (focal length * resolution index) / (sensor width / pixel width).

[0015] The technical solution provided by this invention has the following advantages compared with the known prior art: 1. This invention uses multiple distributed front-end monitoring nodes to conduct comprehensive fire monitoring of the fire monitoring area. It combines two data upload modes, periodic upload and conditional upload, to achieve dynamic monitoring of temperature, humidity and smoke concentration within the fire monitoring area. It can also promptly detect front-end monitoring nodes with abnormal data, and then perform targeted verification on the abnormal front-end monitoring nodes to determine whether the abnormality originates from a forest fire or a sensor malfunction, thereby achieving proactive fault identification and investigation.

[0016] 2. When verifying abnormal front-end monitoring nodes, this invention can intelligently delineate the corresponding monitoring area based on the vegetation coverage around the abnormal monitoring point and the current wind speed and direction, thereby clarifying the affected range of the abnormal monitoring point. Then, it analyzes the infrared images within this range to determine whether a fire source actually exists, thus determining the cause of the data anomaly. Compared with the manual verification method in the prior art, it can timely and accurately delineate the infrared verification range, achieving timely response of one-to-one verification, thereby improving the accuracy and efficiency of fire monitoring and early warning. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the accompanying drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are merely some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without any creative effort.

[0018] Figure 1 This is an overall module block diagram of the present invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, 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 creative effort are within the scope of protection of the present invention.

[0020] The present invention will be further described below with reference to embodiments.

[0021] See Figure 1 A smoke sensor fault monitoring system based on forest fire monitoring includes at least: Distributed monitoring units are used to construct a fire monitoring network, which includes a cloud monitoring center, multiple front-end monitoring nodes, and multiple data processing platforms.

[0022] The front-end monitoring node includes a LoRa communication module (LoRa communication technology is existing technology and will not be elaborated on here), a solar panel (usually 20W-100W), a battery (such as a lithium battery of 12V / 20Ah or higher), and various sensors (fire monitoring areas are usually in the wild forest areas, where it is difficult to achieve power coverage, especially uniform and comprehensive power coverage. Therefore, the front-end monitoring node needs to be equipped with an independent solar power supply device to achieve power self-sufficiency). The types of sensors include, but are not limited to, smoke sensors, temperature sensors, and humidity sensors. Each front-end monitoring node is equipped with a main control unit for control (recording, processing, and uploading of sensor monitoring data).

[0023] Based on the number of data processing platforms, the fire monitoring area is divided into multiple unit monitoring areas. Each data processing platform manages one unit monitoring area. Wireless data transmission is established between the platform and multiple front-end monitoring nodes within the unit monitoring area through a LoRa gateway. The data processing platform is set up at a high point (such as a tower or mountaintop) within the unit monitoring area. The data processing platform forwards all data to the cloud monitoring center (the connection between the data processing platform and the cloud monitoring center adopts one or more of 4G, satellite network and wired network).

[0024] The cloud-based monitoring center receives, stores, and processes data uploaded by front-end monitoring nodes, performs fire analysis, fault diagnosis, and visualization, and sends anomaly signals to management personnel when anomalies are detected.

[0025] It should be noted that the forest fire monitoring network achieves comprehensive monitoring of fire conditions within the forest area through distributed front-end monitoring nodes, and is uniformly monitored and managed by the cloud. This enables rapid detection, response, and handling of fires across the entire area. When a fire occurs at any location within the forest area, it will cause abnormal data from the surrounding sensors, thereby generating a fire warning and effectively improving the response speed and handling efficiency of fires.

[0026] The infrared monitoring unit uses multiple thermal imaging monitoring devices mounted at high points to perform thermal imaging monitoring of the fire monitoring area, and records the thermal imaging images corresponding to the monitoring areas of each unit within the fire monitoring area as unit infrared images.

[0027] The fault detection unit marks abnormal monitoring points based on monitoring data uploaded by multiple front-end monitoring nodes. The uploading of data by the front-end monitoring nodes is divided into periodic uploading and conditional uploading. The uploading period for periodic uploading is a preset value (1 hour in a specific embodiment). The condition for conditional uploading is that the sensing value of any sensor exceeds the corresponding preset threshold or the magnitude of the change in the sensing value of any sensor within a unit of time exceeds the corresponding preset threshold.

[0028] Specifically, a smoke index threshold and a smoke change threshold are set for the smoke sensor, a temperature index threshold and a temperature change threshold are set for the temperature sensor, and a humidity index threshold and a humidity change threshold are set for the humidity sensor. The conditions for uploading include: Condition 1: The smoke index detected by the smoke sensor is greater than the smoke index threshold or the change in the smoke index per unit time is greater than the smoke change threshold. Condition 2: The temperature index captured by the temperature sensor is greater than the temperature index threshold or the change in temperature index per unit time is greater than the temperature change threshold; Condition 3: The humidity index captured by the humidity sensor is greater than the humidity index threshold or the change in humidity index per unit time is greater than the humidity change threshold. When any of the above conditions are met, conditional upload will be performed.

[0029] It should be noted that, under normal circumstances, the monitoring data from multiple sensors at the front-end monitoring node are in a stable state or fluctuate within a small range. When the sensor values ​​fluctuate abnormally, it usually means that there is an anomaly near the front-end monitoring node, such as a forest fire (waiting for regular uploads to provide early warnings for sudden forest fires would delay the time of fire detection). Therefore, conditional uploads can make up for the shortcomings of regular uploads in not being able to detect anomalies in the unit's monitoring area in a timely manner, and the two complement each other.

[0030] It is worth noting that the operation of multiple sensors involves periodically collecting and monitoring data. For example, the smoke sensor collects the smoke index (a dimensionless value) every 30 seconds. Therefore, the data it records can be represented as a sequence with a fixed collection interval between adjacent values ​​and each value in the sequence corresponding to a timestamp (collection time).

[0031] When the uploaded data comes from conditional upload, the corresponding monitoring point is marked as an abnormal monitoring point. When the uploaded data comes from periodic upload, the average value of each monitoring index is calculated. When the difference between any monitoring index of any monitoring point and the corresponding average value is greater than the preset deviation threshold, it is recorded as an abnormal monitoring point.

[0032] Specifically, taking the smoke index as an example, first calculate the average smoke index from the periodically uploaded data, then compare the smoke index of each monitoring point with the average smoke index, and finally divide the difference between the two by the average smoke index to obtain the ratio (the calculation formula is expressed as follows). R represents the smoke index. When the average value of the smoke index is greater than a preset deviation threshold (30% in one specific embodiment), the corresponding monitoring point is marked as an abnormal monitoring point.

[0033] The anomaly analysis unit draws a plan view of the fire monitoring area, recorded as a monitoring map, and marks each abnormal monitoring point on the monitoring map. Based on the location of the abnormal monitoring point, meteorological data (wind direction and wind speed) within the fire monitoring area, and the preset sensing range, it determines the corresponding real-time impact range and records it as the monitoring area of ​​the abnormal monitoring point. The real-time impact range refers to the range within which a forest fire would cause data anomalies at the corresponding abnormal monitoring point. It captures the infrared image corresponding to the monitoring area and records it as the target infrared image. Based on the target infrared image, it determines the temperature value of any point within the monitoring area. When the temperature value of any point in the target infrared image is greater than or equal to a preset temperature threshold, a forest fire alarm signal is generated. Conversely, when the temperature value of any point in the target infrared image is less than the preset temperature threshold, the corresponding abnormal monitoring point is marked as a fault monitoring point, and a sensor fault signal is generated.

[0034] It should be noted that the target infrared image can intuitively and clearly show whether there are abnormal surface temperature monitoring data in the fire monitoring area. This difference is usually caused by forest fires. Therefore, combining the target infrared image with surface temperature verification can determine whether the abnormal data at the monitoring point is caused by a fault or a forest fire, thus improving the efficiency and accuracy of fire monitoring.

[0035] It is worth noting that different sensors have different monitoring functions and ranges. Humidity and temperature sensors can usually only monitor temperature and humidity anomalies in a small surrounding area. In other words, they are difficult to monitor fires that have not yet spread to the vicinity of the front-end monitoring node. On the other hand, smoke sensors monitor smoke and dust that spread with the air, so their effective monitoring range is wider than that of temperature and humidity sensors. This allows them to detect fires at greater distances earlier. As a result, the detection and uploading of smoke index anomalies usually occurs earlier than that of temperature and humidity anomalies. In this case, the authenticity of the data anomaly needs to be verified, and this verification requires infrared monitoring data for corroboration. In addition, forest fire alarm information corroborated by infrared monitoring data is more accurate and effective (the situation where both monitoring methods produce false alarms simultaneously is extremely rare), improving the robustness of the entire forest fire monitoring system.

[0036] Specifically, the process for determining the real-time impact range is as follows: A digital elevation model (DEM) is constructed within the fire monitoring area (obtained through satellite remote sensing, aerial photogrammetry, or UAV aerial photography). Vegetation cover within the fire monitoring area is identified and combined with the DEM to obtain a vegetation cover model. The vegetation cover model provides elevation data and vegetation cover type for any location. Vegetation cover types include shrub cover, forest cover, and grassland cover. The coverage area corresponding to each vegetation cover type is determined. An initial range is drawn with the abnormal monitoring point as the center and the preset sensing range as the radius. Multiple points evenly distributed on the circumference of the initial range are recorded as initial boundary points. Based on the vegetation cover between the abnormal monitoring point and the initial boundary points, combined with wind speed and direction, the positions of multiple initial boundary points are corrected to obtain discrete boundary points. Multiple discrete boundary points are connected sequentially to obtain the real-time influence range.

[0037] It should be noted that existing technologies typically use a circular area centered on the abnormal monitoring point as the monitoring area of ​​the abnormal monitoring point to represent the affected range of the abnormal monitoring point (thereby determining the possible range of the fire). However, in real-world scenarios, the monitoring area corresponding to the abnormal monitoring point is usually affected by a variety of factors and is not suitable to be represented by a simple and fixed circle. By determining the real-time impact range, the monitoring area corresponding to the abnormal monitoring point can be represented more accurately, which helps to determine the location of the fire and monitor the sensor failure at the corresponding location.

[0038] The initial boundary point position correction process is as follows: The connecting line segment between the initial boundary point and the anomaly monitoring point is called the initial line segment. The straight line where the initial line segment is located is called the reference straight line. A rectangular coordinate system is constructed with the anomaly monitoring point as the origin and called the reference coordinate system. The vertical axis of the reference coordinate system extends along the wind direction. A base burning rate is assigned to each vegetation cover type (used to represent the fire spread speed in different vegetation cover areas under windless conditions; this rate is set by staff based on experience, and generally, the fire spread speed of grassland is greater than that of shrubs, which is greater than that of trees). The initial line segment is divided into multiple unit initial line segments in different vegetation cover areas. Based on the base burning rate and combined with wind direction and wind speed, the conduction speed corresponding to the unit initial line segment is analyzed. The conduction distance is divided by the conduction speed to obtain the conduction time of the unit initial line segment. The time difference between the current time and the last upload time of the anomaly monitoring point is recorded as the blind zone time. The sum of multiple propagation times corresponding to the initial line segment is recorded as the influence duration and compared with the blind zone time. When the blind zone time is less than or equal to the influence duration, a corresponding discrete boundary point is determined between the initial boundary point and the anomaly monitoring point. When the blind zone time is greater than the influence duration, a corresponding discrete boundary point is determined on the side of the initial boundary point away from the anomaly monitoring point (on the initial line segment). The discrete boundary point satisfies the following condition: Condition A: It lies on the initial line segment; Condition B: The influence time (calculated in the same way as above) of the discrete boundary point is equal to the blind zone time.

[0039] It should be noted that the basic burning rate corresponding to different vegetation cover types can be obtained by staff through simulation tests. During the simulation tests, the vegetation type should be consistent with that in the fire monitoring area to ensure the accuracy of the simulation results.

[0040] The sensing range is equal to the maximum value of the blind zone time multiplied by the base combustion rate, which gives the initial range a relatively large redundancy space. Then, by analyzing the infrared anomalies in the initial range, the abnormal data reported by the abnormal monitoring points can be checked and verified.

[0041] Furthermore, the conduction velocity analysis process is as follows: The conduction velocity is equal to the vector sum of the wind direction velocity and the natural conduction velocity, with the wind direction velocity along the vertical axis of the reference coordinate system and the natural conduction velocity along the horizontal axis of the reference coordinate system. Wind speed is equal to the natural conduction speed multiplied by the wind force influence coefficient. The natural conduction speed is equal to the base combustion rate of the corresponding vegetation cover route. When the fire spreads along the initial line segment towards the abnormal monitoring point in a headwind manner, the wind force influence coefficient equals 1 minus the wind speed influence coefficient (expressed by the formula). , Indicates the wind force influence coefficient. (This represents the wind speed influence coefficient). When the fire spreads along the initial line segment towards the abnormal monitoring point in a downwind manner, the wind force influence coefficient equals 1 plus the wind speed influence coefficient (expressed by the formula). ); The wind speed influence coefficient is directly proportional to the wind speed. (The formula is expressed as follows) , (This represents wind speed, where k is a preset proportionality coefficient.) More specifically, the vegetation cover identification process is as follows: A drone equipped with a camera performs a comprehensive scan of the fire monitoring area, collecting aerial images of the area. These images are then divided into multiple unit identification regions. Based on image recognition algorithms (such as convolutional neural networks CNN), machine vision recognition is performed on the images within each unit identification region to determine the coverage percentage of each vegetation type. The vegetation type with the largest coverage percentage within a unit identification region is used as the labeled coverage type for that region. Adjacent unit identification regions with the same labeled coverage type are merged to obtain multiple shrub coverage regions, forest coverage regions, and grassland coverage regions.

[0042] It should be noted that shrub cover refers to ground cover primarily consisting of clump-forming shrubs, forest cover refers to ground cover primarily consisting of trees, and grassland cover refers to ground cover primarily consisting of herbaceous plants. The aerial images corresponding to different vegetation cover types show significant differences (due to significant differences in stem structure, core state, etc., the corresponding image texture, color distribution, and outline shape are significantly different, which can be distinguished by image recognition algorithms). By distinguishing different cover areas, we can understand the vegetation distribution within the fire monitoring area, thereby clarifying the impact of vegetation cover on the spread of fire after a fire occurs at different locations.

[0043] In another embodiment, the coverage type of the unit identification area can be divided by manual division and labeling. In other words, the delineation of vegetation coverage area does not have to rely solely on image recognition algorithms.

[0044] Furthermore, the drone's flight altitude control process is as follows: Obtain the camera parameters of the drone, including sensor width, pixel width, and lens focal length. A resolution index range (typically 5-7 cm / pixel) and a flight altitude calculation formula are preset to calculate the flight altitude range, ensuring the drone's flight distance and altitude remain within this range. The flight altitude calculation formula is as follows: Flight altitude = (focal length * resolution index) / (sensor width / pixel width).

[0045] It should be noted that by controlling the flight altitude of the drone during scanning, the vegetation in the photos it takes is clearly visible, which helps to identify the type of vegetation covering the ground using image recognition algorithms.

[0046] A computer device includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps in the above method.

[0047] A computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method.

[0048] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions will not cause the essence of the corresponding technical solutions to deviate from the protection scope of the technical solutions of the embodiments of the present invention.

Claims

1. A smoke sensor fault monitoring system based on forest fire monitoring, characterized in that, include: Distributed monitoring units are used to construct a fire monitoring network, which includes a cloud monitoring center, multiple front-end monitoring nodes, and multiple data processing platforms. The front-end monitoring nodes include smoke sensors, temperature sensors, and humidity sensors. Based on the number of data processing platforms, the fire monitoring area is divided into multiple unit monitoring areas, and multiple front-end monitoring nodes within a unit monitoring area are controlled by the data processing platform. The infrared monitoring unit monitors the thermal imaging of the fire monitoring area and collects thermal imaging images of the monitoring areas of each unit within the fire monitoring area, which are recorded as unit infrared images. The fault detection unit marks abnormal monitoring points based on monitoring data uploaded by multiple front-end monitoring nodes. The uploading of data by the front-end monitoring nodes is divided into periodic uploading and conditional uploading. The anomaly analysis unit determines the corresponding real-time impact range based on the location, wind direction, wind speed, and preset sensing range of the anomaly monitoring point and records it as the monitoring area of ​​the anomaly monitoring point. It extracts the infrared image corresponding to the monitoring area and records it as the target infrared image. Based on the abnormal temperature value in the target infrared image, it checks whether the sensor is faulty.

2. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 1, characterized in that, Set a smoke index threshold and a smoke change threshold for the smoke sensor; set a temperature index threshold and a temperature change threshold for the temperature sensor; set a humidity index threshold and a humidity change threshold for the humidity sensor. The conditions for uploading include: Condition 1: The smoke index detected by the smoke sensor is greater than the smoke index threshold or the change in the smoke index per unit time is greater than the smoke change threshold. Condition 2: The temperature index captured by the temperature sensor is greater than the temperature index threshold or the change in temperature index per unit time is greater than the temperature change threshold; Condition 3: The humidity index captured by the humidity sensor is greater than the humidity index threshold or the change in humidity index per unit time is greater than the humidity change threshold. When any of the above conditions are met, conditional upload will be performed.

3. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 2, characterized in that, When the uploaded data comes from conditional upload, the corresponding monitoring point is marked as an abnormal monitoring point. When the uploaded data comes from periodic upload, the average value of each monitoring index is calculated. When the difference between any monitoring index of any monitoring point and the corresponding average value is greater than the preset deviation threshold, it is recorded as an abnormal monitoring point.

4. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 1, characterized in that, The process for determining the real-time impact range is as follows: A digital elevation model (DEM) is constructed within the fire monitoring area. Vegetation cover within the fire monitoring area is identified and combined with the DEM to obtain a vegetation cover model. The fire monitoring area is divided into multiple coverage areas with different vegetation cover types, including shrub cover, forest cover, and grassland cover. An initial range is drawn with the abnormal monitoring point as the center and the preset sensing range as the radius. Multiple points evenly distributed on the circumference of the initial range are recorded as initial boundary points. Based on the vegetation cover between the abnormal monitoring point and the initial boundary point, combined with wind speed and wind direction, the positions of multiple initial boundary points are corrected to obtain discrete boundary points. Multiple discrete boundary points are connected sequentially to obtain the real-time influence range.

5. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 4, characterized in that, The initial boundary point position correction process is as follows: The connecting line segment between the initial boundary point and the anomaly monitoring point is called the initial line segment. The straight line where the initial line segment is located is called the reference straight line. A rectangular coordinate system is constructed with the anomaly monitoring point as the origin and called the reference coordinate system. The vertical axis of the reference coordinate system extends along the wind direction. A base combustion rate is assigned to each vegetation cover type. The initial line segment is divided into multiple unit initial line segments in different vegetation cover areas. Based on the base combustion rate and combined with wind direction and wind speed, the conduction speed corresponding to the unit initial line segment is analyzed. The conduction distance is divided by the conduction speed to obtain the conduction time of the unit initial line segment. The time difference between the current time and the last upload time of the anomaly monitoring point is recorded as the blind zone time. The sum of multiple transmission times corresponding to the initial line segment is recorded as the influence duration and compared with the blind zone time. When the blind zone time is less than or equal to the influence duration, a corresponding discrete boundary point is determined between the initial boundary point and the anomaly monitoring point. When the blind zone time is greater than the influence duration, a corresponding discrete boundary point is determined on the side of the initial boundary point away from the anomaly monitoring point. The discrete boundary point satisfies the following condition: Condition A: It lies on the initial line segment; Condition B: The influence time corresponding to the discrete boundary point is equal to the blind zone time.

6. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 5, characterized in that, The sensing range is equal to the maximum value of the blind zone time multiplied by the base combustion rate.

7. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 5, characterized in that, The conduction velocity analysis process is as follows: The conduction velocity is equal to the vector sum of the wind direction velocity and the natural conduction velocity, with the wind direction velocity along the vertical axis of the reference coordinate system and the natural conduction velocity along the horizontal axis of the reference coordinate system. Wind speed is equal to the natural conduction speed multiplied by the wind force influence coefficient. The natural conduction speed is equal to the base combustion rate of the corresponding vegetation cover route. When the fire spreads along the initial line segment to the abnormal monitoring point in a headwind manner, the wind force influence coefficient is equal to 1 minus the wind speed influence coefficient. When the fire spreads along the initial line segment to the abnormal monitoring point in a tailwind manner, the wind force influence coefficient is equal to 1 plus the wind speed influence coefficient. The wind speed influence coefficient is equal to the wind speed multiplied by a preset proportional coefficient.

8. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 1, characterized in that, The vegetation cover identification process is as follows: A drone equipped with a camera performs a comprehensive scan of the fire monitoring area, collecting aerial images of the area. These images are then divided into multiple unit identification regions. Based on image recognition algorithms, the coverage percentage of each vegetation type is determined. The vegetation type with the largest coverage percentage within a unit identification region is used as the labeled coverage type for that region. Adjacent unit identification regions with the same labeled coverage type are merged to obtain multiple shrub coverage regions, forest coverage regions, and grassland coverage regions.

9. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 8, characterized in that, The drone flight altitude control process is as follows: The camera parameters of the drone are obtained, including sensor width, pixel width and lens focal length. A sharpness index range and a flight altitude calculation formula are preset to calculate the flight altitude range, so that the drone's altitude during flight is within the flight altitude range.

10. The smoke sensor fault monitoring system based on forest fire monitoring according to claim 9, characterized in that, The formula for calculating flight altitude is: Flight altitude = (focal length * resolution index) / (sensor width / pixel width).

Images