Power transmission line forest fire real-time monitoring system and method based on artificial intelligence

Abstract

The invention relates to a power transmission line forest fire real-time monitoring system based on artificial intelligence. The system comprises a sensor module; the data processing module is used for receiving and preprocessing the environmental data acquired by the sensor module; the artificial intelligence analysis system analyzes the multi-source data collected by the sensor module by using a deep learning algorithm, and identifies the forest fire risk around the power transmission line; and the early warning response system is used for sending out an early warning signal when the forest fire risk is detected, and taking emergency response measures according to the fire risk level. The invention further discloses a power transmission line forest fire real-time monitoring method of the power transmission line forest fire real-time monitoring system based on artificial intelligence. According to the invention, high-precision forest fire detection is realized under different environmental conditions, the false alarm rate is reduced, an early warning signal can be sent in an extremely short time, and safe operation of a power transmission line is ensured; the forest fire risk is automatically judged, the dependence on manual inspection is reduced, and the automation level of power transmission line monitoring is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power system safety monitoring and disaster prevention, and in particular to a real-time monitoring system and method for wildfires on power transmission lines based on artificial intelligence. Background Art

[0002] In recent years, global climate change has triggered increasingly frequent and severe natural disasters, among which wildfires are particularly prominent. Transmission lines usually cross large areas of forests and mountains and are extremely vulnerable to wildfires. Wildfires may not only damage transmission lines, but also cause large-scale power outages, endangering the stability and security of the power grid. Therefore, it is particularly important to develop a system that can monitor the environment around transmission lines in real time, issue early warnings, and effectively respond to wildfires.

[0003] Traditional transmission line monitoring methods mostly rely on manual inspections, which makes it difficult to detect potential fire hazards in a timely manner, and the monitoring range is limited. With the development of technologies such as drones, sensors, and image processing, automated and intelligent monitoring systems have gradually become the research focus of transmission line safety assurance. However, existing automatic monitoring systems still have problems such as low accuracy and slow response speed. Especially in the scenario of wildfire monitoring, these defects may lead to the failure to detect and deal with disasters in a timely manner. Summary of the invention

[0004] In order to solve the problems of slow response speed, low accuracy and limited coverage in the existing technology in fire monitoring, the primary purpose of the present invention is to provide a real-time fire monitoring and early warning system for power transmission line surroundings through sensor acquisition, image processing, artificial intelligence algorithm analysis, etc., and with high accuracy and response speed.

[0005] To achieve the above objectives, the present invention adopts the following technical solutions: A real-time monitoring system for wildfires on power transmission lines based on artificial intelligence, comprising:

[0006] A sensor module, used for collecting environmental data around the transmission line in real time, the sensor module includes a plurality of sensor nodes arranged around the transmission line;

[0007] A data processing module receives and pre-processes the environmental data collected by the sensor module;

[0008] An artificial intelligence analysis system, which uses a deep learning algorithm to analyze multi-source data collected by the sensor module to identify wildfire risks around the transmission line, wherein the artificial intelligence analysis system uses a deep learning model;

[0009] The early warning response system sends out early warning signals when wildfire risks are detected and takes emergency response measures according to the fire risk level.

[0010] The sensor module comprises:

[0011] Temperature sensor, used to monitor changes in ambient temperature;

[0012] Humidity sensor to monitor air humidity levels;

[0013] Smoke sensors, which detect smoke particles in the air;

[0014] Wind speed and direction sensors to monitor wind strength and direction;

[0015] Infrared camera, used to obtain infrared images around the transmission line;

[0016] Visible light camera, used to obtain visible light images around the transmission lines.

[0017] The data processing module comprises:

[0018] A data preprocessing unit cleans, formats and filters the collected sensor data to obtain preprocessed data;

[0019] A feature extraction unit, used to extract key features and patterns from the preprocessed data;

[0020] The data fusion unit integrates and analyzes data from different sensors to generate more comprehensive and accurate monitoring information.

[0021] The early warning response system comprises:

[0022] An alarm system that automatically sounds an alarm and notifies relevant personnel and agencies when a fire risk is detected;

[0023] Information transmission system, used to transmit monitoring data and alarm information to the command center in real time;

[0024] An emergency response planning system for developing pre-set emergency response measures, including fire fighting, evacuation and resource dispatch;

[0025] Decision support system, which helps decision makers assess fire risks and develop response strategies through data analysis and model prediction;

[0026] Feedback system is used to monitor response effects and adjust early warning and response strategies according to actual conditions.

[0027] Another object of the present invention is to provide a method for real-time monitoring of wildfires in power transmission lines based on an artificial intelligence-based real-time monitoring system for wildfires in power transmission lines, the method comprising the following steps in order:

[0028] (1) Multiple sensor nodes are deployed around the transmission lines to collect environmental data in real time, including temperature, humidity, smoke concentration, wind speed and direction, infrared images, and visible light images;

[0029] (2) The collected environmental data is transmitted to the data processing module for preprocessing, noise is removed and key features are extracted to obtain preprocessed data;

[0030] (3) Analyze the preprocessed data through an artificial intelligence analysis system and use deep learning models to identify potential wildfire risks;

[0031] (4) When a wildfire risk is detected, the early warning response system sends out a warning signal and takes emergency response measures based on the risk level.

[0032] In step (3), the deep learning model adopts a lightweight YOLO-LNet model. The total loss function of the lightweight YOLO-LNet model consists of three parts, namely, the regression loss of the bounding box, the confidence loss, and the category loss. Among them, the regression loss adopts the NWD metric design:

[0033] L NWD =1-NWD(N p ,N g )

[0034] Among them, N p is the Gaussian distribution model of the prediction box p, N g is the Gaussian distribution model of the real frame g;

[0035]

[0036] Where C is the number of images in the dataset; cx p is the horizontal coordinate of the upper left corner of the prediction box; cy p is the vertical coordinate of the upper left corner of the prediction box; w p is the width of the prediction box; h p is the height of the prediction box; cx g is the horizontal coordinate of the upper left corner of the real box; cy g is the vertical coordinate of the upper left corner of the real frame; w g is the width of the real frame; h g is the height of the real frame;

[0037] The formula for confidence loss is:

[0038]

[0039] in, represents the binary cross entropy loss, w objrepresents the weight of the positive sample, p0 is the target confidence score in the prediction box, and p NWD It is the NWD value between the predicted box and the true box;

[0040] The formula for class loss is:

[0041]

[0042] in, represents the binary cross entropy loss, w cls represents the weight of the positive sample, c p is the category score of the prediction box, c gt is the one-hot encoding score of the target box category;

[0043] The total loss function of the lightweight YOLO-LNet model is:

[0044]

[0045] Among them, K, S 2 , B are the output feature map, grid, and the number of anchor points on each grid respectively; Indicates whether the i-th grid, j-th anchor box, and k-th output feature map are positive samples. If they are positive samples, then The value of is 1, otherwise, The value of t is 0; p and t gt are the predicted vector and the true vector respectively; Used to balance the weights of output feature maps at different scales; ∝ box To balance the weights of anchor boxes at different scales, ∝ obj is the weight used to balance the confidence at different scales, ∝ cls is the weight used to balance the categories at different scales.

[0046] It can be seen from the above technical scheme that the beneficial effects of the present invention are: first, the present invention combines multi-sensor data fusion and deep learning models to achieve high-precision wildfire detection under different environmental conditions and reduce the false alarm rate; second, the present invention uses an artificial intelligence analysis system to perform real-time analysis of environmental data around the transmission line. Once a wildfire risk is detected, it can issue a warning signal in a very short time to ensure the safe operation of the transmission line; third, the present invention uses an artificial intelligence analysis system to independently judge the wildfire risk, reduces dependence on manual inspections, and improves the automation level of transmission line monitoring; fourth, the substation automatically generates a patrol task strategy based on the current scenario, which not only greatly improves the operation and maintenance efficiency and safety, but also strengthens the stability and emergency response capabilities of the power grid, laying a solid foundation for the sustainable development of the power industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 It is a system structure block diagram of the present invention. DETAILED DESCRIPTION

[0048] like Figure 1 As shown, a real-time monitoring system for wildfires on power transmission lines based on artificial intelligence includes:

[0049] A sensor module, used for collecting environmental data around the transmission line in real time, the sensor module includes a plurality of sensor nodes arranged around the transmission line;

[0050] A data processing module receives and pre-processes the environmental data collected by the sensor module;

[0051] An artificial intelligence analysis system, which uses a deep learning algorithm to analyze multi-source data collected by the sensor module to identify wildfire risks around the transmission line, wherein the artificial intelligence analysis system uses a deep learning model;

[0052] The early warning response system sends out early warning signals when wildfire risks are detected and takes emergency response measures according to the fire risk level.

[0053] The sensor module comprises:

[0054] Temperature sensor, used to monitor changes in ambient temperature;

[0055] Humidity sensor to monitor air humidity levels;

[0056] Smoke sensors, which detect smoke particles in the air;

[0057] Wind speed and direction sensors to monitor wind strength and direction;

[0058] Infrared camera, used to obtain infrared images around the transmission line;

[0059] Visible light camera, used to obtain visible light images around the transmission lines.

[0060] The data processing module comprises:

[0061] A data preprocessing unit cleans, formats and filters the collected sensor data to obtain preprocessed data;

[0062] A feature extraction unit, used to extract key features and patterns from the preprocessed data;

[0063] The data fusion unit integrates and analyzes data from different sensors to generate more comprehensive and accurate monitoring information.

[0064] The early warning response system comprises:

[0065] An alarm system that automatically sounds an alarm and notifies relevant personnel and agencies when a fire risk is detected;

[0066] Information transmission system, used to transmit monitoring data and alarm information to the command center in real time;

[0067] An emergency response planning system for developing pre-set emergency response measures, including fire fighting, evacuation and resource dispatch;

[0068] Decision support system, which helps decision makers assess fire risks and develop response strategies through data analysis and model prediction;

[0069] Feedback system is used to monitor response effects and adjust early warning and response strategies according to actual conditions.

[0070] Another object of the present invention is to provide a method for real-time monitoring of wildfires in power transmission lines based on an artificial intelligence-based real-time monitoring system for wildfires in power transmission lines, the method comprising the following steps in order:

[0071] (1) Multiple sensor nodes are arranged around the transmission lines to collect environmental data in real time, including temperature, humidity, smoke concentration, wind speed and direction, infrared images, and visible light images. Sensor modules are arranged at key nodes of the transmission lines to collect environmental data in real time and transmit them to the data processing center.

[0072] (2) The collected environmental data is transmitted to the data processing module for preprocessing, noise is removed and key features are extracted to obtain preprocessed data; outliers and noise are removed to ensure the accuracy of subsequent data analysis. The system uses multi-sensor data fusion technology to conduct a comprehensive analysis of data from different sensors. For example, when the temperature sensor detects an abnormal increase in temperature and the smoke sensor detects an increase in smoke concentration, the system will determine that the fire risk has increased and thus increase the warning level.

[0073] (3) The pre-processed data is analyzed through an artificial intelligence analysis system, and potential wildfire risks are identified using a deep learning model. After the model is deployed, the system can identify early signs of wildfires around transmission lines, such as the appearance of flames and smoke, and can even detect potential environmental anomalies before a fire, such as high temperature and low humidity.

[0074] (4) When a wildfire risk is detected, the early warning response system sends out an early warning signal and takes emergency response measures according to the risk level, such as shutting down the transmission line or dispatching personnel for on-site processing. When the system detects a wildfire risk, the early warning response system will take different response measures according to the severity of the fire. In the case of a low-level early warning, the system will notify maintenance personnel to go to the site for inspection; when the fire risk is high, the system will automatically activate the emergency plan and shut down the relevant lines to prevent the fire from spreading and causing greater damage to the power grid.

[0075] In step (3), the deep learning model adopts a lightweight YOLO-LNet model. The total loss function of the lightweight YOLO-LNet model consists of three parts, namely, the regression loss of the bounding box, the confidence loss, and the category loss. Among them, the regression loss adopts the NWD metric design:

[0076] L NWD =1-NWD(N p ,N g )

[0077] Among them, N p is the Gaussian distribution model of the prediction box p, N g is the Gaussian distribution model of the real frame g;

[0078]

[0079]

[0080] Where C is the number of images in the dataset; cx p is the horizontal coordinate of the upper left corner of the prediction box; cy p is the vertical coordinate of the upper left corner of the prediction box; w p is the width of the prediction box; h p is the height of the prediction box; cx g is the horizontal coordinate of the upper left corner of the real box; cy g is the vertical coordinate of the upper left corner of the real box; w g is the width of the real frame; h g is the height of the real frame;

[0081] The formula for confidence loss is:

[0082]

[0083] in, represents the binary cross entropy loss, w obj represents the weight of the positive sample, p0 is the target confidence score in the prediction box, and p NWD It is the NWD value between the predicted box and the true box;

[0084] The formula for class loss is:

[0085]

[0086] in, represents the binary cross entropy loss, w cls represents the weight of the positive sample, c p is the category score of the predicted box, c gt is the one-hot encoding score of the target box category;

[0087] The total loss function of the lightweight YOLO-LNet model is:

[0088]

[0089] Among them, K, S 2 , B are the output feature map, grid, and the number of anchor points on each grid respectively; Indicates whether the i-th grid, j-th anchor box, and k-th output feature map are positive samples. If they are positive samples, then The value of is 1, otherwise, The value of t is 0; p and t gt are the predicted vector and the true vector respectively; Used to balance the weights of output feature maps at different scales; ∝ box To balance the weights of anchor boxes at different scales, ∝ obj is the weight used to balance the confidence at different scales, ∝ cls is the weight used to balance the categories at different scales.

[0090] In summary, the present invention provides a real-time monitoring system and method for wildfires on power transmission lines based on artificial intelligence, which can achieve efficient and accurate wildfire identification and early warning in a variety of complex environments. By integrating multi-sensor data and deep learning technology, the system greatly improves the intelligent level of power transmission line safety management and reduces the impact of fire hazards on the power system.

[0091] The specific implementation methods described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific implementation method of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A real-time monitoring system for forest fires on power transmission lines based on artificial intelligence, characterized by: include: A sensor module, used for collecting environmental data around the transmission line in real time, the sensor module includes a plurality of sensor nodes arranged around the transmission line; A data processing module receives and pre-processes the environmental data collected by the sensor module; An artificial intelligence analysis system, which uses a deep learning algorithm to analyze multi-source data collected by the sensor module to identify wildfire risks around the transmission line, wherein the artificial intelligence analysis system uses a deep learning model; The early warning response system sends out early warning signals when wildfire risks are detected and takes emergency response measures according to the fire risk level.

2. The artificial intelligence-based real-time monitoring system for power transmission line wildfires according to claim 1 is characterized in that: The sensor module comprises: Temperature sensor, used to monitor changes in ambient temperature; Humidity sensor to monitor air humidity levels; Smoke sensors, which detect smoke particles in the air; Wind speed and direction sensors to monitor wind strength and direction; Infrared camera, used to obtain infrared images around the transmission line; Visible light camera, used to obtain visible light images around the transmission lines.

3. The power transmission line wildfire real-time monitoring system based on artificial intelligence according to claim 1 is characterized by: The data processing module comprises: A data preprocessing unit cleans, formats and filters the collected sensor data to obtain preprocessed data; A feature extraction unit, used to extract key features and patterns from the preprocessed data; The data fusion unit integrates and analyzes data from different sensors to generate more comprehensive and accurate monitoring information.

4. The artificial intelligence-based real-time monitoring system for power transmission line wildfires according to claim 1 is characterized in that: The early warning response system comprises: An alarm system that automatically sounds an alarm and notifies relevant personnel and agencies when a fire risk is detected; Information transmission system, used to transmit monitoring data and alarm information to the command center in real time; An emergency response planning system for developing pre-set emergency response measures, including fire fighting, evacuation and resource dispatch; Decision support system, which helps decision makers assess fire risks and develop response strategies through data analysis and model prediction; Feedback system is used to monitor response effects and adjust early warning and response strategies according to actual conditions.

5. The method for real-time monitoring of wildfires in power transmission lines according to any one of claims 1 to 4, characterized in that: The method comprises the following steps in order: (1) Multiple sensor nodes are deployed around the transmission lines to collect environmental data in real time, including temperature, humidity, smoke concentration, wind speed and direction, infrared images, and visible light images; (2) The collected environmental data is transmitted to the data processing module for preprocessing, noise is removed and key features are extracted to obtain preprocessed data; (3) Analyze the preprocessed data through an artificial intelligence analysis system and use deep learning models to identify potential wildfire risks; (4) When a wildfire risk is detected, the early warning response system sends out a warning signal and takes emergency response measures based on the risk level.

6. The method for real-time monitoring of forest fires on power transmission lines according to claim 5, characterized in that: In step (3), the deep learning model adopts a lightweight YOLO-LNet model. The total loss function of the lightweight YOLO-LNet model consists of three parts, namely, the regression loss of the bounding box, the confidence loss, and the category loss. Among them, the regression loss adopts the NWD metric design: L NWD =1-NWD(N p ,OF g ) Among them, N p is the Gaussian distribution model of the prediction box p, N g is the Gaussian distribution model of the real frame g; Where C is the number of images in the dataset; cx p is the horizontal coordinate of the upper left corner of the prediction box; cy p is the vertical coordinate of the upper left corner of the prediction box; w p is the width of the prediction box; h p is the height of the prediction box; cx g is the horizontal coordinate of the upper left corner of the real box; cy g is the vertical coordinate of the upper left corner of the real box; w g is the width of the real frame; h g is the height of the real frame; The formula for confidence loss is: in, represents the binary cross entropy loss, w obj represents the weight of the positive sample, p0 is the target confidence score in the prediction box, and p NWD It is the NWD value between the predicted box and the true box; The formula for class loss is: in, represents the binary cross entropy loss, w cls represents the weight of the positive sample, c p is the category score of the predicted box, c gt is the one-hot encoding score of the target box category; The total loss function of the lightweight YOLO-LNet model is: Among them, K, S 2 , B are the output feature map, grid, and the number of anchor points on each grid respectively; Indicates whether the i-th grid, j-th anchor box, and k-th output feature map are positive samples. If they are positive samples, then The value of is 1, otherwise, The value of t is 0; p and t gt are the predicted vector and the true vector respectively; Used to balance the weights of output feature maps at different scales; ∝ box To balance the weights of anchor boxes at different scales, ∝ obj is the weight used to balance the confidence at different scales, ∝ cls is the weight used to balance the categories at different scales.

Images