Mountain fire identification system and method based on unmanned aerial vehicle inspection and related device

Abstract

The embodiment of the invention discloses a forest fire recognition system and method based on unmanned aerial vehicle inspection and a related device. The forest fire recognition system comprises an unmanned aerial vehicle data acquisition system, a deep learning analysis system and a terminal visualization system. The unmanned aerial vehicle data acquisition system is used for acquiring image data of the power transmission line; the deep learning analysis system is used for performing forest fire identification on the image data by using a pre-trained deep learning model, and determining a forest fire identification result of the power transmission line, which is used for reflecting whether the power transmission line has a forest fire; and the terminal visualization system is used for generating a visualization prompt based on the forest fire identification result. Through the forest fire identification system combining the deep learning algorithm and the unmanned aerial vehicle data acquisition, potential dangerous conditions such as forest fire can be automatically identified, and an alarm is given in time, so that the safety of a power transmission line and the efficiency of inspection work are remarkably improved; and an efficient and intelligent forest fire automatic identification and alarm system is provided for a power transmission line.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned aerial vehicle inspection, and in particular to a wildfire identification system, method and related devices based on unmanned aerial vehicle inspection. Background Art

[0002] Wildfires have always been one of the main causes of high-voltage transmission line tripping. When a wildfire occurs within the transmission line corridor, the success rate of the protection device's reclosing is very low. There are many reasons for this. First, the reclosing time is generally in seconds or even milliseconds, but the duration of a wildfire is generally several minutes or even hours, which is far beyond the time range of the reclosing. Therefore, even if the reclosing is successful in a short period of time, the continued spread of the fire may cause another trip, affecting the stability of the power grid.

[0003] Secondly, the high temperature and thick smoke generated by wildfires can not only directly damage power transmission facilities, such as wires and towers, but may also cause the performance of insulation materials to deteriorate and increase the risk of short circuit tripping. In addition, wildfires will also change the conditions of the surrounding environment, such as air composition and humidity. These changes may affect the normal operation of electrical equipment and increase the failure rate. Furthermore, the occurrence of wildfires is often accompanied by extreme weather conditions, such as drought and strong winds, which are themselves not conducive to the safe operation of power grid facilities.

[0004] In the current inspection of high-voltage transmission lines, photos of transmission lines are usually obtained by manual inspection, fixed camera monitoring, and drone photography, and then the presence of wildfires that endanger the safe and stable operation of transmission lines is judged by human visual recognition. However, the manual inspection method has a large workload, and staff are required to take photos and record the situation in real time. Fixed cameras require large investments and have high reliability requirements. Existing drone inspection systems mostly rely on manual analysis of video data, which is inefficient and easily affected by subjective factors. Wildfire detection of high-voltage transmission lines is an important means of sensing information on on-site transmission lines. It not only provides a basis for adjusting the operation mode of the power grid, formulating emergency plans and measures, and evaluating the operation and maintenance status of the lines, but also provides support for relevant government departments to initiate firefighting decisions.

[0005] Currently, photo information of power transmission lines is obtained through manual inspections, fixed camera monitoring, and drone photography, and reported through telephone and WeChat groups. At the same time, data processing personnel confirm the location of wildfire monitoring through the patrol locations of wildfire observers, the locations of fixed cameras, and the shooting directions of drones. Data processing personnel combine the acquired transmission line photo information with the detection location to form an Excel summary table of wildfire observations, and then manually enter it into the wildfire management platform of the power grid. Professionals then review the uploaded information to determine whether there is a wildfire and whether the wildfire will affect the safe and stable operation of the transmission line, and then make decisions. The above traditional wildfire monitoring methods have the following defects:

[0006] 1) Traditional inspection methods cannot achieve full-time, all-round monitoring, and there are monitoring blind spots. Traditional transmission line inspection methods, such as manual inspections or fixed camera monitoring, have significant limitations. Manual inspections are not only time-consuming and labor-intensive, but also often fail to cover all key areas due to geographical and environmental conditions, especially those remote or difficult-to-reach places. Although fixed cameras can provide real-time monitoring, their field of view is limited by the installation location and the viewing angle cannot be adjusted, which easily forms monitoring blind spots. In addition, fixed cameras have insufficient tracking capabilities for moving targets. Once an emergency such as a wildfire occurs, it is difficult to capture the dynamic changes of the fire source in real time. For transmission lines across a wide area, full-time, all-round monitoring is the key to ensuring their safe operation. Traditional inspection methods cannot meet this demand, so there are major safety hazards.

[0007] 2) Manual analysis of video data is time-consuming and labor-intensive, and it is difficult to ensure accuracy and consistency. In traditional monitoring systems, analyzing video data usually relies on manual operations. This process not only consumes a lot of time and human resources, but also due to the interference of human factors, the accuracy and consistency of the results are difficult to guarantee. Human fatigue, distraction or subjective judgment may affect the quality of the analysis. Especially in emergency situations such as wildfires, quickly and accurately identifying the location of the fire and the development of the fire is crucial for timely response. Manual analysis methods are often unable to process large amounts of video data in a short period of time and extract key information from it, which may cause the fire to spread and cause greater losses.

[0008] 3) Existing drone inspection systems lack automated wildfire identification capabilities and cannot respond to wildfire threats in a timely manner. Although drone inspection systems have improved inspection efficiency and coverage to a certain extent, most existing drone systems rely on manual remote control or post-analysis of data. These systems lack automated wildfire identification capabilities, that is, the ability to detect, identify and report wildfire incidents in real time during flight. Wildfires are characterized by suddenness and rapid spread, requiring inspection systems to respond quickly. Manual control of drones or post-analysis of data cannot meet the needs of quickly identifying wildfires and responding in a timely manner. Therefore, drone inspection systems that lack automated identification capabilities have obvious deficiencies in dealing with emergencies such as wildfires.

[0009] 4) The speed of data processing and analysis is not fast enough to meet the needs of real-time monitoring. Data is the core of modern monitoring systems, especially when dealing with emergencies such as wildfires, the speed of data processing and analysis is crucial. However, existing drone inspection systems often have bottlenecks in data processing and analysis. Due to inefficient algorithms or limited computing resources, these systems are unable to process and analyze data immediately after collection, resulting in delayed responses. Real-time monitoring requires the system to be able to process large amounts of data quickly, identify abnormal situations in a timely manner, and respond accordingly. If the speed of data processing and analysis cannot keep up with the speed of data collection, the system will not be able to achieve true real-time monitoring, thus missing the best time to prevent and control wildfires.

[0010] It can be seen that the traditional wildfire monitoring method is inefficient and the accuracy of wildfire identification is also low. There is an urgent need for a wildfire monitoring method that can improve efficiency and accuracy. Summary of the invention

[0011] The main purpose of the present invention is to provide a wildfire identification system, method and related devices based on drone inspection, which can solve the problems of low efficiency of wildfire monitoring methods in the prior art and low wildfire identification accuracy.

[0012] To achieve the above-mentioned object, the first aspect of the present invention provides a wildfire identification system based on drone inspection, the wildfire identification system comprising: a drone data acquisition system, a deep learning analysis system and a terminal visualization system, wherein the drone data acquisition system, the deep learning analysis system and the terminal visualization system are connected in communication;

[0013] The drone data acquisition system is used to collect image data of the power transmission line and send the image data to the deep learning analysis system;

[0014] The deep learning analysis system is used to receive the image data, use a pre-trained deep learning model to perform wildfire identification on the image data, determine a wildfire identification result of the transmission line, and send the wildfire identification result to the terminal visualization system, the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line;

[0015] The terminal visualization system is used to receive the wildfire identification result and generate a visualization prompt based on the wildfire identification result.

[0016] In a feasible implementation, the drone data acquisition system includes at least: a high-voltage transmission line ledger module, a sensor camera control module, a data acquisition storage unit module, and a data communication upload module, and there is a communication connection between the high-voltage transmission line ledger module, the sensor camera control module, the data acquisition storage unit module, and the data communication upload module;

[0017] The high-voltage transmission line ledger module is used to store the ledger information of the high-voltage transmission line and send the ledger information to the sensor camera control module;

[0018] The sensor camera control module is used to receive the ledger information, perform data acquisition planning processing based on the ledger information, and obtain data acquisition instructions; perform image acquisition processing on the transmission line based on the data acquisition instructions to obtain the image data, and send the image data to the data acquisition storage unit module;

[0019] The data acquisition storage unit module is used to receive and store the image data, and send the image data to the data communication upload module;

[0020] The data communication upload module is used to receive the image data and upload the image data to the deep learning analysis system.

[0021] In a feasible implementation, the drone data acquisition system further includes a data analysis and processing module, the input end of the data analysis and processing module is communicatively connected to the output end of the data acquisition and storage unit module, and the output end of the data acquisition and storage unit module is communicatively connected to the input end of the data communication upload module;

[0022] The data acquisition storage unit module is also used to send the image data to the data analysis and processing module;

[0023] The data analysis and processing module is used to receive the image data, perform denoising on the image data, obtain denoised image data, and send the denoised image data to the data communication upload module;

[0024] The data communication upload module is used to receive the denoised image data and upload the denoised image data to the deep learning analysis system.

[0025] In a feasible implementation, the deep learning analysis system at least includes: a data preprocessing module, a data annotation processing module, an image cutting processing module, a neural network feature extraction module, a model training module, a regression prediction module and an output analysis result module, wherein the data preprocessing module, the data annotation processing module, the image cutting processing module, the neural network feature extraction module, the model training module, the regression prediction module and the output analysis result module are connected in communication;

[0026] The data preprocessing module is used to receive the historical wildfire image data and image data, preprocess the historical wildfire image data, and obtain the preprocessed historical wildfire image data;

[0027] The data labeling processing module is used to label the preprocessed historical wildfire image data with training labels to obtain the historical wildfire image data labeled with training labels;

[0028] The image cutting processing module is used to perform image cutting on the historical wildfire image data marked with training labels using a preset image segmentation rule to obtain a plurality of image blocks;

[0029] The neural network feature extraction module is used to extract features from the image block using a neural network model to obtain image features of the image block;

[0030] The model training module is used to perform wildfire recognition training on the deep learning model using image features and training labels to obtain a trained deep learning model;

[0031] The regression prediction module is used to perform regression prediction using the trained deep learning model and the image data to determine the wildfire recognition result of the transmission line;

[0032] The output analysis result module is used to send the wildfire identification result to the terminal visualization system.

[0033] In a feasible implementation, the terminal visualization system at least includes: a short message alarm push module and a visualization display module, and the short message alarm push module and the visualization display module are connected in communication;

[0034] The SMS alarm push module is used to receive the wildfire identification result, and output SMS alarm information to the visual display module according to the wildfire identification result and the current operating status of the transmission line;

[0035] The visual display module is used to receive the SMS warning information and output a visual prompt based on the SMS warning information.

[0036] In a feasible implementation, the terminal visualization system further includes a decision-making auxiliary display module and a feedback closed-loop module, and the decision-making auxiliary display module and the feedback closed-loop module are communicatively connected with each other;

[0037] The auxiliary decision display module is used to receive the manual auxiliary decision result and send the auxiliary decision result to the feedback closed-loop module, wherein the auxiliary decision result is used to indicate the confirmation result of the manual auxiliary decision on the accuracy of the wildfire identification result;

[0038] The feedback closed-loop module is used to receive the auxiliary decision result, and use the auxiliary decision result as feedback data to optimize the trained deep learning model to obtain an optimized deep learning model.

[0039] To achieve the above-mentioned object, the second aspect of the present invention provides a method for identifying wildfires based on drone inspection, which is applied to the wildfire identification system based on drone inspection as described in the first aspect and any feasible implementation manner, and the method comprises:

[0040] Collect image data of power transmission lines;

[0041] Using a pre-trained deep learning model to perform wildfire recognition on the image data, and determining a wildfire recognition result of the transmission line, wherein the wildfire recognition result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line;

[0042] A visual prompt is generated based on the wildfire identification result.

[0043] To achieve the above-mentioned object, the third aspect of the present invention provides a wildfire identification device based on drone inspection, which is applied to the wildfire identification system based on drone inspection as described in the first aspect and any feasible implementation method, and the device includes:

[0044] Data acquisition module: used to collect image data of transmission lines;

[0045] A wildfire identification module: used to identify wildfires on the image data using a pre-trained deep learning model, and determine a wildfire identification result of the transmission line, wherein the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line;

[0046] Prompt generation module: used to generate visual prompts based on the wildfire identification results.

[0047] To achieve the above-mentioned purpose, the fourth aspect of the present invention provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, the processor executes the steps shown in the wildfire identification method based on drone inspection as described in the second aspect.

[0048] To achieve the above-mentioned purpose, the fifth aspect of the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps shown in the wildfire identification method based on drone inspection as described in the second aspect.

[0049] The embodiments of the present invention have the following beneficial effects:

[0050] The present invention provides a mountain fire identification system based on drone inspection, and the mountain fire identification system includes: a drone data acquisition system, a deep learning analysis system and a terminal visualization system, and there is a communication connection between the drone data acquisition system, the deep learning analysis system and the terminal visualization system; the drone data acquisition system is used to collect image data of the transmission line and send the image data to the deep learning analysis system; the deep learning analysis system is used to receive the image data, use the pre-trained deep learning model to perform mountain fire identification on the image data, determine the mountain fire identification result of the transmission line, and send the mountain fire identification result to the terminal visualization system, the mountain fire identification result is used to reflect whether there is a mountain fire on the transmission line, and the deep learning model is pre-trained based on the historical mountain fire image data of the transmission line; the terminal visualization system is used to receive the mountain fire identification result and generate a visual prompt based on the mountain fire identification result. Through the above-mentioned mountain fire identification system, an advanced drone inspection system can be provided, which combines a deep learning algorithm to achieve real-time monitoring of the transmission line and its surrounding environment. The advantage of this combined technology is that it can automatically identify potential dangerous situations such as wildfires and issue alarms in a timely manner, thereby significantly improving the safety of transmission lines and the efficiency of inspection work, and providing an efficient and intelligent wildfire automatic identification and alarm system for transmission lines. This system not only improves the safety management standards of transmission lines, but also provides strong technical support for the stable operation of power systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0052] in:

[0053] Figure 1 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 1 ;

[0054] Figure 2 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 2 ;

[0055] Figure 3 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 3 ;

[0056] Figure 4 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 4 ;

[0057] Figure 5 This is a flow chart of a method for identifying wildfires based on drone inspection in an embodiment of the present invention;

[0058] Figure 6 This is a structural block diagram of a wildfire identification device based on drone inspection in an embodiment of the present invention;

[0059] Figure 7 4 is a structural block diagram of a computer device in an embodiment of the present invention. DETAILED DESCRIPTION

[0060] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0061] The present invention is mainly composed of a background management system and a terminal visualization system. The background management system mainly includes a drone data acquisition system 10 and a deep learning analysis system 20. The drone data acquisition system 10 is equipped with high-definition cameras and sensors to inspect transmission lines and collect images and environmental data. The deep learning analysis system 20 is equipped with a deep learning server, which uses a pre-trained deep learning model to perform real-time analysis of image data. On this basis, a terminal visualization system 30 is established and deployed in line inspection and maintenance centers at the municipal level to visualize the analysis results of the deep learning analysis module 20 and assist in decision-making.

[0062] See also Figure 1 , Figure 1 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 1 ,like Figure 1 The wildfire identification system 00 shown includes: a drone data collection system 10, a deep learning analysis system 20 and a terminal visualization system 30, wherein there are multiple terminal visualization systems, which are deployed in various places, wherein the drone data collection system 10, the deep learning analysis system 20 and the terminal visualization system 30 are connected in communication;

[0063] Furthermore, the drone data acquisition system 10 is used to collect image data of the transmission line and send the image data to the deep learning analysis system 20. The image data is image data of the environment in which the transmission line is located. The image data is collected in real time, and the image data can be used for real-time monitoring to know whether there is a wildfire around the transmission line; the deep learning analysis system 20 is used to receive the image data, use a pre-trained deep learning model to identify the image data for wildfires, determine the wildfire identification result of the transmission line, and send the wildfire identification result to the terminal visualization system 30, wherein the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on the historical wildfire image data of the transmission line, wherein the wildfire identification result may include a first result that a wildfire exists and a second result that a wildfire does not exist. The results, as well as the corresponding image data, are convenient for subsequent intuitive display and prompting based on the image data for visualization, wherein the historical wildfire image data is the image data of the power transmission line collected when the wildfire occurred in historical time. The deep learning model is trained by various images of the wildfire, so that the deep learning model can learn the image features of the wildfire. In this way, the wildfire phenomenon can be analyzed and identified based on the real-time collected image data, thereby improving the efficiency and accuracy of wildfire identification; the terminal visualization system 30 is used to receive the wildfire identification result, and generate a visual prompt based on the wildfire identification result, wherein the wildfire identification result may include the first result or the second result, and the corresponding image data, and the first result and the image data or the second result and the image data are visualized so that relevant personnel can review the wildfire identification result more intuitively and promptly.

[0064] By introducing deep learning algorithms, the automatic identification and analysis of wildfire characteristics is realized, reducing the reliance on manual intervention. The real-time collection method of drones is highly real-time, and can detect wildfires and issue alarms in the first time, improving the speed of emergency response. The system integrates data analysis and detection functions, and can optimize inspection strategies based on historical data and environmental changes to improve prevention capabilities. The system structure is well designed and can adapt to different environments and meteorological conditions to ensure the continuity and stability of inspection work.

[0065] The present invention provides a mountain fire identification system based on drone inspection, and the mountain fire identification system includes: a drone data acquisition system, a deep learning analysis system and a terminal visualization system, and there is a communication connection between the drone data acquisition system, the deep learning analysis system and the terminal visualization system; the drone data acquisition system is used to collect image data of the transmission line and send the image data to the deep learning analysis system; the deep learning analysis system is used to receive the image data, use the pre-trained deep learning model to perform mountain fire identification on the image data, determine the mountain fire identification result of the transmission line, and send the mountain fire identification result to the terminal visualization system, the mountain fire identification result is used to reflect whether there is a mountain fire on the transmission line, and the deep learning model is pre-trained based on the historical mountain fire image data of the transmission line; the terminal visualization system is used to receive the mountain fire identification result and generate a visual prompt based on the mountain fire identification result. Through the above-mentioned mountain fire identification system, an advanced drone inspection system can be provided, which combines a deep learning algorithm to achieve real-time monitoring of the transmission line and its surrounding environment. The advantage of this combined technology is that it can automatically identify potential dangerous situations such as wildfires and issue alarms in a timely manner, thereby significantly improving the safety of transmission lines and the efficiency of inspection work, and providing an efficient and intelligent wildfire automatic identification and alarm system for transmission lines. This system not only improves the safety management standards of transmission lines, but also provides strong technical support for the stable operation of power systems.

[0066] See also Figure 2 , Figure 2 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 2 , Figure 2 The composition of the UAV data acquisition system is specifically demonstrated, including Figure 2 The drone data acquisition system 10 shown includes at least: a high-voltage transmission line ledger module 11, a sensor camera control module 12, a data acquisition storage unit module 13, a data analysis and processing module 14, and a data communication upload module 15, wherein the high-voltage transmission line ledger module 11, the sensor camera control module 12, the data acquisition storage unit module 13, the data analysis and processing module 14, and the data communication upload module 15 are connected in communication;

[0067] Furthermore, the high-voltage transmission line ledger module 11 is used to store the ledger information of the high-voltage transmission lines and send the ledger information to the sensor camera control module 12. Specifically, the high-voltage transmission line ledger module 11 contains the ledger information of the detailed archives of all high-voltage transmission lines, such as the direction of the lines, the location of the towers, the specifications of the wires, and related geographical and environmental information. This information is crucial for planning inspection routes, analyzing risk points, and formulating maintenance plans. The ledger module also supports data updating and maintenance to ensure that all information is up to date for effective monitoring and management.

[0068] The sensor camera control module 12 is used to receive ledger information, perform data collection planning and processing based on the ledger information, and obtain data collection instructions; perform image collection and processing on the transmission line based on the data collection instructions to obtain image data, and send the image data to the data collection storage unit module 13. Specifically, the sensor camera control module 12 is mainly used to control and coordinate various sensors and cameras carried on the UAV, and automatically collect photos of the high-voltage transmission lines that need to be inspected according to the ledger information of the high-voltage transmission line ledger module 11. This module 12 ensures that during the flight, data can be collected in the correct manner according to the predetermined plan or real-time instructions. It handles image stabilization, focus, zoom and switching between sensors to optimize data quality and collection efficiency.

[0069] The data acquisition storage unit module 13 is used to receive and store the image data, and send the image data to the data communication upload module 15. Specifically, the data acquisition storage unit module 13 is mainly used to receive raw data from sensors and cameras, and store it securely on the memory card of the drone or a similar storage device. The data acquisition storage unit module 13 has high-speed writing capability and large-capacity storage space to adapt to long-term flight missions and high-resolution video or image storage requirements, and also ensures data integrity to prevent data loss or damage before transmission.

[0070] The data communication upload module 15 is used to receive the image data and upload the image data to the deep learning analysis system 20. Specifically, the data communication upload module 15 mainly uploads the pre-processed transmission line photo information to the subsequent deep learning analysis system 20 to provide the original data for the system 20.

[0071] In a feasible implementation, since the photo information collected by the drone sensor and camera is affected by the weather, environment, etc., the photo data may have large background noise and low signal-to-noise ratio. Therefore, in order to improve the quality of the image data, the original image data may be subjected to some preliminary preprocessing operations such as denoising, and the image data may be sent to the deep learning analysis system 20 after denoising, that is, the drone data collection system 10 may also include a data analysis processing module 14, wherein the input end of the data analysis processing module 14 is in communication connection with the output end of the data collection storage unit module 13, and the output end of the data collection storage unit module 13 is in communication connection with the input end of the data communication upload module 15;

[0072] Furthermore, the data acquisition storage unit module 13 is also used to send the image data to the data analysis and processing module 14; the data analysis and processing module 14 is used to receive the image data, and perform denoising on the image data to obtain denoised image data, and send the denoised image data to the data communication upload module 15. It can be understood that since the photo information collected by the drone sensor and camera is affected by weather, environment, etc., the photo data will have large background noise and low signal-to-noise ratio. The module 14 mainly performs simple preprocessing on the original photo data, mainly including measures such as removing background noise and enhancing the signal-to-noise ratio to ensure the validity of the data subsequently entering the deep learning analysis system 20; and then the data communication upload module 15 is used to receive the denoised image data, and upload the denoised image data to the deep learning analysis system 20.

[0073] See also Figure 3 , Figure 3 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 3 , Figure 3 The composition of the deep learning analysis system 20 is specifically shown, such as Figure 3 The deep learning analysis system 20 shown includes at least: a data preprocessing module 21, a data annotation processing module 22, an image cutting processing module 23, a neural network feature extraction module 24, a model training module 25, a regression prediction module 26 and an output analysis result module 27, wherein the data preprocessing module 21, the data annotation processing module 22, the image cutting processing module 23, the neural network feature extraction module 24, the model training module 25, the regression prediction module 26 and the output analysis result module 27 are connected in communication;

[0074] Further, the data preprocessing module 21 is used to receive the historical wildfire image data and the image data, preprocess the historical wildfire image data, and obtain the preprocessed historical wildfire image data. Specifically, the preprocessing includes but is not limited to image classification, format standardization, cropping of photos, etc. to ensure that the photo sizes are consistent;

[0075] The data labeling processing module 22 is used to label the pre-processed historical wildfire image data with training labels, so as to obtain historical wildfire image data labeled with training labels, such as labeling the wildfire location, wildfire type, etc., to clarify the photo category;

[0076] The image cutting processing module 23 is used to perform image cutting on the historical wildfire image data marked with training labels using a preset image segmentation rule to obtain a plurality of image blocks, and the image segmentation rule is not limited to seamless cutting;

[0077] The neural network feature extraction module 24 is used to extract features from the image block using a neural network model to obtain image features of the image block;

[0078] The model training module 25 is used to perform wildfire recognition training on the deep learning model using image features and training labels to obtain a trained deep learning model;

[0079] The regression prediction module 26 is used to perform regression prediction using the trained deep learning model and the image data to determine the wildfire recognition result of the transmission line;

[0080] The analysis result output module 27 is used to send the wildfire identification result to the terminal visualization system 30 .

[0081] Among them, the main working processes of the deep learning analysis system 20 are as follows A01 to A08:

[0082] A01. Collect data from the selected high-voltage transmission line area through the above-mentioned UAV data collection system 10 to establish a massive wildfire detection database, the elements of which are transmission line photos;

[0083] A02, the data preprocessing module 21 performs further data preprocessing on the data samples in the database, mainly including image classification, format standardization, photo cropping, etc. to ensure that the photo sizes are consistent;

[0084] A03, data labeling processing module 22, creates labels for supervised learning, and manually labels the photo data to clarify the category of the photos.

[0085] A04. Since the collected and annotated photos are relatively large, the image cutting processing module 23 performs seamless cutting on the data set of the image sample library to obtain small image blocks of 320×320 in size. In order to achieve the purpose of seamless cutting, the present invention fills black pixels around the image during the cutting process so that its width and height are multiples of 320.

[0086] A05. After the cutting is completed, the obtained small image blocks are sent to the neural network model to complete feature extraction.

[0087] A06, the model training module 25 divides the massive photo data into a training set and a validation set according to a certain ratio, uses the training set to train the model, and uses the test set to verify the training results. If the accuracy requirement is not met, the number of training set samples is increased for training until the accuracy requirement is met, thereby obtaining a trained deep learning model.

[0088] A07. After the model is trained, the regression prediction module 26 will perform regression prediction on photos newly entered into the system (such as image data of power transmission lines collected in real time during actual application), quickly detect the wildfire area in the image, and mark the location of the wildfire in the image with a box to obtain the wildfire recognition result.

[0089] A08, output analysis result module 27 outputs the wildfire identification result to the terminal visualization system, and the wildfire photo information confirmed by the visualization system is added to the training set, and the model is continuously trained to improve the prediction accuracy.

[0090] See also Figure 4 , Figure 4 The structural framework of a wildfire identification system based on drone inspection in an embodiment of the present invention is shown in FIG. Figure 4 ; Figure 4 The structure of the terminal visualization system 30 is specifically demonstrated, wherein the terminal visualization system 30 is built on the basis of the drone data acquisition system 10 and the deep learning analysis system 20, and is displayed according to the output results of the deep learning analysis system 20, and is deployed in various prefecture-level power transmission patrol and maintenance centers. It mainly includes a short message alarm push module 31, a visualization display module 32, an auxiliary decision display module 33, and a feedback closed-loop module 34, wherein the short message alarm push module 31, the visualization display module 32, the auxiliary decision display module 33, and the feedback closed-loop module 34 are connected in communication;

[0091] Furthermore, the SMS alarm push module 31 is used to receive the wildfire identification result, and output SMS alarm information to the visual display module 32 according to the wildfire identification result and the current operating status of the transmission line. Specifically, the SMS alarm push module 31 sends abnormal alarm information to the person in charge of the corresponding municipal-level transmission patrol and maintenance center based on the detection result transmitted by the deep learning analysis system, that is, the wildfire identification result, combined with the transmission line operation status, voltage level, importance and other information, so that the person in charge can grasp the abnormal information of the transmission line in the first time;

[0092] The visualization display module 32 is used to receive the SMS alarm information and output a visualization prompt based on the SMS alarm information. Specifically, after receiving the alarm information, the person in charge quickly logs into the system, queries the wildfire photo annotation results corresponding to the alarm information through the visualization display module 32, and reviews them. If it is a misjudgment, click to close the system loop and end the task.

[0093] The auxiliary decision display module 33 is used to receive the manual auxiliary decision results and send the auxiliary decision results to the feedback closed-loop module 34. The auxiliary decision results are used to indicate the confirmation results of the manual auxiliary decision on the accuracy of the wildfire identification results. Specifically, the confirmation results can also include emergency plans. If the detection result system of the auxiliary decision module 33 is correct, the auxiliary decision display module considers the ledger information and operation mode of the power grid, and gives auxiliary decision information according to the consequences of the corresponding power accident events, including handling the wildfire hidden dangers within the specified time or shutting down the line in advance to ensure the safe and stable operation of the power grid.

[0094] The feedback closed-loop module 34 is used to receive the auxiliary decision results, and use the auxiliary decision results as feedback data to optimize the trained deep learning model to obtain an optimized deep learning model. Specifically, after processing the wildfire information, the operator clicks to complete the closed loop of the wildfire information, and at the same time evaluates the wildfire detection results and the auxiliary decision information. The evaluation results are fed back to the deep learning analysis module for system optimization.

[0095] The beneficial effects of the present invention include at least the following four points:

[0096] 1) The present invention collects real-time data through sensors and cameras carried by drones, and uses deep learning algorithms to analyze images and data, which can quickly and accurately identify wildfires. This method greatly improves the efficiency and accuracy of monitoring. Compared with traditional manual patrols or monitoring systems based on simple sensors, it can detect fires in a shorter time and reduce delays and misjudgments caused by human factors;

[0097] 2) The present invention combines deep learning algorithms to achieve real-time monitoring of the environment around the transmission line and automatically identify abnormal situations such as wildfires. Once signs of wildfires are detected, the system will automatically send an alarm to relevant departments or personnel, shortening the response time and providing a valuable time window for decision-making and action in emergency situations, so that fire extinguishing or isolation measures can be taken in advance to prevent the spread of the fire;

[0098] 3) The automated monitoring and identification of the present invention reduces the reliance on manual inspections, saves human resources, and reduces the cost of manual monitoring. At the same time, drone inspections can be carried out in bad weather or complex terrain, reducing the risk of personnel working in dangerous environments and improving the overall safety of work;

[0099] 4) The present invention can automatically adjust the inspection route and frequency according to environmental changes and historical data, optimize the inspection strategy, and ensure the effectiveness and coverage of monitoring. In addition, data query, statistics and analysis are carried out through the wildfire management platform, which facilitates scientific data management. All calculations are performed by software according to unified requirements, with high accuracy, and historical data query, statistics and management will not be affected by factors such as job changes.

[0100] See also Figure 5 , Figure 5 FIG. 1 is a flow chart of a method for identifying wildfires based on drone inspection in an embodiment of the present invention. Figure 5 The method is applied to Figures 1 to 4 The method of the mountain fire identification system based on drone inspection can be applied to both a terminal and a server. The terminal can be a desktop terminal or a mobile terminal. The mobile terminal can be at least one of a mobile phone, a tablet computer, a laptop computer, etc. The server can be implemented by an independent server or a server cluster composed of multiple servers. This embodiment is described by applying to a server as an example. The method includes the following steps:

[0101] 501. Collect image data of power transmission lines;

[0102] 502. Perform wildfire recognition on the image data using a pre-trained deep learning model to determine a wildfire recognition result of the transmission line, wherein the wildfire recognition result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line;

[0103] 503. Generate a visual prompt based on the wildfire identification result.

[0104] Understandably, Figure 5 The contents of each step in the method shown are Figure 1The functions of each system in the wildfire identification system based on drone inspection are similar. To avoid repetition, they are not described here. For details, please refer to Figure 1 The functions of each system in the wildfire identification system based on drone inspection are shown.

[0105] The present invention provides a method for identifying wildfires based on drone inspections, the method comprising: collecting image data of a transmission line; using a pre-trained deep learning model to identify wildfires on the image data, determining a wildfire identification result of the transmission line, the wildfire identification result being used to reflect whether there is a wildfire on the transmission line, the deep learning model being pre-trained based on historical wildfire image data of the transmission line; and generating a visual prompt based on the wildfire identification result. Through the above-mentioned method, a wildfire monitoring method combining a deep learning algorithm and a drone inspection can be provided to achieve real-time monitoring of the transmission line and its surrounding environment. The advantage of this combined technology is that it can automatically identify potential dangerous situations such as wildfires, and issue an alarm in a timely manner, thereby significantly improving the safety of the transmission line and the efficiency of the inspection work, and providing an efficient and intelligent wildfire automatic identification and alarm system for the transmission line. It not only improves the safety management standards of the transmission line, but also provides strong technical support for the stable operation of the power system.

[0106] See also Figure 6 , Figure 6 FIG. 1 is a structural block diagram of a wildfire identification device based on drone inspection in an embodiment of the present invention. Figure 6 The device shown is used for Figures 1 to 4 Any of the above-mentioned wildfire identification systems based on drone inspection, wherein the device comprises:

[0107] Data acquisition module 601: used to acquire image data of power transmission lines;

[0108] A wildfire identification module 602 is used to identify wildfires on the image data using a pre-trained deep learning model to determine a wildfire identification result of the transmission line, wherein the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line;

[0109] Prompt generation module 603: used to generate visual prompts based on the wildfire identification results.

[0110] Understandably, Figure 6 The functions of each module in the system are shown in Figure 5 The contents of each step in the wildfire identification method based on drone inspection are similar, so they are not described here to avoid repetition. For details, please refer to Figure 5 The contents of each step in the wildfire identification method based on drone inspection are shown.

[0111] The present invention provides a wildfire identification device based on drone inspection, the device includes: a data acquisition module: used to collect image data of a transmission line; a wildfire identification module: used to use a pre-trained deep learning model to identify the image data, and determine the wildfire identification result of the transmission line, the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on the historical wildfire image data of the transmission line; a prompt generation module: used to generate a visual prompt based on the wildfire identification result. Through the above device, a wildfire monitoring that combines deep learning algorithms and drone inspections can be provided to achieve real-time monitoring of transmission lines and their surroundings. The advantage of this combined technology is that it can automatically identify potential dangerous situations such as wildfires and issue alarms in a timely manner, thereby significantly improving the safety of transmission lines and the efficiency of inspection work, and providing an efficient and intelligent wildfire automatic identification and alarm system for transmission lines. It not only improves the safety management standards of transmission lines, but also provides strong technical support for the stable operation of power systems.

[0112] Figure 7 FIG. 1 shows an internal structure diagram of a computer device in an embodiment. The computer device may be a terminal or a server. Figure 7 As shown, the computer device includes a processor, a memory and a network interface connected via a system bus. The memory includes a non-volatile storage medium and an internal memory. The non-volatile storage medium of the computer device stores an operating system and may also store a computer program. When the computer program is executed by the processor, the processor can implement the above method. The internal memory may also store a computer program. When the computer program is executed by the processor, the processor can implement the above method. Those skilled in the art can understand that Figure 7 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0113] In one embodiment, a computer device is provided, comprising a memory and a processor, wherein the memory stores a computer program, and when the computer program is executed by the processor, the processor performs the following steps: Figure 5 The steps of the method shown.

[0114] In one embodiment, a computer-readable storage medium is provided, which stores a computer program. When the computer program is executed by a processor, the processor performs the following steps: Figure 5 The steps of the method shown.

[0115] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the program can be stored in a non-volatile computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0116] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0117] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A wildfire identification system based on drone inspection, characterized in that: The mountain fire identification system includes: a drone data acquisition system, a deep learning analysis system and a terminal visualization system, and the drone data acquisition system, the deep learning analysis system and the terminal visualization system are connected in communication; The drone data acquisition system is used to collect image data of the power transmission line and send the image data to the deep learning analysis system; The deep learning analysis system is used to receive the image data, use a pre-trained deep learning model to perform wildfire identification on the image data, determine a wildfire identification result of the transmission line, and send the wildfire identification result to the terminal visualization system, the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line; The terminal visualization system is used to receive the wildfire identification result and generate a visualization prompt based on the wildfire identification result.

2. The wildfire identification system according to claim 1, characterized in that: The drone data acquisition system at least includes: a high-voltage transmission line ledger module, a sensor camera control module, a data acquisition storage unit module, and a data communication upload module, and the high-voltage transmission line ledger module, the sensor camera control module, the data acquisition storage unit module, and the data communication upload module are connected to each other by communication; The high-voltage transmission line ledger module is used to store the ledger information of the high-voltage transmission line and send the ledger information to the sensor camera control module; The sensor camera control module is used to receive the ledger information, perform data acquisition planning processing based on the ledger information, and obtain data acquisition instructions; perform image acquisition processing on the transmission line based on the data acquisition instructions to obtain the image data, and send the image data to the data acquisition storage unit module; The data acquisition storage unit module is used to receive and store the image data, and send the image data to the data communication upload module; The data communication upload module is used to receive the image data and upload the image data to the deep learning analysis system.

3. The forest fire identification system according to claim 2, characterized in that: The drone data acquisition system also includes a data analysis and processing module, the input end of the data analysis and processing module is communicatively connected to the output end of the data acquisition and storage unit module, and the output end of the data acquisition and storage unit module is communicatively connected to the input end of the data communication upload module; The data acquisition storage unit module is also used to send the image data to the data analysis and processing module; The data analysis and processing module is used to receive the image data, perform denoising on the image data, obtain denoised image data, and send the denoised image data to the data communication upload module; The data communication upload module is used to receive the denoised image data and upload the denoised image data to the deep learning analysis system.

4. The forest fire identification system according to any one of claims 1 to 3, characterized in that: The deep learning analysis system at least includes: a data preprocessing module, a data annotation processing module, an image cutting processing module, a neural network feature extraction module, a model training module, a regression prediction module and an output analysis result module, wherein the data preprocessing module, the data annotation processing module, the image cutting processing module, the neural network feature extraction module, the model training module, the regression prediction module and the output analysis result module are connected in communication; The data preprocessing module is used to receive the historical wildfire image data and image data, preprocess the historical wildfire image data, and obtain the preprocessed historical wildfire image data; The data labeling processing module is used to label the preprocessed historical wildfire image data with training labels to obtain the historical wildfire image data labeled with training labels; The image cutting processing module is used to perform image cutting on the historical wildfire image data marked with training labels using a preset image segmentation rule to obtain a plurality of image blocks; The neural network feature extraction module is used to extract features from the image block using a neural network model to obtain image features of the image block; The model training module is used to perform wildfire recognition training on the deep learning model using image features and training labels to obtain a trained deep learning model; The regression prediction module is used to perform regression prediction using the trained deep learning model and the image data to determine the wildfire recognition result of the transmission line; The output analysis result module is used to send the wildfire identification result to the terminal visualization system.

5. The wildfire identification system according to claim 1, characterized in that: The terminal visualization system at least includes: a short message alarm push module and a visualization display module, and there is a communication connection between the short message alarm push module and the visualization display module; The SMS alarm push module is used to receive the wildfire identification result, and output SMS alarm information to the visual display module according to the wildfire identification result and the current operating status of the transmission line; The visual display module is used to receive the SMS warning information and output a visual prompt based on the SMS warning information.

6. The wildfire identification system according to claim 5, characterized in that: The terminal visualization system further includes a decision-making auxiliary display module and a feedback closed-loop module, and the decision-making auxiliary display module and the feedback closed-loop module are communicatively connected with each other; The auxiliary decision display module is used to receive the manual auxiliary decision result and send the auxiliary decision result to the feedback closed-loop module, wherein the auxiliary decision result is used to indicate the confirmation result of the manual auxiliary decision on the accuracy of the wildfire identification result; The feedback closed-loop module is used to receive the auxiliary decision result, and use the auxiliary decision result as feedback data to optimize the trained deep learning model to obtain an optimized deep learning model.

7. A method for identifying wildfires based on drone inspection, characterized in that: The method is applied to the wildfire identification system based on drone inspection as described in any one of claims 1 to 6, and the method comprises: Collect image data of power transmission lines; Using a pre-trained deep learning model to perform wildfire recognition on the image data, and determining a wildfire recognition result of the transmission line, wherein the wildfire recognition result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line; A visual prompt is generated based on the wildfire identification result.

8. A wildfire identification device based on drone inspection, characterized in that: The device is applied to the wildfire identification system based on drone inspection as described in any one of claims 1 to 6, and the device comprises: Data acquisition module: used to collect image data of transmission lines; A wildfire identification module: used to identify wildfires on the image data using a pre-trained deep learning model, and determine a wildfire identification result of the transmission line, wherein the wildfire identification result is used to reflect whether there is a wildfire on the transmission line, and the deep learning model is pre-trained based on historical wildfire image data of the transmission line; Prompt generation module: used to generate visual prompts based on the wildfire identification results.

9. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the processor is caused to perform the steps of the method according to claim 7.

10. A computer device comprising a memory and a processor, characterized in that: The memory stores a computer program, and when the computer program is executed by the processor, the processor executes the steps of the method according to claim 7.

Images