Photovoltaic station intelligent fire extinguishing method and system based on unmanned aerial vehicle cluster

Abstract

The invention discloses a photovoltaic station intelligent fire extinguishing method and system based on an unmanned aerial vehicle cluster, and relates to the technical field of new energy power station safety and fire protection. The unmanned aerial vehicle cluster and a fixed monitoring point cooperatively collect multi-source data, and after fusion analysis, a common equipment fire and a transformer insulating oil fire are accurately distinguished; according to the fire behavior type and the real-time state of the unmanned aerial vehicle, assigning an adaptive unmanned aerial vehicle and planning an optimal path; differential fire extinguishing strategies are executed according to different fire behaviors, and transformer oil fire is sprayed in a safe distance covering mode; after fire extinguishing, the unmanned aerial vehicle is automatically dispatched to return to the corresponding guarantee station to complete accurate supply. The system solves the problems that in the prior art, all links are split, fire behavior recognition is not accurate, dispatching is low in efficiency, supply is insufficient and the like, unmanned, precise and uninterrupted guarding of the whole fire fighting process of the photovoltaic station is achieved, and the fire fighting safety and efficiency are greatly improved.

Description

Technical Field

[0001] This invention relates to the field of safety and fire protection technology for new energy power plants, specifically to an intelligent fire extinguishing method and system for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms. Background Technology

[0002] As an important clean energy facility, photovoltaic power generation stations are expanding in scale and becoming increasingly densely equipped, often containing high-risk oil-filled electrical equipment such as main transformers in substations. These stations are generally located in remote, open, or even arid areas, posing a severe and unique challenge to fire safety. Currently, photovoltaic power stations mainly rely on traditional manual inspections, fixed smoke and heat detectors, and limited fixed fire-fighting facilities. These methods have significant limitations when dealing with photovoltaic power station scenarios: manual inspections are infrequent and inefficient, making it difficult to detect hidden or initial fires in a timely manner; fixed detectors have limited coverage and cannot meet the needs of comprehensive monitoring of widely distributed equipment and complex environments; and fixed fire-fighting facilities (such as fire hydrants and sprinklers) are ineffective in extinguishing special fires, especially those involving main transformers (such as insulating oil fires), due to high pipeline laying costs, numerous blind spots, and insufficient extinguishing efficiency.

[0003] In recent years, drone technology has been attempted to be applied in the field of firefighting, but its application in photovoltaic power plant scenarios is still in its early stages. Existing solutions mostly focus on the function of drones as a single firefighting vehicle, generally suffering from a broken chain in the "early warning-identification-decision-firefighting-support" process. Specifically, this manifests as: a lack of comprehensive, real-time fire awareness capabilities integrated with fixed monitoring networks; insufficient fire identification accuracy, particularly difficulty in accurately distinguishing between ordinary equipment fires and dangerous transformer oil fires, leading to an inability to automatically match the optimal firefighting strategy; and limitations in drone endurance and payload capacity, lacking a complete, automated ground energy and diversified fire extinguishing agent supply support system, hindering unmanned, uninterrupted routine monitoring and immediate response. Therefore, how to construct a complete closed-loop system integrating automatic inspection and early warning, accurate fire identification, intelligent task scheduling, adaptive and efficient firefighting, and continuous logistical support has become a pressing technical challenge to improve the fire safety level of photovoltaic power plants. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the purpose of this invention is to provide an intelligent fire suppression method and system for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms. This method solves problems such as fragmented processes, inaccurate fire identification, inefficient dispatching, and insufficient supplies in existing technologies. It enables unmanned, precise, and uninterrupted fire suppression throughout the entire process of photovoltaic power plant fire suppression, significantly improving fire safety and efficiency.

[0005] To achieve the above objectives, the embodiments of this invention provide the following technical solutions:

[0006] This application provides an intelligent fire suppression method for photovoltaic power plants based on a drone swarm, comprising the following steps: S1, controlling the drone swarm to automatically inspect the photovoltaic power plant according to a preset gridded route, while simultaneously acquiring sensor data from fixed monitoring points deployed in key equipment areas, and fusing them to form a comprehensive monitoring information for the entire power plant area; S2, based on the comprehensive monitoring information, using a central control platform to fuse and analyze visible light images, infrared thermal imaging data, and environmental sensor data, identifying the location of the fire, and classifying the fire as a common equipment fire or a transformer insulating oil fire based on flame morphology, smoke color, and heat radiation distribution characteristics; S3, according to the fire type... The system monitors the location, scale, and real-time status of each drone in the drone swarm, intelligently assigns drones equipped with matching types of extinguishing agents, and plans the optimal flight path for the drones, balancing safety and efficiency; S4, controls the assigned drones to fly to the fire scene and executes corresponding fire extinguishing strategies according to the type of fire, including controlling the drones to perform covering spraying operations at a preset safe distance from the fire source for transformer insulating oil fires; S5, during or after the fire extinguishing operation, monitors the drone status in real time, automatically dispatches the drones to return to the corresponding ground support station, and sequentially or simultaneously completes the replenishment of extinguishing agents and energy to support uninterrupted duty and operation cycles.

[0007] Further, step S1 includes: S11, the drone flies along a preset route and collects visual images and temperature data of the photovoltaic equipment in real time using its onboard high-definition visible light camera and infrared thermal imager; S12, it simultaneously receives monitoring data from fixed temperature, smoke, and flame detectors deployed in key areas of the substation, transformer, and combiner box; S13, it receives wind speed, wind direction, and humidity data collected by environmental sensors; S14, it fuses the visual images with the temperature data, monitoring data, and wind speed, wind direction, and humidity data, and uploads them to the central intelligent control platform.

[0008] Further, in step S2, classifying the fire as a general equipment fire or a transformer insulating oil fire based on flame morphology, smoke color, and heat radiation distribution characteristics specifically includes: S21, analyzing visible light images to identify whether there are intense flames accompanied by thick black smoke; S22, analyzing infrared thermal imaging data to detect whether the abnormally high-temperature area highly matches the physical outline of the transformer equipment and exhibits liquid flow characteristics; S23, if both intense black smoke and the flowing high-temperature characteristics of a specific outline are met, it is determined to be a transformer insulating oil fire, triggering the highest level alarm; otherwise, it is determined to be a general equipment fire.

[0009] Furthermore, in step S3, planning the optimal flight path for the UAV that balances safety and efficiency specifically includes: S31, obtaining the precise coordinates of the fire scene and, in conjunction with real-time wind speed and direction data, constructing an initial candidate flight path set considering the influence of airflow; S32, for transformer insulating oil fires, screening and excluding flight paths that pass directly above the transformer body and the potential coverage area of ​​the oil drain pit; S33, based on the real-time remaining power and payload of the assigned UAV and the flight path set, constructing an optimization model with the fastest safe arrival time as the objective, and dynamically solving for the optimal flight path.

[0010] Further, in step S4, the corresponding fire extinguishing strategy according to the type of fire specifically includes: S41, if it is a fire involving ordinary equipment, dispatching a water-carrying drone to fly directly above the fire point, maintaining the optimal spraying height through laser ranging, and performing precise spraying at a fixed point; S42, if it is a fire involving transformer insulating oil, executing a special fire extinguishing procedure: S421, dispatching a drone carrying compressed air foam or dry powder extinguishing agent; S422, controlling the drone to fly to a predetermined safe working area 10-20 meters away from the transformer body; S423, controlling the drone to fly along a preset rectangular or spiral path covering the transformer body and the surface of the drain pit, while simultaneously activating the integrated foam generating device for large-scale coverage spraying; S424, after the open flame is extinguished, instructing the drone to perform periodic infrared scanning at a safe distance until it is confirmed that there is no risk of reignition.

[0011] Further, in step S5, automatically scheduling the UAV to return to the corresponding ground support station specifically includes: S51, real-time monitoring of the UAV's remaining fire extinguishing agent and battery power; S52, when the remaining agent or battery power is lower than a preset threshold, or when a fire extinguishing mission completion signal is received, automatically generating a return command containing the target support station type; S53, the UAV flies to the corresponding intelligent water storage station or dedicated fire extinguishing agent replenishment station according to the command; S54, using a combination of visual recognition and UWB precise positioning technology, guiding the UAV to complete an autonomous landing with centimeter-level precision on the replenishment station docking platform.

[0012] Further, in step S5, the sequential or simultaneous completion of fire extinguishing agent replenishment and energy replenishment specifically includes: S55, after the UAV lands, the automatic docking robotic arm of the replenishment station completes a sealed connection with the quick-connect interface of its fire extinguishing agent storage tank; S56, the net weight change of the UAV storage tank is monitored in real time by a high-precision weighing sensor suspended below the docking mechanism; S57, the central control system presets a target weight according to the fire extinguishing agent requirements of the next task and sends a control command to the replenishment station; S58, the replenishment station opens the corresponding solenoid valve or pneumatic delivery valve according to the command, injects the fire extinguishing agent, and provides real-time feedback of weight data; wherein, when injecting dry powder or foam concentrate, the fire extinguishing agent is delivered through a special pipeline with a Teflon coating on the inner wall to reduce adhesion and residue; S59, when the real-time net weight reaches the target weight, the valve is automatically closed, the robotic arm is disconnected, and a precise replenishment is completed.

[0013] Furthermore, in step S423 of the special fire extinguishing process, the large-area coverage spraying specifically includes: S423a, during the spraying operation, acquiring the current wind speed and direction data collected by the environmental sensors in real time; S423b, based on the current wind speed and direction, the real-time hovering or flight position of the drone, and the initial velocity and particle size distribution of the extinguishing agent, predicting the dispersion trajectory and landing point distribution of the extinguishing agent in the air; S423c, comparing the predicted landing point distribution with the target coverage area, if the predicted landing point deviates from the target area by more than a preset threshold due to the influence of wind, dynamically adjusting the horizontal position and / or spray angle of the drone, or adjusting the spray pressure to change the initial velocity of the extinguishing agent, so that the compensated predicted landing point re-covers the target area.

[0014] Furthermore, the method also includes step S6, which specifically includes: S61, after step S4 is completed, assigning another drone equipped with an infrared thermal imager to continuously or periodically scan the area where the fire has been extinguished; S62, comparing and analyzing the temperature data obtained from the scan with historical normal data or data from the surrounding area; S63, if an abnormal temperature rise point is detected, an automatic secondary alarm is triggered, and nearby standby drones are dispatched to handle the situation.

[0015] Accordingly, this application also provides an intelligent fire suppression system for photovoltaic power plants based on a drone swarm, comprising: an acquisition component for controlling the drone swarm to automatically inspect the photovoltaic power plant according to a preset gridded route, while acquiring sensor data from fixed monitoring points deployed in key equipment areas, and fusing them to form comprehensive monitoring information for the entire power plant area; a discrimination component for identifying the location of the fire based on the comprehensive monitoring information of the power plant area, through a central control platform fusing and analyzing visible light images, infrared thermal imaging data, and environmental sensor data, and classifying the fire as a common equipment fire or a transformer insulating oil fire based on flame morphology, smoke color, and heat radiation distribution characteristics; and a planning component for planning based on the fire situation. The system intelligently assigns drones carrying matching types of extinguishing agents based on their type, location, scale, and the real-time status of each drone in the drone swarm, and plans the optimal flight path for each drone while balancing safety and efficiency. An execution component controls the assigned drones to reach the fire scene and executes corresponding firefighting strategies based on the fire type, including controlling drones to perform smothering spraying operations at a preset safe distance from the fire source for transformer insulating oil fires. A resupply component monitors the drone status in real time during or after firefighting operations, automatically scheduling the drones to return to the corresponding ground support station to sequentially or simultaneously replenish extinguishing agents and energy, supporting uninterrupted monitoring and operational cycles.

[0016] The beneficial effects of this invention are as follows: Through a closed-loop design encompassing "full-area collaborative inspection - multi-mode fire classification - adaptive dispatch - precise fire suppression - intelligent continuous support," the unmanned, automated, and precise fire protection of photovoltaic power plants is achieved. On the one hand, by integrating data from drone swarms and fixed monitoring points, monitoring blind spots are eliminated, ensuring early detection of fires. On the other hand, specific fire classification and matching dispatch, along with fire suppression strategies, improve fire suppression efficiency and reduce secondary risks. Simultaneously, the automated resupply mechanism solves the bottlenecks of drone endurance and fuel delivery, enabling uninterrupted monitoring and significantly enhancing the comprehensiveness and reliability of fire safety protection at photovoltaic power plants. Attached Figure Description

[0017] Figure 1 A flowchart illustrating an intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms, provided as an embodiment of this application;

[0018] Figure 2 This is a schematic diagram of the structure of an intelligent fire extinguishing system for photovoltaic power plants based on a drone swarm, provided as an embodiment of this application. Detailed Implementation

[0019] The specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are for illustration and explanation only and are not intended to limit the scope of the present invention.

[0020] In this invention, the terms "system" and "network" are used interchangeably. "Multiple" refers to two or more; therefore, in this invention, "multiple" can also be understood as "at least two." "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. Additionally, the character " / ", unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship. Furthermore, it should be understood that in the description of this invention, terms such as "first" and "second" are used only for descriptive purposes and should not be construed as indicating or implying relative importance or order.

[0021] In existing photovoltaic power plant fire protection technologies, the processes of inspection, fire identification, drone dispatch, fire suppression execution, and logistical support are fragmented and lack a comprehensive collaborative mechanism. Either they rely solely on drone inspections, neglecting data support from fixed monitoring points and resulting in monitoring blind spots; or fire identification fails to distinguish between ordinary equipment fires and high-risk transformer insulating oil fires, leading to blind fire suppression strategies; or dispatching drones fails to consider their real-time status (power level, agent type), resulting in resource waste or fire suppression delays. Furthermore, the lack of a continuous energy and extinguishing agent replenishment loop makes it impossible to achieve 24 / 7 uninterrupted monitoring and is insufficient to meet the fire protection needs of photovoltaic power plants covering vast areas and involving complex equipment.

[0022] like Figure 1 As shown, this application provides an intelligent fire suppression method for photovoltaic power plants based on a drone swarm, including the following steps: S1, controlling the drone swarm to automatically inspect the photovoltaic power plant according to a preset gridded route, while acquiring sensor data from fixed monitoring points deployed in key equipment areas, and fusing them to form a comprehensive monitoring information for the entire power plant; S2, based on the comprehensive monitoring information, using a central control platform to fuse and analyze visible light images, infrared thermal imaging data, and environmental sensor data, identifying the location of the fire, and classifying the fire into a common equipment fire or a transformer insulating oil fire based on flame morphology, smoke color, and heat radiation distribution characteristics; S3, according to the fire type... The system intelligently assigns drones equipped with matching fire extinguishing agents based on their type, location, scale, and the real-time status of each drone in the drone swarm, and plans the optimal flight path for each drone while balancing safety and efficiency; S4, it controls the assigned drones to fly to the fire scene and executes corresponding fire extinguishing strategies according to the type of fire, including controlling the drones to perform covering spraying operations at a preset safe distance from the fire source for transformer insulating oil fires; S5, during or after the fire extinguishing operation, it monitors the drone status in real time and automatically schedules the drones to return to the corresponding ground support station to complete the replenishment of fire extinguishing agents and energy in sequence or simultaneously, so as to support uninterrupted duty and operation cycle.

[0023] In another possible embodiment, step S1: The central intelligent control platform pre-plans a gridded inspection route based on the distribution of photovoltaic power station equipment and terrain characteristics, and distributes it to the drone swarm. The drone swarm flies autonomously according to the route. At the same time, fixed monitoring points (temperature, smoke, and flame detectors) deployed in key areas such as the booster station, transformers, and combiner boxes continuously collect sensor data. This data is fused with the data collected by the drones and meteorological data collected by environmental sensors (wind speed, wind direction, and humidity) to form monitoring information covering the entire power station area, and is uploaded to the central intelligent control platform in real time. Step S2: After receiving the monitoring information covering the entire area, the central intelligent control platform analyzes the flame and smoke features in the visible light image using image recognition algorithms, processes abnormally high temperature areas in the infrared thermal imaging data using temperature analysis algorithms, and corrects the identification results by combining environmental data. Further, based on flame morphology (whether it burns intensely), smoke color (whether it is thick black smoke), and heat radiation distribution (whether it exhibits liquid flow characteristics), the fire is accurately classified into ordinary equipment fire or transformer insulating oil fire, and the fire location is marked on the electronic map, triggering the corresponding alarm level (ordinary fire triggers a level 2 alarm, transformer insulating oil fire triggers the highest level alarm). Step S3: The central intelligent control platform acquires real-time status data of each drone in the drone cluster, including remaining battery power, carrier type (water / foam / dry powder), carrier remaining amount, and current location. Based on the fire type (e.g., matching foam or dry powder extinguishing agent for transformer insulating oil fire), location (straight-line distance from the drone and flight path difficulty), and scale (small fire, single drone; large fire, multiple drones working together), the optimal drone is selected. At the same time, the optimal flight path is planned: avoiding obstacles and high-risk areas (e.g., avoiding the area covered by oil spill pits in the case of transformer fire), and optimizing the flight trajectory in combination with real-time wind speed and direction, balancing flight speed and safety. Step S4: After receiving the flight command, the assigned drone autonomously flies to the fire scene and locates the center of the fire source using a visual perception system and a ranging sensor. If it is a fire involving ordinary equipment, the drone hovers directly above the fire source at the optimal spraying height and activates the spraying device for precise, targeted spraying. If it is a transformer insulating oil fire, the drone flies to a preset safe distance of 10-20 meters from the fire source and follows a preset rectangular or spiral coverage path, simultaneously activating the foam generator or dry powder spraying device to ensure the extinguishing agent completely covers the transformer body and the surface of the drain pit. Step S5: During the firefighting operation, the central intelligent control platform monitors the drone's remaining extinguishing agent and battery level in real time. When the remaining amount or battery level falls below a preset threshold (e.g., 20% battery remaining, 30% extinguishing agent remaining), or when the firefighting mission is completed, the platform automatically generates a return command, specifying the target support station type (return to the intelligent water storage station for ordinary fires, return to the dedicated extinguishing agent replenishment station for transformer oil fires).After receiving instructions, the drone autonomously flies to the target support station to replenish fire extinguishing agents and energy. After replenishment, it enters standby mode to await the next mission scheduling, forming a "operation-replenishment-standby" cycle to support uninterrupted duty.

[0024] Through a closed-loop design encompassing "full-area collaborative inspection - multi-mode fire classification - adaptive dispatch - precise fire suppression - intelligent continuous support," the system achieves unmanned, automated, and precise fire protection for photovoltaic power plants. On one hand, it integrates data from drone swarms and fixed monitoring points to eliminate monitoring blind spots and ensure early fire detection. On the other hand, it improves fire suppression efficiency and reduces secondary risks through specific fire classification and matching dispatch and fire suppression strategies. Simultaneously, the automated resupply mechanism solves the bottlenecks of drone endurance and fuel delivery, enabling uninterrupted monitoring and significantly enhancing the comprehensiveness and reliability of fire safety protection at photovoltaic power plants.

[0025] Existing photovoltaic power plant inspection technologies either rely solely on drones to collect visual or temperature data, resulting in a single data dimension that fails to comprehensively reflect the equipment status; or they rely solely on fixed monitoring points, leading to limited coverage and numerous monitoring blind spots. Furthermore, the lack of effective integration of drone data, fixed-point data, and environmental data results in fragmented monitoring information, failing to provide comprehensive and accurate basic data for subsequent fire identification, thus affecting the timeliness and accuracy of fire identification.

[0026] In this embodiment of the application, step S1 includes: S11, the drone flies along a preset route and collects visual images and temperature data of the photovoltaic equipment in real time using a high-definition visible light camera and an infrared thermal imager; S12, it simultaneously receives monitoring data from fixed temperature, smoke, and flame detectors deployed in key areas of the substation, transformer, and combiner box; S13, it receives wind speed, wind direction, and humidity data collected by environmental sensors; S14, it fuses the visual images with the temperature data, monitoring data, and wind speed, wind direction, and humidity data, and uploads them to the central intelligent control platform.

[0027] In another possible embodiment, step S11: The drone flies along a gridded flight path preset by the central intelligent control platform. During flight, the onboard high-definition visible light camera captures real-time visual images of the photovoltaic equipment (photovoltaic panels, combiner boxes, transformers, etc.) and records the equipment's appearance (whether there is damage, smoke, or open flame); the infrared thermal imager simultaneously collects the equipment's temperature data and captures abnormally high-temperature areas (such as combiner box temperatures exceeding 85°C). All data is transmitted to the central intelligent control platform in real time. Step S12: Fixed monitoring points deployed in key areas such as the substation, transformers, and combiner boxes continuously monitor the surface temperature of the equipment with temperature sensors, smoke sensors monitor the surrounding smoke concentration, and flame detectors detect the presence of open flames in real time. This monitoring data is uploaded to the central intelligent control platform at a preset frequency (e.g., once every 5 seconds), complementing the data collected by the drone. Step S13: Environmental sensors deployed within the site collect real-time meteorological data such as wind speed, wind direction, and humidity. Wind speed and wind direction data are used for subsequent flight path planning and fire suppression spray compensation, while humidity data is used to assist in judging the speed of fire spread. Environmental data is simultaneously uploaded to the central intelligent control platform. Step S14: After receiving data collected by the UAV, fixed-point monitoring data, and environmental data, the central intelligent control platform performs correlation analysis on the multi-source data through data fusion algorithms (such as combining smoke images taken by a UAV in a certain area with the concentration data of fixed smoke sensors in that area), eliminates abnormal data (such as invalid data caused by sensor failure), and forms complete and accurate full-domain monitoring information to provide data support for subsequent fire identification.

[0028] Through multi-dimensional data collection and fusion, comprehensive monitoring of photovoltaic power plants without blind spots was achieved. The mobile inspection by drones and the fixed-point monitoring by fixed monitoring points complement each other, covering a wide area and key equipment. The fusion of visual images, temperature data, and environmental data enriches the dimensions of monitoring information, providing comprehensive and accurate data support for subsequent multi-mode fire identification, significantly reducing the probability of missed or false fire detection, and buying time for early fire response.

[0029] Current fire detection technologies can only determine whether a fire has occurred, but cannot accurately distinguish the type of fire, especially high-risk Class B liquid fires such as transformer insulating oil fires. Because the combustion characteristics of transformer insulating oil fires (intense burning, thick black smoke, and liquid flow and spread) differ greatly from those of ordinary solid equipment fires, treating them as ordinary fires using water-based extinguishing agents not only results in poor extinguishing effects but may also trigger secondary risks such as oil spraying and explosions, seriously threatening the safety of photovoltaic power plants.

[0030] In this embodiment of the application, step S2, classifying the fire as a general equipment fire or a transformer insulating oil fire based on flame morphology, smoke color, and thermal radiation distribution characteristics, specifically includes: S21, analyzing visible light images to identify whether there are intense flames accompanied by thick black smoke; S22, analyzing infrared thermal imaging data to detect whether the abnormally high-temperature area highly matches the physical outline of the transformer equipment and exhibits liquid flow characteristics; S23, if both intense black smoke and the flowing high-temperature characteristics of a specific outline are met, it is determined to be a transformer insulating oil fire, triggering the highest level alarm; otherwise, it is determined to be a general equipment fire.

[0031] In another possible embodiment, step S21: The central intelligent control platform analyzes the visible light images uploaded by the drone frame by frame, extracting flame morphology and smoke color features through image recognition algorithms. If a fiercely burning flame is detected in the image, accompanied by a large amount of thick black smoke (distinct from the gray or white smoke of ordinary solid fires), it is initially determined to be a suspected transformer insulating oil fire. Step S22: Infrared thermal imaging data is analyzed simultaneously to extract the contour and distribution characteristics of the abnormal high-temperature area. If the physical contour of the abnormal high-temperature area highly matches the preset transformer equipment contour of the photovoltaic power station, and the high-temperature area exhibits obvious liquid flow characteristics (i.e., the high-temperature range extends along the ground or equipment surface, consistent with the distribution pattern of combustion after insulating oil leakage), the suspected conclusion is further confirmed. Step S23: If the "intense black smoke" characteristic of step S21 and the "specific outline + flowing high temperature" characteristic of step S22 are met simultaneously, it is finally determined to be a transformer insulating oil fire. The central intelligent control platform immediately triggers the highest level alarm and sends an emergency notification to the site staff at the same time. If only one of the characteristics is met or neither is met, it is determined to be a fire of ordinary equipment and triggers a level two alarm, providing a clear basis for the formulation of subsequent fire extinguishing strategies.

[0032] By employing a dual-feature judgment logic, the system achieves specific identification of transformer insulating oil fires, accurately distinguishing between ordinary equipment fires and high-risk transformer insulating oil fires. This provides crucial information for subsequently matching specialized extinguishing agents and firefighting strategies. It avoids problems such as ineffective firefighting and secondary risks caused by misjudging the fire type, significantly improving the safety and effectiveness of fire response in high-risk areas.

[0033] Existing drone firefighting path planning technologies mostly focus on the "shortest distance" without fully considering the safety risks at the fire site (such as the oil spill pit coverage area and high-temperature radiation area in transformer insulating oil fires) and the real-time status of the drone (remaining battery power and payload). This may result in the planned path passing through high-risk areas, threatening the safety of the drone; or, due to the lack of consideration of remaining battery power, the drone may lose power midway and be unable to reach the fire site, affecting firefighting efficiency.

[0034] In this embodiment of the application, step S3, which plans the optimal flight path for the UAV that takes into account both safety and efficiency, specifically includes: S31, obtaining the precise coordinates of the fire scene and, in conjunction with real-time wind speed and direction data, constructing an initial candidate flight path set considering the influence of airflow; S32, for transformer insulating oil fires, screening and excluding flight paths that pass directly above the transformer body and the potential coverage area of ​​the oil drain pit; S33, based on the real-time remaining power and payload of the assigned UAV and the flight path set, constructing an optimization model with the fastest safe arrival time as the objective, and dynamically solving for the optimal flight path.

[0035] In another possible embodiment, step S31: The central intelligent control platform acquires the precise coordinates of the fire site (determined through UAV positioning and image calibration), and combines this with obstacle information (such as utility poles and buildings) from the site's electronic map to generate multiple initial candidate flight paths from the UAV's current location to the fire site. Simultaneously, real-time wind speed and direction data are incorporated to adjust the flight path's difficulty (e.g., a headwind path requires more power), forming an initial candidate flight path set that considers airflow influence. Step S32: Risk screening of the initial candidate flight paths is performed based on the fire type. If it is a transformer insulating oil fire, the system automatically retrieves the preset coordinate range of the area directly above the transformer and the potential coverage area of ​​the oil drain pit (calculated based on the transformer model and oil storage capacity), screening and excluding all flight paths passing through this high-risk area to ensure the UAV's flight path stays away from explosion and high-temperature radiation risks. Step S33: An optimization model is constructed based on the assigned UAV's real-time status data (remaining battery power, payload, current location). The model aims for the "fastest safe arrival time," with constraints including: the total flight path length does not exceed the maximum flight distance supported by the UAV's remaining battery power, and the flight path avoids all high-risk areas and obstacles. The model is solved by dynamic programming algorithm to output the optimal flight path, ensuring that the drone can reach the fire site as quickly as possible while ensuring safety.

[0036] By employing a path optimization logic of "risk screening + status adaptation," an optimal flight path that balances safety and efficiency is planned. This not only avoids high-risk areas of the fire scene, ensuring the safety of the drone's flight, but also combines real-time dynamic adjustments to the drone's status to ensure that the drone can safely and quickly reach the fire scene, gaining valuable time for early firefighting and improving the reliability and timeliness of the firefighting mission.

[0037] Existing drone firefighting strategies lack specificity, employing a uniform spraying method for both ordinary equipment fires and high-risk transformer insulating oil fires, resulting in poor firefighting effectiveness: if the spraying area is too large for ordinary equipment fires, it wastes water resources; if the spraying is too close or the oil drain pit is not covered for transformer insulating oil fires, it cannot completely isolate the oil surface from the air, which can easily cause reignition, and close-range operations threaten the safety of drones.

[0038] In this embodiment, step S4, which involves executing a corresponding fire extinguishing strategy based on the type of fire, specifically includes: S41, if it is a fire involving ordinary equipment, dispatching a water-carrying drone to fly directly above the fire point, maintaining the optimal spraying height through laser ranging, and performing precise, targeted spraying; S42, if it is a fire involving transformer insulating oil, executing a special fire extinguishing procedure: S421, dispatching a drone carrying compressed air foam or dry powder extinguishing agent; S422, controlling the drone to fly to a predetermined safe working area 10-20 meters away from the transformer body; S423, controlling the drone to fly along a preset rectangular or spiral path covering the transformer body and the surface of the drain pit, while simultaneously activating the integrated foam generator for large-scale coverage spraying; S424, after the open flame is extinguished, instructing the drone to perform periodic infrared scanning at a safe distance until it is confirmed that there is no risk of reignition.

[0039] In another possible embodiment, step S41: If the fire is determined to be a fire involving ordinary equipment (such as photovoltaic panels or combiner boxes), the central intelligent control platform instructs a water-carrying drone to fly to the fire scene. After the drone arrives, it measures the vertical distance to the fire source using a laser rangefinder, adjusts its flight altitude to the optimal spraying height (e.g., 5-8 meters), and after hovering and stabilizing, activates the spraying device to precisely spray the center of the fire source until the open flames are extinguished.

[0040] Step S42: Special fire extinguishing procedure for transformer insulating oil fire:

[0041] Step S421: The central intelligent control platform selects a dedicated drone equipped with compressed air foam or dry powder fire extinguishing agent and issues fire extinguishing mission instructions to it.

[0042] Step S422: The drone flies along the planned optimal path and arrives at the preset safe working area 10-20 meters away from the transformer body (this distance is calculated based on the transformer's oil storage capacity and fire radiation range to avoid high temperature and explosion risks), hovering stably.

[0043] Step S423: The drone flies along a preset rectangular or spiral path (the path covers the transformer body and the surface of the drain pit), and simultaneously starts the integrated foam generator to spray high-quality compressed air foam to ensure that the foam evenly covers the oil surface and isolates the air to block combustion.

[0044] Step S424: After the open flames are extinguished, the drone maintains a safe distance and performs an infrared scan of the fire area every 30 seconds to collect temperature data and upload it to the central intelligent control platform. If no abnormal high-temperature points are found in three consecutive scans, it is determined that there is no risk of reignition, and the drone executes the return-to-home command; if an abnormality is found, it continues to monitor and provides timely feedback.

[0045] By designing differentiated fire extinguishing strategies for different types of fires, "precise spraying at fixed points" is used for ordinary equipment fires to save fire extinguishing resources; a special process of "safe distance + coverage spraying + reignition monitoring" is used for transformer insulating oil fires to ensure that the extinguishing agent fully covers the fire source, completely isolates the oil surface from the air, and avoids drones from coming into close contact with high-risk areas, thereby improving the fire extinguishing effect and operational safety, and preventing fire rebound through reignition monitoring.

[0046] The existing drone resupply system lacks an automated scheduling mechanism, requiring manual control of drones to return to the resupply point, which is inefficient. In addition, the landing positioning accuracy is insufficient, causing drones to fail to accurately dock with the resupply equipment, affecting the resupply progress. Furthermore, the lack of clear distinction between the types of resupply stations corresponding to different fire conditions may cause drones to go to the wrong resupply station, delaying the execution of subsequent tasks.

[0047] In this embodiment, step S5, automatically scheduling the UAV to return to the corresponding ground support station, specifically includes: S51, real-time monitoring of the UAV's remaining fire extinguishing agent and battery power; S52, automatically generating a return command containing the target support station type when the remaining agent or battery power is below a preset threshold, or when a fire extinguishing mission completion signal is received; S53, the UAV flies to the corresponding intelligent water storage station or dedicated fire extinguishing agent replenishment station according to the command; S54, using a combination of visual recognition and UWB precise positioning technology, guiding the UAV to complete an autonomous landing with centimeter-level precision on the replenishment station docking platform.

[0048] In another possible embodiment, step S51: Real-time status monitoring: The central intelligent control platform acquires the battery power and fire extinguishing agent balance data of the drone in real time through the power sensor and fire extinguishing agent balance sensor carried by the drone, with a monitoring frequency of once per second.

[0049] Step S52: Return-to-Home Command Generation: When the drone's battery level is below a preset threshold (e.g., 20%) or the remaining fire extinguishing agent is below a preset threshold (e.g., 30%), or the central intelligent control platform receives a fire extinguishing mission completion signal, the system automatically generates a return-to-home command. The command specifies the target support station type: drones for ordinary equipment fires go to the intelligent water storage station, and drones for transformer insulating oil fires go to the dedicated fire extinguishing agent replenishment station. Step S53: Fly to the Target Support Station: After receiving the return-to-home command, the drone autonomously plans its flight path to the target support station, avoiding obstacles and fire areas, and flies to the target support station. Step S54: Precise Autonomous Landing: When the drone approaches the support station, it activates a landing guidance system that integrates visual recognition and UWB precise positioning. The visual recognition module captures the visual identifiers of the support station docking platform, while simultaneously receiving positioning signals from the UWB base station. This dual positioning corrects the flight attitude, ultimately achieving centimeter-level precision autonomous landing on the docking platform, preparing for subsequent resupply.

[0050] The "state-triggered + precise positioning" return-to-home resupply mechanism enables automated scheduling and centimeter-level precise landing of UAVs. Automatic matching of resupply station types avoids resupply errors; precise docking ensures efficient resupply, significantly improving the automation level and efficiency of logistical support and providing support for uninterrupted UAV swarm monitoring.

[0051] Existing drone fire extinguishing agent replenishment technologies mostly use liquid level sensors to monitor the replenishment amount, which has low accuracy and is prone to over-replenishment (increasing drone payload and affecting flight efficiency) or under-replenishment (failing to meet fire extinguishing needs) due to sensor errors. In addition, the replenishment pipelines are not designed for the characteristics of different fire extinguishing agents, causing highly viscous fire extinguishing agents such as dry powder and foam concentrate to easily adhere to the inner wall of the pipeline, resulting in residue waste, or even blockage of the pipeline, affecting the service life of the replenishment system.

[0052] In this embodiment, step S5, which sequentially or simultaneously completes the replenishment of extinguishing agent and energy, specifically includes: S55, after the UAV lands, the automatic docking robotic arm of the replenishment station completes a sealed connection with the quick-connect interface of its extinguishing agent storage tank; S56, the net weight change of the UAV storage tank is monitored in real time by a high-precision weighing sensor suspended below the docking mechanism; S57, the central control system presets a target weight according to the extinguishing agent requirements of the next task and sends a control command to the replenishment station; S58, the replenishment station opens the corresponding solenoid valve or pneumatic delivery valve according to the command, injects the extinguishing agent, and provides real-time feedback on the weight data; wherein, when injecting dry powder or foam concentrate, the extinguishing agent is delivered through a dedicated pipeline with a Teflon coating on the inner wall to reduce adhesion and residue; S59, when the real-time net weight reaches the target weight, the valve is automatically closed, the robotic arm is disconnected, and a precise replenishment is completed.

[0053] In another possible embodiment, step S55: After the UAV completes a precise landing on the docking platform of the resupply station, the central control system of the resupply station sends a command to activate the automatic docking robotic arm. The robotic arm adjusts its attitude according to the UAV's positioning data, aligns with the quick-connect interface of the UAV's fire extinguishing agent storage tank, and completes a sealed connection to ensure no leakage during the resupply process. Step S56: The net weight data of the UAV's storage tank is collected in real time by a high-precision weighing sensor (accuracy ±5g) suspended below the docking mechanism, and the data is synchronously transmitted to the central control system to form a weight change curve. Step S57: The central control system calculates the required amount of fire extinguishing agent based on the type and scale of the fire in the next task, presets the target weight of the storage tank (e.g., 10kg of water for ordinary fires, 8kg of foam fire extinguishing agent for transformer oil fires), and sends a resupply control command to the resupply station. Step S58: After receiving the command, the resupply station opens the solenoid valve (water or foam concentrate) or pneumatic delivery valve (dry powder) of the corresponding storage tank, and injects the fire extinguishing agent into the UAV's storage tank through a dedicated pipeline. When injecting dry powder or foam concentrate, pipes with Teflon coating on the inner wall are used to reduce the adhesion and residue of extinguishing agent. During the delivery process, a flow sensor monitors the delivery speed in real time, and a pressure regulating valve dynamically adjusts the pressure to ensure stable delivery. Step S59: When the high-precision weighing sensor detects that the real-time net weight of the storage tank has reached the preset target weight, the central control system immediately sends a shutdown command to close the corresponding valve, and the robotic arm disconnects from the drone, completing a precise resupply.

[0054] By adopting a closed-loop control method based on high-precision weighing, the precise replenishment of extinguishing agents was achieved, ensuring that the replenishment amount perfectly matched the target demand. At the same time, a special pipeline was designed for viscous extinguishing agents to reduce adhesion and residue, lower the risk of pipeline blockage, improve the reliability of the replenishment system and the utilization rate of extinguishing agents, and ensure the smooth execution of subsequent fire fighting tasks.

[0055] Current drone-based fire suppression spraying technologies mostly involve spraying at fixed angles and pressures, without considering the impact of wind conditions on the landing point of the extinguishing agent. Photovoltaic power plants are often located in open areas, and changes in wind speed and direction can cause the extinguishing agent to drift through the air, deviating from the target coverage area and reducing fire suppression efficiency. This is especially true in the handling of transformer insulating oil fires, where if the foam cannot cover the oil surface, the fire cannot be completely extinguished.

[0056] In this embodiment of the application, step S423 of the special fire extinguishing process specifically includes the following steps: S423a, during the spraying operation, real-time acquisition of current wind speed and direction data collected by the environmental sensors; S423b, based on the current wind speed and direction, the real-time hovering or flight position of the drone, and the initial velocity and particle size distribution of the extinguishing agent, predicting the dispersion trajectory and landing point distribution of the extinguishing agent in the air; S423c, comparing the predicted landing point distribution with the target coverage area, and if the predicted landing point deviates from the target area by more than a preset threshold due to the influence of wind, dynamically adjusting the horizontal position and / or spray angle of the drone, or adjusting the spray pressure to change the initial velocity of the extinguishing agent, so that the compensated predicted landing point re-covers the target area.

[0057] In another possible embodiment, step S423a: During the drone's coverage spraying operation, the central intelligent control platform receives real-time wind speed and direction data collected by environmental sensors (updated twice per second) and transmits it synchronously to the drone's flight control system. Step S423b: Based on the current wind speed and direction data, combined with its real-time hovering or flight position (via GPS / RTK positioning), the initial velocity of the extinguishing agent (preset or adjustable), and particle size distribution (based on spraying device parameters), the drone's flight control system predicts the dispersion trajectory and final landing point distribution of the extinguishing agent in the air using a fluid dynamics model. Step S423c: The predicted landing point distribution is compared with the preset target coverage area (transformer body and oil drain pit surface), and the deviation difference is calculated. If the deviation exceeds the preset threshold (e.g., 5cm), the system will dynamically adjust according to the direction and degree of deviation: when there is a headwind, the spray pressure will be increased to increase the initial velocity of the extinguishing agent, or the horizontal position of the drone will be adjusted to shift towards the headwind; when there is a crosswind, the spray angle will be adjusted to tilt in the opposite direction of the crosswind, or the drone's flight path will be finely adjusted to ensure that the predicted landing point after compensation completely covers the target area again.

[0058] Through real-time wind field data-driven dynamic spray compensation control, the extinguishing agent dispersion trajectory is accurately predicted, and the drone position, spray angle or spray pressure are dynamically adjusted to counteract the influence of the wind field on the spraying effect, ensuring that the extinguishing agent accurately covers the target area. This significantly improves the fire extinguishing effect under complex weather conditions, and is especially suitable for fire disposal in open areas.

[0059] Current drone firefighting technologies mostly focus on extinguishing open flames, neglecting the risk of reignition after fire suppression. Hidden ignition sources (such as smoldering internal wiring) may exist inside photovoltaic power plant equipment (e.g., transformers, combiner boxes). These can easily reignite after an open flame is extinguished, and if not detected in time, the fire could spread and cause even greater losses.

[0060] In this embodiment of the application, the method further includes step S6, which specifically includes: S61, after step S4 is completed, assigning another drone equipped with an infrared thermal imager to continuously or periodically scan the area of ​​the extinguished fire; S62, comparing and analyzing the temperature data obtained from the scan with historical normal data or data from the surrounding area; S63, if an abnormal temperature rise point is found, an automatic secondary alarm is triggered, and nearby standby drones are dispatched to handle the situation.

[0061] In another possible embodiment, step S61: After the fire extinguishing operation in step S4 is completed, the central intelligent control platform immediately dispatches another drone equipped with an infrared thermal imager (preferably a standby drone that is close to the fire and has sufficient power) to perform a reignition monitoring task. Step S62: The monitoring drone flies to the area where the fire has been extinguished, performs a wide-area infrared scan along a preset scanning route, and collects temperature data in the area in real time. The central intelligent control platform compares and analyzes the collected temperature data with the historical normal temperature data of the area (or the real-time temperature data of surrounding areas where no fire has occurred) to identify whether there are any abnormal temperature rise points (such as temperatures exceeding the normal range by more than 50°C). Step S63: If an abnormal temperature rise point is detected, the central intelligent control platform automatically triggers a secondary alarm, simultaneously marks the location of the temperature rise point, and dispatches nearby standby drones (matching extinguishing agents according to the temperature rise point situation) to handle the situation; if no abnormal temperature rise point is detected after multiple consecutive scans (such as 3 times, with an interval of 5 minutes each time), it is determined that there is no risk of reignition, and the monitoring drone returns to standby status.

[0062] Continuous or periodic infrared scanning monitoring after fire extinguishing enables the early detection of hidden fire sources and timely handling of reignition risks, forming a complete closed loop of "fire extinguishing-monitoring-secondary treatment." This avoids fire rebound caused by reignition and further improves the thoroughness and safety of fire handling at photovoltaic power plants.

[0063] like Figure 2As shown in the illustration, this application also provides an intelligent fire suppression system for photovoltaic power plants based on a drone swarm, comprising: an acquisition component for controlling the drone swarm to automatically inspect the photovoltaic power plant according to a preset gridded route, and simultaneously acquiring sensor data from fixed monitoring points deployed in key equipment areas, fusing them to form comprehensive monitoring information for the entire power plant area; a discrimination component for identifying the location of the fire based on the comprehensive monitoring information for the entire power plant area, through a central control platform fusing and analyzing visible light images, infrared thermal imaging data, and environmental sensor data, and classifying the fire as a common equipment fire or a transformer insulating oil fire based on flame morphology, smoke color, and heat radiation distribution characteristics; and a planning component for planning based on the fire situation. The system intelligently assigns drones carrying matching types of extinguishing agents based on their type, location, scale, and the real-time status of each drone in the drone swarm, and plans the optimal flight path for each drone while balancing safety and efficiency. An execution component controls the assigned drones to reach the fire scene and executes corresponding firefighting strategies based on the fire type, including controlling drones to perform smothering spraying operations at a preset safe distance from the fire source for transformer insulating oil fires. A resupply component monitors the drone status in real time during or after firefighting operations, automatically scheduling the drones to return to the corresponding ground support station to sequentially or simultaneously replenish extinguishing agents and energy, supporting uninterrupted monitoring and operational cycles.

[0064] The optional embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the embodiments of the present invention are not limited to the specific details in the above embodiments. Within the scope of the technical concept of the embodiments of the present invention, various simple modifications can be made to the technical solutions of the embodiments of the present invention, and these simple modifications all fall within the protection scope of the embodiments of the present invention.

[0065] It should also be noted that the various specific technical features described in the above embodiments can be combined in any suitable manner without contradiction. To avoid unnecessary repetition, the embodiments of the present invention will not describe the various possible combinations separately.

[0066] Furthermore, various different implementations of the present invention can be combined arbitrarily, as long as they do not violate the spirit of the present invention, they should also be regarded as the content disclosed in the present invention.

Claims

1. A smart fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms, characterized in that, Includes the following steps: S1. Control the drone swarm to automatically inspect the photovoltaic power station according to the preset grid route, and at the same time acquire sensor data from fixed monitoring points deployed in key equipment areas, and integrate them to form the full-area monitoring information of the power station. S2. Based on the monitoring information of the entire station, the visible light image, infrared thermal imaging data and environmental sensor data are integrated and analyzed through the central control platform to identify the location of the fire, and the fire is classified as a general equipment fire or a transformer insulating oil fire based on the flame shape, smoke color and heat radiation distribution characteristics. S3. Based on the type, location, scale of the fire and the real-time status of each drone in the drone swarm, intelligently assign drones carrying matching types of fire extinguishing agents and plan the optimal flight path for the drones that balances safety and efficiency. S4. Control the assigned drone to fly to the fire scene and execute the corresponding fire extinguishing strategy according to the type of fire, including controlling the drone to carry out a covering spraying operation at a preset safe distance from the fire source for transformer insulating oil fire. S5. During or after firefighting operations, monitor the status of the UAV in real time and automatically dispatch the UAV back to the corresponding ground support station to complete the replenishment of fire extinguishing agent and energy in sequence or simultaneously, so as to support uninterrupted duty and operation cycle.

2. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, Step S1 includes: S11. The drone flies along a preset route and collects visual images and temperature data of the photovoltaic equipment in real time using the high-definition visible light camera and infrared thermal imager it is equipped with. S12. Simultaneously receive monitoring data from fixed temperature, smoke, and flame detectors deployed in key areas of the substation, transformer, and combiner box; S13. Receive wind speed, wind direction and humidity data collected by environmental sensors; S14. The visual image is fused with temperature data, monitoring data, and wind speed, wind direction, and humidity data, and then uploaded to the central intelligent control platform.

3. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, In step S2, classifying the fire into a general equipment fire or a transformer insulating oil fire based on flame shape, smoke color, and heat radiation distribution characteristics specifically includes: S21. Analyze the visible light image to identify whether there is a violent flame accompanied by thick black smoke; S22. Analyze infrared thermal imaging data to detect whether the abnormally high temperature area closely matches the physical outline of the transformer equipment and exhibits liquid flow characteristics; S23. If both intense black smoke and flowing high temperature with a specific outline are simultaneously present, it is determined to be a transformer insulating oil fire, triggering the highest level alarm; otherwise, it is determined to be a common equipment fire.

4. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, In step S3, planning the optimal flight path for the UAV that balances safety and efficiency specifically includes: S31. Obtain the precise coordinates of the fire site and, in conjunction with real-time wind speed and direction data, construct an initial candidate flight path set that takes into account the influence of airflow. S32. For transformer insulating oil fires, screen and eliminate routes that pass directly above the transformer body and the potential coverage area of ​​the oil drain pit. S33. Based on the real-time remaining battery power and payload of the assigned drone and the set of flight routes, construct an optimization model with the goal of the fastest safe arrival time, and dynamically solve for the optimal flight path.

5. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, In step S4, the specific implementation of the corresponding fire extinguishing strategy based on the fire type includes: S41. If it is a fire involving ordinary equipment, dispatch a water-carrying drone to fly directly above the fire point, maintain the optimal spraying height through laser ranging, and perform precise spraying at the designated point. S42. If the fire is caused by transformer insulating oil, a special fire extinguishing procedure shall be followed: S421. Dispatch drones carrying compressed air foam or dry powder fire extinguishing agents; S422. Control the drone to fly to the designated safe working area 10-20 meters away from the transformer body; S423. Control the drone to fly along a preset rectangular or spiral path covering the transformer body and the surface of the drain pit, while simultaneously activating the integrated foam generator to carry out large-area coverage spraying. S424. After the open flame is extinguished, instruct the drone to conduct periodic infrared scans at a safe distance until it is confirmed that there is no risk of reignition.

6. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, In step S5, automatically scheduling the UAV to return to the corresponding ground support station specifically includes: S51. Real-time monitoring of the remaining fire extinguishing agent and battery power of the drone; S52. When the remaining power or battery level is lower than the preset threshold, or when a fire-fighting mission completion signal is received, a return command containing the target support station type is automatically generated. S53. The drone flies to the corresponding smart water storage station or dedicated fire extinguishing agent supply station according to the instructions. S54. Employing a fusion of visual recognition and UWB precise positioning technology, it guides the UAV to complete an autonomous landing with centimeter-level precision on the docking platform of the resupply station.

7. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 6, characterized in that, In step S5, the sequential or simultaneous replenishment of extinguishing agent and energy specifically includes: S55. After the drone lands, the automatic docking robotic arm of the supply station completes a sealed connection with the quick-plug interface of its fire extinguishing agent storage tank. S56. Real-time monitoring of the net weight change of the UAV storage tank using a high-precision weighing sensor suspended below the docking mechanism; S57. The central control system presets the target weight based on the extinguishing agent requirements of the next task and sends control commands to the supply station. S58. The supply station opens the corresponding solenoid valve or pneumatic delivery valve according to the instruction, injects the extinguishing agent, and provides real-time feedback of weight data; wherein, when injecting dry powder or foam concentrate, the extinguishing agent is delivered through a special pipeline with a Teflon coating on the inner wall to reduce adhesion and residue. S59. When the real-time net weight reaches the target weight, the valve is automatically closed, the robotic arm is disconnected, and a precise refueling is completed.

8. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 5, characterized in that, In step S423 of the special fire extinguishing process, the large-area coverage spraying specifically includes: S423a. During the spraying operation, the current wind speed and wind direction data collected by the environmental sensor are acquired in real time; S423b: Based on the current wind speed and direction, the real-time hovering or flight position of the UAV, and the initial velocity and particle size distribution of the extinguishing agent, predict the dispersion trajectory and landing point distribution of the extinguishing agent in the air. S423c: Compare the predicted impact point distribution with the target coverage area. If the predicted impact point deviates from the target area by more than a preset threshold due to the influence of wind, dynamically adjust the horizontal position and / or spray angle of the UAV, or adjust the spray pressure to change the initial velocity of the extinguishing agent, so that the compensated predicted impact point can once again cover the target area.

9. The intelligent fire suppression method for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms according to claim 1, characterized in that, The method further includes step S6, which specifically includes: S61. After step S4 is completed, another drone equipped with an infrared thermal imager is assigned to conduct a continuous or periodic wide-area scan of the area where the fire has been extinguished. S62. Compare and analyze the temperature data obtained from the scan with historical normal data or data from the surrounding area; S63. If an abnormal temperature rise is detected, a secondary alarm will be automatically triggered, and nearby standby drones will be dispatched to handle the situation.

10. A smart fire suppression system for photovoltaic power plants based on unmanned aerial vehicle (UAV) swarms, characterized in that, include: The component is used to control a cluster of drones to automatically inspect photovoltaic power stations along a preset gridded route, while also acquiring sensor data from fixed monitoring points deployed in key equipment areas, and integrating them to form a comprehensive monitoring information for the entire power station. The discrimination component is used to identify the location of the fire based on the monitoring information of the entire station, and to integrate and analyze visible light images, infrared thermal imaging data and environmental sensor data through the central control platform. Based on the flame shape, smoke color and heat radiation distribution characteristics, the fire is classified as a general equipment fire or a transformer insulating oil fire. The planning component is used to intelligently assign drones carrying matching types of fire extinguishing agents based on the type, location, scale of the fire, and the real-time status of each drone in the drone swarm, and to plan the optimal flight path for the drones that balances safety and efficiency. The execution component is used to control the assigned drone to fly to the fire scene and execute the corresponding fire extinguishing strategy according to the type of fire, including controlling the drone to perform a covering spraying operation at a preset safe distance from the fire source for transformer insulating oil fire; The resupply component is used to monitor the status of the UAV in real time during or after firefighting operations, automatically dispatch the UAV to the corresponding ground support station, and complete the resupply of fire extinguishing agent and energy in sequence or simultaneously to support uninterrupted duty and operation cycle.

Images