Fire extinguishing unmanned aerial vehicle cluster control method for new energy station fire emergency

Abstract

The invention provides a fire extinguishing unmanned aerial vehicle cluster control method for new energy station fire emergency, and the method comprises the steps: dynamically sensing the fire information of a new energy station through a three-dimensional fire scene model, and the fire information comprises the size of a fire, the position of an ignition point and meteorological information; based on the fire behavior size, the ignition point position and the meteorological information, multi-target optimization task distribution is realized for the unmanned aerial vehicle cluster; based on the distributed multi-objective optimization task, planning a three-dimensional path for the unmanned aerial vehicle cluster through a composite artificial potential field; a fire extinguishing instruction is sent to an unmanned aerial vehicle cluster by using a dual-mode redundancy transmission and distributed instruction storage mechanism, the unmanned aerial vehicle cluster is controlled to carry out fire extinguishing of different target optimization tasks according to a three-dimensional path, and fire extinguishing is carried out by using the three-dimensional path produced in a mode of a composite artificial potential field and a three-dimensional fire scene model. The unmanned aerial vehicle cluster fire extinguishing method closely fits the environment information of a new energy station, and compared with a pure unified control mode, the fire extinguishing efficiency of the unmanned aerial vehicle cluster is effectively improved in a fire extinguishing mode based on the distributed tasks.

Description

Technical Field

[0001] The present invention relates to the field of drone fire control technology, and in particular to a fire-fighting drone cluster control method for emergency fire response at new energy stations. Background Art

[0002] New energy stations are collections of renewable energy generation facilities, such as wind farms and solar power stations, that are centrally connected to the power system. These facilities include generators, transformers, energy storage devices, and other supporting equipment. Therefore, firefighting at these stations is particularly important. Current drone swarm firefighting methods rely solely on the simultaneous control of multiple drones, which fails to fully utilize the firefighting advantages of drone swarms.

[0003] Therefore, how to improve the fire-fighting efficiency of drone clusters has become a technical problem that technical personnel in this field urgently need to solve. Summary of the Invention

[0004] The present invention provides a fire-fighting drone cluster control method for emergency fire response at new energy stations, which is used to solve the defect of low fire-fighting efficiency of drone clusters in the prior art.

[0005] In a first aspect, the present invention provides a method for controlling a cluster of fire-fighting drones for fire emergencies at new energy stations, comprising: Through the three-dimensional fire scene model, dynamic perception of fire information at new energy stations, including fire size, fire location and meteorological information; Based on the fire size, fire location and meteorological information, multi-objective optimization task distribution is achieved for the drone cluster; Based on the distributed multi-objective optimization task, a three-dimensional path is planned for the UAV cluster through a composite artificial potential field; The dual-mode redundant transmission and distributed instruction storage mechanism are used to send fire-fighting instructions to the drone cluster, and the drone cluster is controlled to perform fire-fighting tasks with different target optimization tasks according to the three-dimensional path.

[0006] According to the present invention, a fire-fighting drone cluster control method for new energy station fire emergency is provided, which implements multi-objective optimization task distribution for the drone cluster based on the fire size, fire point location and meteorological information, including: Determining fire point value assessment parameters based on the fire size, fire point location and meteorological information; Determine the drone capability matrix in the drone swarm, which includes payload type, payload quantity, flight time, and current location; According to the size of the fire point value assessment parameter, a target optimization task is assigned to the UAV capability matrix.

[0007] According to a fire-fighting drone cluster control method for new energy station fire emergency response provided by the present invention, the fire point value assessment parameters are determined based on the fire size, fire point location and meteorological information, including: Determine the fire intensity based on the fire size, determine the equipment value based on the fire location, and determine the fire spread coefficient based on the meteorological information; A fire point value assessment parameter is determined based on the fire intensity, the equipment value and the fire spread coefficient.

[0008] According to a fire-fighting drone swarm control method for new energy station fire emergencies provided by the present invention, the method of determining the drone capability matrix in the drone swarm, which is composed of payload type, payload quantity, flight time, and current position, includes: The type of payload of each drone is determined through micro-electronic tags, and the number of payloads of each drone is determined through weight sensors; Use the drone endurance prediction model to predict the endurance of each drone under different conditions; The current position of each drone is determined through the positioning model; The drone capability matrix is ​​constructed by using rows to represent drones and columns to represent the payload type, payload quantity, flight time, and current location.

[0009] According to a fire-fighting drone cluster control method for new energy station fire emergencies provided by the present invention, the target optimization task is assigned to the drone capability matrix according to the size of the fire point value assessment parameter, including: Classify the fire point value assessment parameters into different levels and extract the fire characteristics of each level; Based on the fire characteristics of different levels, dynamic weights are assigned to the payload type, payload quantity, flight time and current position in the UAV capability matrix; The divided fire level and fire characteristics are used as environmental state input, and the parameters in the drone capability matrix are used as action space to output the drone capability matching strategy. The UAVs are scheduled based on the UAV capability matching strategy and task priorities.

[0010] According to a fire-fighting drone swarm control method for new energy station fire emergencies provided by the present invention, the distributed multi-objective optimization task plans a three-dimensional path for the drone swarm through a composite artificial potential field, including: Establish a new energy station model based on the location information of the new energy station and the information of the new energy power generation equipment; Construct a composite artificial potential field through the gravitational potential field and the repulsive potential field; Based on the new energy station model and the composite artificial potential field, the gradient descent method is used to perform three-dimensional path planning for the UAV cluster.

[0011] According to a fire-fighting drone swarm control method for new energy station fire emergencies provided by the present invention, the method uses a gradient descent method to perform three-dimensional path planning for the drone swarm based on the new energy station model and the composite artificial potential field, including: Initializing parameters of the new energy station model and the composite artificial potential field; After initialization is completed, the potential field force of each UAV in the UAV cluster is calculated in the new energy station model according to the composite artificial potential field to obtain the composite potential field force; Determine the gradient direction based on the composite potential field force; According to the preset step size and the gradient direction, the next position coordinates of the UAV are determined until a three-dimensional path is generated.

[0012] According to a fire-fighting drone cluster control method for new energy station fire emergency provided by the present invention, after generating a three-dimensional path, the method further includes: Optimize the three-dimensional path using path smoothing algorithm; The optimized three-dimensional path is further optimized based on the obstacles, endurance time and composite artificial potential field parameters in the new energy station.

[0013] In a second aspect, the present invention provides a fire-fighting drone cluster control device for fire emergencies at new energy stations, comprising: The perception module is used to dynamically perceive the fire information of the new energy station through the three-dimensional fire scene model, including the fire size, fire point location and meteorological information; A distribution module is used to implement multi-objective optimization task distribution for the UAV cluster based on the fire size, fire location and meteorological information; A planning module, configured to plan a three-dimensional path for the UAV cluster through a composite artificial potential field based on a distributed multi-objective optimization task; The fire extinguishing module is used to send fire extinguishing instructions to the drone cluster using dual-mode redundant transmission and distributed instruction storage mechanism, and control the drone cluster to perform fire extinguishing for different target optimization tasks according to the three-dimensional path.

[0014] In a third aspect, the present invention also provides an electronic device comprising a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, it implements any of the above-described methods for controlling a cluster of fire-fighting drones for fire emergencies in new energy stations.

[0015] In a fourth aspect, the present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, it implements any of the above-mentioned methods for controlling a cluster of fire-fighting drones for emergency fire response in new energy stations.

[0016] In a fifth aspect, the present invention also provides a computer program product, including a computer program, which, when executed by a processor, implements any of the above-mentioned methods for controlling a cluster of fire-fighting drones for fire emergencies in new energy stations.

[0017] The fire-fighting drone cluster control method for fire emergency response in new energy stations provided by the present invention dynamically senses the fire information of the new energy station through a three-dimensional fire scene model, including the fire size, fire point location and meteorological information; based on the fire size, fire point location and meteorological information, multi-objective optimization task distribution is implemented for the drone cluster; based on the distributed multi-objective optimization tasks, a three-dimensional path is planned for the drone cluster through a composite artificial potential field; a dual-mode redundant transmission and a distributed instruction storage mechanism are used to send fire-fighting instructions to the drone cluster, and the drone cluster is controlled to perform fire-fighting for different target optimization tasks according to the three-dimensional path. The three-dimensional path produced by the composite artificial potential field and the three-dimensional fire scene model is used to extinguish the fire, which is closely aligned with the environmental information of the new energy station. The fire-fighting method based on the assigned tasks effectively improves the fire-fighting efficiency of the drone cluster compared with a simple unified control method. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the technical solutions in the present invention or the prior art, a brief introduction is given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1 This is a flow chart of a method for controlling a cluster of fire-fighting drones for emergency fire response at a new energy station, as provided in this embodiment; Figure 2 This is a structural diagram of a fire-fighting drone cluster control device for new energy station fire emergencies provided by this embodiment; Figure 3 Schematic diagram of the structure of the electronic device provided in this embodiment. DETAILED DESCRIPTION

[0020] To make the objectives, technical solutions, and advantages of the present invention more clear, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0021] Figure 1 This is a flow chart of the fire-fighting drone cluster control method for new energy station fire emergencies provided in this embodiment.

[0022] like Figure 1 As shown, the fire-fighting drone cluster control method for new energy station fire emergency provided by the embodiment of the present invention mainly includes the following steps: 101. Through the three-dimensional fire scene model, the fire information of the new energy station can be dynamically perceived, including the fire size, fire point location and meteorological information.

[0023] In a specific implementation process, multiple types of sensors and equipment are deployed at new energy stations. Drones equipped with high-definition cameras, thermal imagers, and lidar are used to conduct full-scale scans of the station. Cameras and thermal imagers capture flame images and temperature distribution, while lidar obtains the three-dimensional coordinates of terrain and equipment. Weather stations collect real-time meteorological data such as wind speed, direction, temperature, humidity, and air pressure. Smoke sensors and temperature sensors are installed around key facilities to monitor local fire conditions.

[0024] The collected 3D terrain and equipment data is imported into the modeling software to construct a basic 3D model of the site. Based on the thermal imaging data, the fire area is marked in the model, with different colors and transparency levels used to distinguish the fire intensity. The fire point is located using LiDAR data and assigned precise 3D coordinates. Meteorological data is overlaid onto the model as a dynamic layer, such as using arrows to indicate wind direction and numerical values ​​to indicate wind speed, to form an initial 3D fire scene model.

[0025] Drones continuously patrol the area, transmitting new fire images and data in real time. Image recognition algorithms compare previous and subsequent data to update the fire's scope and direction of spread. Sensors monitor temperature and smoke concentration changes in real time, triggering parameter adjustments for corresponding areas in the model. Weather stations frequently transmit meteorological data, promptly refreshing the model's weather layers. Based on this dynamic data, the system automatically updates the 3D fire scene model, achieving real-time and accurate perception of fire information.

[0026] Therefore, the collected data can be input into the three-dimensional fire scene model to quickly and accurately output the corresponding fire size, fire point location and meteorological information.

[0027] 102. Based on the fire size, fire location and meteorological information, multi-objective optimization task distribution is achieved for drone clusters.

[0028] Specifically, task priorities are comprehensively assessed based on fire size, location, and meteorological information. Large, rapidly spreading fires near critical infrastructure, combined with unfavorable weather conditions (e.g., strong winds), are assigned high priority. Small fires far from critical areas, with stable weather conditions, are assigned low priority. A quantitative assessment model is established, assigning different weights to fire size, location, and meteorological information (e.g., 0.5, 0.3, and 0.2, respectively). Task priority scores are then calculated to clarify the order in which tasks are executed.

[0029] Utilize the established drone capability matrix and select suitable drones based on mission requirements. For high-priority missions, prioritize drones with large payloads, targeted payload types (such as dry powder fire extinguishing bombs for electrical fires), long flight times, and proximity to the fire. For low-priority missions, select drones with moderate payloads and closer locations. Also, consider the collaborative operation requirements of drone swarms and select drones with collaborative communication capabilities for mission execution.

[0030] Tasks are assigned based on the matching of mission priorities and drone capabilities. Mission instructions, including target location, firefighting requirements, and flight path planning, are sent to selected drones via the management platform. During mission execution, fire conditions and drone status are monitored in real time. If a new high-priority fire occurs, or if a drone performing the mission experiences a malfunction or low ammunition, a dynamic adjustment mechanism is immediately activated to reassess mission priorities and deploy other available drones to ensure efficient firefighting mission completion.

[0031] 103. Based on the distributed multi-objective optimization task, a three-dimensional path is planned for the UAV cluster through a composite artificial potential field.

[0032] Specifically, a new energy station model is first established based on the location information of the new energy station and its power generation equipment. High-precision LiDAR scanning and drone aerial photography are used to collect comprehensive data about the new energy station. LiDAR quickly captures the three-dimensional coordinates and shape of station equipment, buildings, and terrain, while aerial imagery provides additional texture and detail. Using point cloud processing algorithms and 3D modeling software, the collected data is converted into a high-precision 3D digital model. The model accurately annotates the location, size, and height of various facilities and terrain, including photovoltaic arrays, energy storage equipment, substations, transmission lines, buildings, hills, and ravines. A model database is also established, linking attribute information (such as traversability and hazard level) of various equipment and terrain types with the 3D model, providing foundational data for the subsequent construction of the artificial potential field.

[0033] Then, a composite artificial potential field is constructed using gravitational and repulsive potential fields. Gravitational Field Setup: Based on the multi-objective optimization task, a gravitational source is set for each mission target (e.g., a fire point). The magnitude of the gravitational force is inversely proportional to the distance from the drone to the target point, and is weighted based on the mission's urgency. For example, for high-risk fire point missions, the gravitational coefficient is increased to enable the drone to approach the target more quickly; for low-priority mission targets, the gravitational coefficient is appropriately reduced. The direction of gravity always points toward the target point. In three-dimensional space, the drone is subjected to the combined horizontal and vertical gravitational forces from the mission target point, guiding it toward the target. Repulsive Field Setup: Specific repulsive fields are set for various types of equipment and terrain within the new energy station. For impassable equipment (e.g., substations and buildings) and hazardous terrain (e.g., deep trenches and steep slopes), a strong repulsive field is set, with the repulsive force inversely proportional to the square of the distance from the drone to the obstacle, ensuring that the drone can sense and avoid it at a relatively long distance. For traversable but risky areas (such as narrow corridors between photovoltaic arrays), a weaker repulsive field is set to prevent the drone from deviating from the optimal path due to excessive avoidance. Furthermore, the system takes into account the operating status of equipment. When certain equipment is undergoing maintenance or malfunctioning, the repulsive force around it is dynamically increased. The repulsive force is directed along the normal line connecting the drone and the obstacle, acting both horizontally and vertically in three-dimensional space to keep the drone at a safe distance.

[0034] The gravitational field and the repulsive field are superimposed to form a composite artificial potential field. At each location in three-dimensional space, the net force acting on the drone is the vector sum of the gravitational and repulsive forces. By calculating the magnitude and direction of the net force at different locations, a potential field distribution map is constructed. Using visualization technology, the potential field distribution is displayed as a heat map on the management platform, allowing operators to intuitively understand the forces acting on the drone in different areas. Furthermore, to avoid singularities in the net force calculation (e.g., multiple repulsive sources causing confusion in the net force direction), a smoothing algorithm is used to optimize the net force, ensuring that the forces acting on the drone in the potential field are continuous and stable.

[0035] Finally, based on the new energy station model and the composite artificial potential field, the gradient descent method is used to perform three-dimensional path planning for the UAV cluster, including: Initialize the parameters of the new energy station model and the composite artificial potential field. Before using the gradient descent method for 3D path planning for a drone swarm, the relevant parameters must be initialized. Based on the new energy station model, determine the starting position coordinates of each drone. For example, accurate 3D coordinates (x0, y0, z0) can be obtained using a positioning base station within the station or a high-precision positioning module onboard the drone. Simultaneously, based on the composite artificial potential field model, determine the parameters required for potential field force calculation, such as the attraction coefficient and repulsion coefficient. These coefficients must be set based on the actual conditions of the new energy station and the mission requirements. For example, in areas with dense wind turbines, the repulsion coefficient can be appropriately increased to ensure drone safety. Furthermore, set the step size parameter for the gradient descent method. The step size determines the distance each drone moves. Excessively large step sizes may result in inaccurate paths, while too small a step size increases the computational complexity. Multiple experiments are required to find an appropriate value for the new energy station scenario.

[0036] After initialization, the composite artificial potential field is used to calculate the potential force of each drone in the swarm within the new energy station model. The composite potential field force is then calculated for each drone at its current position within the three-dimensional space of the new energy station using the composite artificial potential field model.

[0037] For the gravitational potential field, the equipment the drone needs to inspect (such as a solar panel in a solar array or the nacelle of a wind turbine) is used as the gravitational source. Based on the relationship between the gravitational force and the distance between the drone and the target point, combined with a set gravitational coefficient, the magnitude and direction of the gravitational force are calculated. For example, for a drone close to the target point, the gravitational force is relatively small but clearly pointed toward the target point. For a target point farther away and with a higher mission priority, the gravitational force is stronger due to the larger gravitational coefficient.

[0038] For the repulsive potential field, various obstacles within the new energy station (such as buildings, rotating wind turbine blades, and neatly arranged solar panel racks) are used as repulsive sources. Based on the rule that the repulsive force is inversely proportional to the square of the distance from the drone to the obstacle, combined with the repulsive coefficients set for each obstacle, the magnitude and direction of the repulsive force are calculated. For example, when approaching rotating wind turbine blades, the repulsive force will be significantly increased due to the high repulsive coefficient and close proximity, and the direction will be away from the blades. Finally, the attractive and repulsive forces are vectorially synthesized to obtain the composite potential field force at the drone's current location, which will guide the drone's movement.

[0039] Based on the composite potential force, the gradient direction is determined. The gradient direction is the direction of the function's fastest growth, and the gradient descent method moves in the direction of the function's fastest decline. Therefore, the direction of the gradient descent is opposite to the direction of the composite potential force. In the three-dimensional spatial coordinate system of the new energy station, the components of the composite potential force along the x, y, and z axes are used to determine the trend of the gradient descent direction in these three dimensions, thereby clarifying the general direction of the drone's next movement.

[0040] According to the preset step size and gradient direction, the next position coordinates of the UAV are determined until a three-dimensional path is generated. According to the set step size and the determined gradient descent direction, the next position coordinates of the UAV are calculated. In the three-dimensional spatial grid of the new energy station, the UAV is moved along the gradient descent direction by one step size to obtain the new position coordinates (x1, y1, z1). Check whether the new position exceeds the flight performance limits of the UAV, such as whether it exceeds the maximum flight altitude, whether it exceeds the maximum climb / descent angle, etc. If so, adjust the step size or recalculate the direction. At the same time, determine whether the new position is the target point or meets the mission completion conditions. If not, the new position is used as the current position, and the steps of potential field force calculation, gradient direction determination, and path iterative calculation are repeated to gradually generate the three-dimensional path of the UAV.

[0041] After obtaining a preliminary three-dimensional path through the gradient descent method, the path is optimized and verified. A path smoothing algorithm is used to address sharp corners in the path, making the drone's flight trajectory smoother and reducing energy loss and control difficulty during flight. Based on the actual situation of the new energy station, the path is checked to ensure that all obstacles are avoided and that the mission time requirements and the drone's endurance are met. If any problems exist, the path is replanned and optimized by adjusting potential field parameters (such as the gravitational coefficient and repulsive coefficient) or resetting the step size until a high-quality three-dimensional path that meets the operational requirements of the new energy station is generated.

[0042] 104. Use dual-mode redundant transmission and distributed instruction storage mechanism to send fire-fighting instructions to the drone cluster, and control the drone cluster to carry out fire-fighting tasks with different target optimization tasks along a three-dimensional path.

[0043] Specifically, dual-mode redundant transmission ensures command delivery: a dual-mode transmission channel combining wireless communication and satellite communication is established. During the initial stages of drone swarm operations, high-speed wireless communication networks (such as 5G) are prioritized to quickly transmit firefighting instructions (including three-dimensional path data, target mission information, etc.) to the drones. Meanwhile, the satellite communication link remains on standby. If the wireless communication signal is interfered with or interrupted, the system automatically switches to satellite communication mode to continue transmitting commands, ensuring uninterrupted command transmission. Both communication modes transmit the same command data, which the drone receives and verifies. If the dual-mode received data is consistent, the command is confirmed to be valid. In the event of inconsistency, the satellite communication data prevails, improving the accuracy and reliability of command transmission.

[0044] Distributed instruction storage enables autonomous execution. Each drone is equipped with a local storage module. Upon receiving a firefighting command, it immediately stores the command data (including mission objectives, 3D path nodes, flight parameters, and firefighting procedures) locally. Even if communication with the control center is completely interrupted, the drone can autonomously fly to the target along the preset 3D path based on the locally stored instructions and execute the firefighting mission. The drone also regularly performs hash checks on the locally stored instructions to prevent data errors or tampering during storage due to factors such as electromagnetic interference, ensuring stable command execution.

[0045] Command execution and dynamic adjustment. The drone flies toward the target along a three-dimensional path based on the received and stored commands. During flight, the drone uses onboard sensors to perceive its own status (such as position, speed, battery power, and payload) and environmental information (such as weather changes and the appearance of obstacles) in real time. If an emergency situation occurs (such as the appearance of a new obstacle ahead on the path), the drone uses its own path planning algorithm based on the locally stored emergency response strategy to replan the local path and continue to perform the mission. At the same time, the status changes and adjustments are fed back to the control center through the communication link. Based on the global information, the control center can choose to send new commands to the drone or allow it to make autonomous decisions, thereby realizing dynamic control and collaborative optimization of the drone cluster firefighting mission.

[0046] Furthermore, based on the above embodiment, this embodiment implements multi-objective optimization task distribution for drone clusters based on fire size, fire location, and meteorological information, including: First, based on the fire size, fire location, and meteorological information, determine the fire point value assessment parameters: Specifically, the fire intensity is determined based on the size of the fire, the equipment value is determined based on the location of the fire point, and the fire spread coefficient is determined based on meteorological information; the fire point value assessment parameters are determined based on the fire intensity, equipment value and fire spread coefficient.

[0047] Using the aforementioned data on fire size, location, and meteorological information as input parameters, a mathematical model is constructed. For example, a weighted summation approach can be used, assigning different weights to different parameters. For fire size, the weights for flame area are set to 0.4, energy release to 0.3, and fire spread rate to 0.3; the weights for the distance between the fire location and key facilities are set to 0.5, the weights for spatial location accuracy are set to 0.3, and the weights for the relationship with the surrounding environment are set to 0.2; and for meteorological information, the weights for wind speed are set to 0.4, wind direction to 0.3, temperature to 0.15, and humidity to 0.15, etc. This model can be used to calculate a comprehensive fire value assessment parameter value.

[0048] Risk Grading: Fire risk is categorized into different levels based on the calculated fire value assessment parameters. For example, a value between 0 and 30 indicates low risk, meaning the fire is relatively easy to control and poses a minimal threat to overall station safety. A value between 31 and 60 indicates medium risk, requiring prompt and effective firefighting measures to prevent the fire from spreading. A value between 61 and 100 indicates high risk, indicating a severe fire that could pose a significant threat to station facilities and personnel, necessitating immediate large-scale firefighting operations and emergency evacuation. This risk grading provides a clear reference for subsequent firefighting decisions.

[0049] Then, determine the drone capability matrix in the drone swarm, which consists of payload type, payload quantity, flight time, and current location: Specifically, each drone's payload type is determined using microelectronic tags, and its payload quantity is determined using weight sensors. Before a drone takes off, a microelectronic tag is attached to its ammunition, storing detailed information such as ammunition type, specifications, and applicable fire conditions. Once the drone is loaded with ammunition, an RFID (radio frequency identification) reader deployed in the station's ammunition storage area quickly reads the electronic tag and transmits the payload type data in real time to the drone's control system and the station's central management platform. Image recognition technology is also used to assist with confirmation. During the drone's ammunition loading process, a high-definition camera installed in the loading area captures the ammunition loading image. Image recognition algorithms analyze the ammunition's appearance and compare it with standard ammunition images in a database. This doubles down on ensuring the accuracy of the payload type information, avoids misjudgments of payload type, and ensures the reliability of this parameter in the capability matrix.

[0050] During the drone ammunition loading process, a combination of weighing sensors and intelligent algorithms is used to count the number of ammunition carried. High-precision weighing sensors are installed at the bottom of the loading platform, measuring the total weight of the drone after loading in real time. The system pre-enters the standard weight of each ammunition type. Once loaded, the algorithm calculates the actual number of ammunition carried based on the difference between the total weight and the drone's unladen weight, combined with the standard ammunition weight. For example, if a single fire extinguisher bomb weighs 5 kg, the drone weighs 20 kg unladen, and the total loaded weight is 50 kg, then the number of ammunition carried is (50 - 20) ÷ 5 = 6. Furthermore, a mechanical counting device can be used to assist with counting. A photoelectric counter is installed on the ammunition loading track. Each time an ammunition item passes, the counter records the number. The count result is then compared with the weighing result, further improving the accuracy of ammunition counts and integrating precise ammunition load data into the capability matrix.

[0051] Using a drone endurance prediction model, we estimate the flight time of each drone under different conditions. Specifically, the endurance assessment comprehensively considers multiple factors. First, a high-precision power sensor is integrated into the drone's battery management system to monitor the remaining battery charge in real time and transmit this data to the drone's control system and management platform. Simultaneously, a drone endurance prediction model is developed using historical flight data and machine learning algorithms. This model analyzes the drone's past power consumption under various flight speeds, altitudes, and payloads. Based on the drone's current flight parameters (such as speed, altitude, and payload weight), it predicts the flight time supported by the remaining battery charge. For example, analysis shows that when a drone flies at 20 km / h, an altitude of 100 meters, and a payload of 10 kg, it can fly for 15 minutes for every 10% of its battery consumed. If the current remaining battery charge is 60%, the predicted endurance is approximately 90 minutes. Furthermore, environmental factors that affect endurance are considered. For example, battery performance degrades in cold temperatures. The system then adjusts the endurance prediction based on real-time meteorological data to obtain accurate endurance parameters and improve the capability matrix.

[0052] Each drone's current location is determined through a positioning model. Specifically, multi-mode positioning technology is employed to ensure precise positioning of the drone's current position. Using the Beidou satellite navigation system as the primary navigation system and GPS as a supplement, the drones are equipped with dual-mode positioning modules that receive satellite signals in real time and obtain highly accurate latitude and longitude coordinates. This is supplemented by base station positioning. When a drone is in areas with weak satellite signals (such as near tall buildings or in complex terrain), it uses triangulation to estimate its position by interacting with surrounding communication base stations. Furthermore, UWB (ultra-wideband) positioning base stations are deployed in key areas within the station. When a drone enters these areas, it communicates with the UWB base stations at high speed, achieving centimeter-level positioning accuracy. These three positioning methods work together, automatically switching or fusing positioning data based on different scenarios. This allows the drone's real-time location information to be quickly and accurately fed back to the management platform, accurately populating the location parameters in the capability matrix.

[0053] A drone capability matrix is ​​constructed, with rows representing drones and columns representing payload type, payload quantity, flight time, and current location. Specifically, after obtaining data such as payload type, payload quantity, flight time, and current location, the management platform integrates this data in a unified format to construct the drone capability matrix. Each row in the matrix represents a drone, and each column corresponds to attributes such as payload type, payload quantity, flight time, and current location. For example, the first row contains capability information for drone 1, including its payload type (dry powder fire extinguisher), payload quantity (8), flight time (120 minutes), and current location coordinates (X1, Y1, Z1). The second row contains the corresponding data for drone 2, and so on. This capability matrix allows the management platform to intuitively display the capability status of each drone in the drone swarm, enabling commanders to quickly select drones suitable for firefighting tasks based on fire conditions, rationally plan the drone swarm's operational strategies, and achieve efficient firefighting operations.

[0054] Finally, according to the size of the fire point value evaluation parameters, the target optimization task is assigned to the UAV capability matrix: Specifically, fire point value assessment parameters are categorized into different levels, and fire characteristics for each level are extracted. Fire point value assessment parameters are divided into multiple, detailed levels based on risk, such as extremely high risk (90-100 points), high risk (70-89 points), medium risk (40-69 points), and low risk (0-39 points). Unique characteristics are extracted for each level of fire point. For example, extremely high-risk fire points typically exhibit rapid fire spread, proximity to critical infrastructure, and unfavorable weather conditions for firefighting. Low-risk fire points may exhibit small fires, be located far from critical areas, and experience stable weather conditions. A fire point feature database is also established, recording typical parameter combinations for fire points of different levels in terms of fire size, location, and weather information, providing data support for subsequent task matching.

[0055] Based on the characteristics of different fire levels, dynamic weights are assigned to payload type, payload quantity, flight time, and current location in the drone capability matrix. For different fire levels, dynamic weights are assigned to payload type, payload quantity, flight time, and current location. For example, for extremely high-risk fires, due to the need to quickly control the spread of the fire, the weights for payload quantity and payload type are increased to 40% and 30%, respectively, emphasizing that the drone carries sufficient and effective firefighting ammunition. The weight for flight time is adjusted to 15%, and the weight for current location is 15%. For low-risk fires, considering mission flexibility and efficient resource utilization, the weight for current location is increased to 35%, facilitating the dispatch of drones to the nearest location. The weight for payload quantity is set to 25%, the weight for payload type is 20%, and the weight for flight time is 20%. Dynamic weighting is adjusted based on the fire risk level and the actual contribution of each capability element to the firefighting mission. A machine learning algorithm is used to continuously optimize the weighting model to better align with actual operational needs.

[0056] A reinforcement learning-based task-drone capability matching algorithm is constructed, using the classified fire level and characteristics as environmental inputs and the parameters in the drone capability matrix as the action space. The algorithm uses the fire point level and characteristics as environmental inputs and the parameters in the drone capability matrix as the action space. Through trial and error and learning, the algorithm assigns rewards or penalties based on the results of each task (such as firefighting efficiency, task completion time, and resource consumption), gradually exploring the optimal drone capability matching strategy for different fire point levels. For example, when encountering a high-risk fire point, the algorithm prioritizes drones with large payloads, payload types suitable for the fire situation, and close proximity. For low-risk fire points, it prefers drones that are closer and have sufficient flight time to achieve reasonable resource allocation and efficient task completion.

[0057] Drone scheduling is based on a drone capability matching strategy and task priorities. Task priority is determined based on the fire point value assessment parameters. Extremely high-risk fire points receive the highest priority, requiring immediate deployment of the most capable drones. Low-risk fire points receive lower priority, allowing for flexible resource allocation. Game theory is incorporated into resource scheduling to construct a drone swarm task allocation game model. Each drone, acting as a participant in the game, maximizes overall firefighting effectiveness through strategic selection while meeting its own mission requirements. For example, when multiple drones are assigned to the same fire point, the system uses a game model to calculate the benefits and costs of each drone's mission and prioritizes the drone with the highest overall benefit. The system also considers inter-drone collaboration to avoid resource waste and task conflicts.

[0058] During a drone firefighting mission, changes in fire point value assessment parameters (e.g., fire expansion due to sudden changes in weather conditions) and the status of the drone's capability matrix (e.g., drone payload depletion and reduced flight time) are monitored in real time. Based on this real-time data, an adaptive dynamic programming algorithm is used to reassess the match between mission requirements and drone capabilities. If the currently executing drone cannot meet the changed mission requirements, the system immediately activates a task reallocation mechanism, selects new suitable drones from the drone capability matrix, and adjusts the task allocation plan to ensure the continued and efficient progress of the firefighting mission. Furthermore, actual data from the mission execution is fed back into the model training process to continuously optimize fire point value assessment, capability matrix weight allocation, and task matching algorithms, thereby improving the adaptability and decision-making accuracy of the entire system.

[0059] Based on the same general inventive concept, the present invention also protects a fire-fighting drone cluster control device for fire emergencies in new energy stations. The fire-fighting drone cluster control device for fire emergencies in new energy stations described below and the fire-fighting drone cluster control method for fire emergencies in new energy stations described above can be referenced to each other.

[0060] Figure 2 This is a structural diagram of the fire-fighting drone cluster control device for new energy station fire emergencies provided in this embodiment.

[0061] like Figure 2 As shown, this embodiment provides a fire-fighting drone cluster control device for new energy station fire emergency, including: The sensing module 201 is used to dynamically sense the fire information of the new energy station through the three-dimensional fire scene model, including the fire size, fire point location and meteorological information; The distribution module 202 is used to implement multi-objective optimization task distribution for the UAV cluster based on the fire size, fire location and meteorological information; Planning module 203, for planning a three-dimensional path for the UAV cluster through a composite artificial potential field based on a distributed multi-objective optimization task; The fire extinguishing module 204 is used to send fire extinguishing instructions to the drone cluster using dual-mode redundant transmission and distributed instruction storage mechanism, and control the drone cluster to perform fire extinguishing tasks with different target optimization tasks according to a three-dimensional path.

[0062] Figure 3 Schematic diagram of the structure of the electronic device provided in this embodiment.

[0063] like Figure 3 As shown, the electronic device may include: a processor (processor) 310, a communication interface (Communications Interface) 320, a memory (memory) 330 and a communication bus 340, wherein the processor 310, the communication interface 320, and the memory 330 communicate with each other through the communication bus 340. The processor 310 can call the logic instructions in the memory 330 to execute a fire-fighting drone cluster control method for new energy station fire emergency, the method comprising: dynamically sensing the fire information of the new energy station through a three-dimensional fire scene model, including the fire size, the location of the fire point, and the meteorological information; based on the fire size, the location of the fire point, and the meteorological information, implementing multi-objective optimization task distribution for the drone cluster; based on the distributed multi-objective optimization tasks, planning a three-dimensional path for the drone cluster through a composite artificial potential field; using dual-mode redundant transmission and a distributed instruction storage mechanism to send fire-fighting instructions to the drone cluster, and controlling the drone cluster to perform fire extinguishing tasks of different objectives according to the three-dimensional path.

[0064] Furthermore, the logic instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the portion that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product, stored in a storage medium, includes several instructions for enabling a computer device (which can be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0065] On the other hand, the present invention also provides a computer program product, which includes a computer program, which can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the fire-fighting drone cluster control method for new energy station fire emergency provided by the above methods. The method includes: dynamically sensing the fire information of the new energy station through a three-dimensional fire scene model, including the fire size, fire point location and meteorological information; based on the fire size, fire point location and meteorological information, implementing multi-objective optimization task distribution for the drone cluster; based on the distributed multi-objective optimization tasks, planning a three-dimensional path for the drone cluster through a composite artificial potential field; using dual-mode redundant transmission and distributed instruction storage mechanism to send fire-fighting instructions to the drone cluster, and controlling the drone cluster to perform fire extinguishing of different target optimization tasks according to the three-dimensional path.

[0066] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a fire-fighting drone cluster control method for fire emergencies in new energy stations provided by the above-mentioned methods. The method includes: dynamically sensing the fire information of the new energy station through a three-dimensional fire scene model, including the size of the fire, the location of the fire point and meteorological information; based on the size of the fire, the location of the fire point and meteorological information, implementing multi-objective optimization task distribution for the drone cluster; based on the distributed multi-objective optimization tasks, planning a three-dimensional path for the drone cluster through a composite artificial potential field; using dual-mode redundant transmission and a distributed instruction storage mechanism to send fire-fighting instructions to the drone cluster, and controlling the drone cluster to perform fire extinguishing for different target optimization tasks according to the three-dimensional path.

[0067] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, i.e., they may be located in one location or distributed across multiple network units. Some or all of the modules may be selected based on actual needs to achieve the objectives of the present embodiment. Persons of ordinary skill in the art will be able to understand and implement the present invention without inventive effort.

[0068] Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented using software plus a necessary general-purpose hardware platform, or of course, hardware. Based on this understanding, the essence of the above technical solution, or the portion that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a magnetic disk, or an optical disk, and includes a number of instructions for causing a computer device (such as a personal computer, server, or network device) to execute the methods described in each embodiment or certain portions of the embodiments.

[0069] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A fire-fighting drone cluster control method for new energy station fire emergency response, characterized in that: include: Through the three-dimensional fire scene model, dynamic perception of fire information at new energy stations, including fire size, fire location and meteorological information; Based on the fire size, fire location and meteorological information, multi-objective optimization task distribution is achieved for the drone cluster; Based on the distributed multi-objective optimization task, a three-dimensional path is planned for the UAV cluster through a composite artificial potential field; The dual-mode redundant transmission and distributed instruction storage mechanism are used to send fire-fighting instructions to the drone cluster, and the drone cluster is controlled to perform fire-fighting tasks with different target optimization tasks according to the three-dimensional path.

2. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 1 is characterized in that: The multi-objective optimization task distribution for the UAV cluster based on the fire size, fire location and meteorological information includes: Determining fire point value assessment parameters based on the fire size, fire point location and meteorological information; Determine the drone capability matrix in the drone swarm, which includes payload type, payload quantity, flight time, and current location; According to the size of the fire point value assessment parameter, a target optimization task is assigned to the UAV capability matrix.

3. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 2 is characterized in that: The fire point value assessment parameters are determined based on the fire size, fire point location and meteorological information, including: Determine the fire intensity based on the fire size, determine the equipment value based on the fire location, and determine the fire spread coefficient based on the meteorological information; A fire point value assessment parameter is determined based on the fire intensity, the equipment value and the fire spread coefficient.

4. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 2 is characterized in that: Determining the drone capability matrix in the drone swarm, which includes payload type, payload quantity, flight time, and current location, includes: The type of payload of each drone is determined through micro-electronic tags, and the number of payloads of each drone is determined through weight sensors; Use the drone endurance prediction model to predict the endurance of each drone under different conditions; The current position of each drone is determined through the positioning model; The drone capability matrix is ​​constructed by using rows to represent drones and columns to represent the payload type, payload quantity, flight time, and current location.

5. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 2 is characterized in that: The step of allocating target optimization tasks to the UAV capability matrix according to the size of the fire point value assessment parameter includes: Classify the fire point value assessment parameters into different levels and extract the fire characteristics of each level; Based on the fire characteristics of different levels, dynamic weights are assigned to the payload type, payload quantity, flight time and current position in the UAV capability matrix; The divided fire level and fire characteristics are used as environmental state input, and the parameters in the drone capability matrix are used as action space to output the drone capability matching strategy. The UAVs are scheduled based on the UAV capability matching strategy and task priorities.

6. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 1 is characterized in that: The distribution-based multi-objective optimization task plans a three-dimensional path for the UAV cluster through a composite artificial potential field, including: Establish a new energy station model based on the location information of the new energy station and the information of the new energy power generation equipment; Construct a composite artificial potential field through the gravitational potential field and the repulsive potential field; Based on the new energy station model and the composite artificial potential field, the gradient descent method is used to perform three-dimensional path planning for the UAV cluster.

7. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 6 is characterized in that: The method of using a gradient descent method to perform three-dimensional path planning for a UAV cluster based on the new energy station model and the composite artificial potential field includes: Initializing parameters of the new energy station model and the composite artificial potential field; After initialization is completed, the potential field force of each UAV in the UAV cluster is calculated in the new energy station model according to the composite artificial potential field to obtain the composite potential field force; Determine the gradient direction based on the composite potential field force; According to the preset step size and the gradient direction, the next position coordinates of the UAV are determined until a three-dimensional path is generated.

8. The fire-fighting drone cluster control method for new energy station fire emergency according to claim 7 is characterized in that: After generating the three-dimensional path, the method further includes: Optimize the three-dimensional path using path smoothing algorithm; The optimized three-dimensional path is further optimized based on the obstacles, endurance time and composite artificial potential field parameters in the new energy station.

9. A fire-fighting drone cluster control device for new energy station fire emergency response, characterized in that: include: The perception module is used to dynamically perceive the fire information of the new energy station through the three-dimensional fire scene model, including the fire size, fire point location and meteorological information; A distribution module is used to implement multi-objective optimization task distribution for the UAV cluster based on the fire size, fire location and meteorological information; A planning module, configured to plan a three-dimensional path for the UAV cluster through a composite artificial potential field based on a distributed multi-objective optimization task; The fire extinguishing module is used to send fire extinguishing instructions to the drone cluster using dual-mode redundant transmission and distributed instruction storage mechanism, and control the drone cluster to perform fire extinguishing for different target optimization tasks according to the three-dimensional path.

10. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, it implements the fire-fighting drone cluster control method for new energy station fire emergency as described in any one of claims 1 to 8.

Images