A forest fire intervention based unmanned aerial vehicle path planning system

Abstract

The application discloses a kind of unmanned vehicle path planning systems based on forest fire intervention, comprising: data acquisition module, environmental data is collected, environmental data is transmitted to cloud computing module after airborne edge preprocessing;Cloud computing module, fire source is identified using improved YOLOv8 model, and rescue path is optimized using IAO-CFOA fusion algorithm;Intelligent control module receives the rescue path of cloud computing module, uses real-time differential positioning technology to correct the trajectory of unmanned vehicle, and integrates wind speed compensation algorithm, carries out fire rescue;Ecological evaluation module, based on the evaluation model constructed, uses Monte Carlo simulation to evaluate the ecological benefits after fire rescue.The application uses the above system, effectively overcomes the limitations of traditional unmanned vehicle path planning in dynamic wild fire environment using IAO-CFOA fusion algorithm, improves response speed and flight stability, and further quantifies the ecological benefits of unmanned vehicle intervention system in different regions.

Description

Technical Field

[0001] This invention relates to the field of path planning technology, and in particular to a drone path planning system based on forest fire intervention. Background Technology

[0002] Timely and precise intervention in the early stages of a fire is crucial to containing its spread, but existing traditional methods such as satellite monitoring and ground patrols suffer from limitations in response time and coverage. Aerial monitoring, currently the most efficient means of forest fire monitoring, primarily relies on aircraft or drones. However, traditional path planning algorithms (such as A / B) are limited in their effectiveness. Algorithms (such as Dijkstra's algorithm) are ill-suited to the dynamic and unpredictable nature of wildfire spread. Furthermore, existing systems lack real-time closed-loop control and quantitative assessment of ecological benefits. Summary of the Invention

[0003] The purpose of this invention is to provide a drone path planning system based on forest fire intervention to solve the problems mentioned in the background art.

[0004] To achieve the above objectives, the present invention provides a drone path planning system based on forest fire intervention, including a data acquisition module, a cloud computing module, an intelligent control module, and an ecological assessment module;

[0005] The data acquisition module uses a drone equipped with multiple sensor components to collect environmental data such as fire source coordinates, wind speed, and wind direction. After the environmental data is preprocessed by the airborne edge computing module, it is transmitted to the cloud computing module.

[0006] The cloud computing module uses the improved YOLOv8 model to identify fire sources, and based on the identified fire sources, it uses the IAO-CFOA fusion algorithm to optimize the rescue path.

[0007] The intelligent control module receives the rescue path from the cloud computing module, uses real-time differential positioning technology to correct the drone trajectory, and integrates a wind speed compensation algorithm for fire rescue.

[0008] The ecological assessment module, based on the constructed assessment model, uses Monte Carlo simulation to evaluate the ecological benefits after fire rescue and quantify net carbon emission reduction.

[0009] Preferably, the multi-sensor components include a wide-angle camera, a lidar, and an infrared thermometer.

[0010] Preferably, the cloud computing module includes a terrain construction unit, a fire source identification unit, and a path planning unit;

[0011] The fire source identification unit uses the improved YOLOv8 model to identify fire sources and outputs a set of fire source coordinates. ;

[0012] The path planning unit is based on the set of fire source coordinates output by the fire source identification unit. The final rescue route is generated through optimization using the IAO-CFOA fusion algorithm. ;

[0013] The terrain construction unit is based on MATLAB to build a digital twin model, integrates GIS terrain data, constructs a terrain model, and deploys rescue routes.

[0014] Preferably, the YOLOv8 model improvement process includes:

[0015] Add deformable convolutional modules to the backbone network to dynamically adjust the receptive field of the convolutional kernels, using the following formula:

[0016] ;

[0017] in, Indicates the input feature map, Indicates the convolution kernel weights, Indicates the standard sampling location. Indicates the position offset;

[0018] A convolutional attention module is embedded in the neck layer of the network, and its attention weights are calculated as follows:

[0019] ;

[0020] in, yes No. 1 eigenvector This represents the scaling function. Indicates the total number of features;

[0021] Introducing the HIoU loss function, the HIoU loss function for bounding box regression is:

[0022] ;

[0023] in, It is the intersection, and it is a small constant. and These are predicted bounding boxes and ground truth bounding boxes. It is Euclidean distance. Indicates bounding box The length of the diagonal.

[0024] Preferably, the route planning unit optimizes the route using the IAO-CFOA fusion algorithm to generate the final rescue route. The process includes:

[0025] Based on the set of fire source coordinates output by the fire source identification unit Based on real-time environmental data, the IAO algorithm continuously adjusts path points through a weighted fusion model to generate an initial path solution. ;

[0026] Fire source coordinate set output by the fire source identification unit Based on the UAV performance parameters, the CFOA algorithm is used to calculate compensation path solutions for dynamic environments and mission energy consumption. ;

[0027] The IAO-CFOA fusion algorithm is used to jointly optimize the two algorithms and generate the final rescue path. The formula for the synergistic effect of the IAO-CFOA fusion algorithm is as follows:

[0028] ;

[0029] in, It is the synergy coefficient that balances the two algorithms.

[0030] Preferably, the weight of the k-th sensor in the multi-sensor input data dynamically fused by the weight fusion model is:

[0031] ;

[0032] in, Indicates the first The measurement variance of each sensor, Indicates the first The measurement variance of each sensor, It refers to the number of sensors;

[0033] Fusion The parameters after that are:

[0034] ;

[0035] in, As measured by the kth sensor, For the first Sensor weights.

[0036] Preferably, the CFOA algorithm is used to calculate the compensation path solution for dynamic environments and task energy consumption. for:

[0037] Define the fitness function, and the fitness function formula is:

[0038] ;

[0039] in, Indicates the task time. This indicates energy consumption, used to evaluate and filter pathways. , The weighting coefficients are used to update the position of the solution in CFOA. ;

[0040] During the iteration process, the solution is updated using the position update formula, which is:

[0041] ;

[0042] in, As a tendency factor, it controls the intensity with which an individual learns towards the current optimal solution. As a disturbance factor, it controls the strength of random disturbances; for example, the random disturbance term. , For the updated path solution, For the current generation's path solution, This is the optimal path solution found in the entire population.

[0043] Through continuous iterative updates, a compensation path solution is generated. .

[0044] Preferably, the wind speed compensation algorithm includes flight speed correction and material conveying displacement correction;

[0045] The formula for calculating the corrected flight speed is as follows:

[0046] ;

[0047] in, It is the expected speed. Wind speed;

[0048] The formula for calculating the material conveying displacement correction is as follows:

[0049] ;

[0050] in, The time for release to the target.

[0051] Preferably, the construction of the evaluation model includes:

[0052] The formula for setting the prevention and control efficiency is:

[0053] ;

[0054] in, To ensure accuracy, For dynamic correction coefficients, These are the discrete coefficients;

[0055] The formula for calculating carbon loss due to fire is as follows:

[0056] ;

[0057] in, This is the fire area. For carbon dioxide conversion rate, Carbon loss rate;

[0058] The carbon sequestration gain achieved by setting preventative measures is given by the following formula:

[0059] ;

[0060] in, For the sequestration rate, For the recycling period, Represents the discrete coefficients;

[0061] The net emission reduction is set using the following formula:

[0062] .

[0063] Preferably, the assessment of the ecological benefits after fire relief using Monte Carlo simulation includes:

[0064] Set the fire zone extent, carbon density, and number of simulations;

[0065] The ecological benefits after fire rescue are assessed based on the evaluation model. For N simulations, the mean value is calculated. and standard deviation And set a 95% confidence interval, the formula is:

[0066] ;

[0067] ;

[0068] .

[0069] Therefore, the present invention employs the above-mentioned UAV path planning system based on forest fire intervention, which has the following beneficial effects:

[0070] (1) The IAO algorithm is used to improve the dynamic fusion of multi-sensor data, enhance the accuracy of environmental perception, and the CFOA algorithm is used to compensate for path calculation in real time to reduce transmission error. The IAO-CFOA fusion algorithm effectively overcomes the limitations of traditional UAV path planning algorithms in dynamic wildfire environments. The coupling strategy is better than traditional algorithms, and high fire source coverage is achieved in dynamic scenarios, improving response speed and flight stability.

[0071] (2) It can adjust the drone path in real time, making up for the lack of dynamic feedback in traditional systems;

[0072] (3) Monte Carlo simulation was used to assess the ecological benefits after fire rescue and quantify the net carbon emission reduction.

[0073] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0074] Figure 1 This is a flowchart of the method system according to an embodiment of the present invention;

[0075] Figure 2 This is a system response diagram according to an embodiment of the present invention;

[0076] Figure 3 This is a diagram of the improved YOLOv8 model framework according to an embodiment of the present invention;

[0077] Figure 4 The two-dimensional path planning visualization flight trajectory of the UAV swarm in this embodiment of the invention;

[0078] Figure 5 The three-dimensional path visualization flight trajectory of the drone swarm in this embodiment of the invention;

[0079] Figure 6 This is a comparison diagram of the improved algorithm and the baseline algorithm in a dynamic fire scene environment according to an embodiment of the present invention;

[0080] Figure 7 The following is a probability distribution map of net emission reduction of the environmental system in an embodiment of the present invention: (a) Northern forest, (b) Southern forest;

[0081] Figure 8 The following is a diagram showing the relationship between the disaster prevention efficiency of the environmental system and the net emission reduction in an embodiment of the present invention: (a) Northern forests; (b) Southern forests. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. In the description of the present invention, it should be noted that the terms "upper," "lower," "inner," "outer," etc., indicating orientation or positional relationships are based on the orientation or positional relationships shown in the accompanying drawings, or the orientation or positional relationships commonly used when the product of the invention is in use. They are only for the convenience of describing the present invention and simplifying the description, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the present invention.

[0083] Example

[0084] Reference Figures 1-3 This invention provides a drone path planning system based on forest fire intervention, including a data acquisition module, a cloud computing module, an intelligent control module, and an ecological assessment module.

[0085] The data acquisition module uses a drone equipped with multiple sensor components, including a wide-angle camera, lidar, and infrared thermometer, to collect environmental data such as fire source coordinates, wind speed, and wind direction. After onboard edge preprocessing, the environmental data is transmitted to the cloud computing module via the MAVLink protocol.

[0086] In this embodiment, the UAV adopts a novel tiltrotor electric vertical takeoff and landing (EVL) aircraft equipped with a distributed electric propulsion system. It features a conventional layout with high tail volume and strong static stability margin, combined with a high aspect ratio rectangular wing (AR=10) and a V-tail design, aiming to provide excellent pitch stability and directional control efficiency in complex flow fields. The main wing uses the NACA6412 airfoil, which exhibits a high lift coefficient and good stall characteristics at low and medium speeds, with a stall angle of attack of 13.7°, ensuring that it is not easily stalled under strong disturbances. The V-tail uses the NACA0012 airfoil, reducing structural weight and simplifying the control surface configuration through aerodynamic coupling, thus improving the system's response reliability in turbulent flow.

[0087] The cloud computing module uses an improved YOLOv8 model to identify fire sources, and based on the identified fire sources, it uses the IAO-CFOA fusion algorithm to optimize the rescue path.

[0088] In this embodiment, the cloud computing module includes a terrain construction unit, a fire source identification unit, and a path planning unit.

[0089] The fire source identification unit uses an improved YOLOv8 model to identify fire sources and outputs a set of fire source coordinates. The YOLOv8 model improvement process includes:

[0090] Adding deformable convolutional modules to the backbone network dynamically adjusts the receptive field of the convolutional kernels, enabling the model to adaptively focus on feature regions based on the actual shape of different fire sources. This solves the problem of insufficient feature extraction for diverse fire sources, directly improving fire source detection accuracy, especially significantly enhancing the identification capability of small fire sources. The operating formula is:

[0091] ;

[0092] in, Indicates the input feature map, Indicates the convolution kernel weights, Indicates the standard sampling location. Indicates the position offset;

[0093] A convolutional attention module is embedded in the neck layer of the network, comprising two sub-modules: channel attention and spatial attention. Channel attention uses "average pooling + max pooling" on the channel dimensions of the feature map to select the channel features more critical for fire source identification and weaken irrelevant channels. Spatial attention focuses on the spatial location of the fire source through convolution, suppressing spatial interference from background noise such as smoke and building shadows. The attention weights are calculated as follows:

[0094] ;

[0095] in, yes No. 1 eigenvector This represents the scaling function. Indicates the total number of features;

[0096] Introducing the HIoU loss function, the HIoU loss function for bounding box regression is:

[0097] ;

[0098] in, It is the intersection, and it is a small constant. and These are predicted bounding boxes and ground truth bounding boxes. It is Euclidean distance. Indicates bounding box The length of the diagonal.

[0099] The path planning unit is based on the set of fire source coordinates output by the fire source identification unit. The final rescue route is generated through optimization using the IAO-CFOA fusion algorithm. Specifically:

[0100] Input the set of fire source coordinates output by the fire source identification unit into the IAO algorithm. With real-time environmental data (such as obstacle locations and wind speed), the IAO algorithm leverages its information perception advantage to continuously adjust path points through a weighted fusion model, generating an initial path solution. The solution is a sequence of path points, represented as , of which each It is a three-dimensional coordinate point. Among them, the weighted fusion model dynamically fuses the multi-sensor input data... The weights of each sensor are:

[0101] ;

[0102] in, Indicates the first The measurement variance of each sensor, Indicates the first The measurement variance of each sensor, It refers to the number of sensors;

[0103] Fusion The parameters after that are:

[0104] ;

[0105] in, As the first Each sensor measures, For the first Sensor weights.

[0106] The CFOA algorithm takes the set of fire source coordinates output by the fire source identification unit as input. Based on the UAV performance parameters, the CFOA algorithm calculates compensation path solutions for dynamic environments and mission energy consumption. Specifically:

[0107] First, define the fitness function. The fitness function formula is:

[0108] ;

[0109] in, Indicates the task time. This indicates energy consumption, used to evaluate and filter pathways. , The weighting coefficients are used to update the position of the solution in CFOA. ;

[0110] Secondly, the solution is iterated. During the iteration, the position update formula is used to update the solution. The position update formula is:

[0111] ;

[0112] in, As a tendency factor, it controls the intensity with which an individual learns towards the current optimal solution. As a disturbance factor, it controls the strength of random disturbances; for example, the random disturbance term. , For the updated path solution, For the current generation's path solution, This is the optimal path solution found in the entire population.

[0113] By continuously iterating and updating, we find solutions with lower fitness and generate compensation path solutions. .

[0114] The IAO-CFOA fusion algorithm is used to jointly optimize the two algorithms and generate the final rescue path. The formula for the synergistic effect of the IAO-CFOA fusion algorithm is as follows:

[0115] ;

[0116] in, It is the synergy coefficient that balances the two algorithms.

[0117] The terrain construction unit uses MATLAB to build a digital twin model, integrates GIS terrain data, constructs a terrain model, and deploys rescue routes.

[0118] The intelligent control module receives the rescue path from the cloud computing module, uses real-time differential positioning technology to correct the drone's trajectory, and integrates a wind speed compensation algorithm for fire rescue. The wind speed compensation algorithm includes flight speed correction and material conveying displacement correction.

[0119] The formula for calculating corrected flight speed is:

[0120] ;

[0121] in, It is the expected speed. Wind speed;

[0122] The formula for calculating the material conveying displacement correction is as follows:

[0123] ;

[0124] in, The time for release to the target.

[0125] The ecological assessment module, based on the constructed assessment model, uses Monte Carlo simulation to evaluate the ecological benefits after fire relief and quantify net carbon emission reductions.

[0126] The evaluation model includes:

[0127] The formula for setting the prevention and control efficiency is:

[0128] ;

[0129] in, To ensure accuracy, , For dynamic correction coefficients, , For discrete coefficients, ;

[0130] The formula for calculating carbon loss due to fire is as follows:

[0131] ;

[0132] in, This is the fire area. For carbon dioxide conversion rate, , Carbon loss rate;

[0133] The carbon sequestration gain achieved by setting preventative measures is given by the following formula:

[0134] ;

[0135] in, The sequestration rate is 3 t CO2 / ha / year in the north and 7.5 t CO2 / ha / year in the south. The recycling period is 40 years in the north and 15 years in the south. These represent the discrete coefficients, used to reflect the uncertainty and volatility during the system's execution.

[0136] The net emission reduction is set using the following formula:

[0137] .

[0138] The use of Monte Carlo simulations to assess the ecological benefits of fire relief efforts includes:

[0139] Define the fire zone Carbon density and number of simulations The fire area is defined as follows: The carbon density follows a normal distribution. , This represents the average area of ​​the fire. Indicates the standard deviation of the fire area. This represents the mean carbon density. This represents the standard deviation of carbon density.

[0140] The ecological benefits after fire rescue are assessed based on the evaluation model. For N simulations, the mean value is calculated. and standard deviation And set a 95% confidence interval, the formula is:

[0141] ;

[0142] ;

[0143] .

[0144] Example 1

[0145] To verify the effectiveness of the fusion algorithm in UAV path planning, multiple fire scenarios were constructed for testing. Figure 4 and Figure 5 The flight trajectories of the drone swarm in the experiment were shown.

[0146] To quantitatively evaluate the performance of different path planning strategies in dynamic fire environments, traditional A... The core differences between the algorithm, Information Acquisition Optimizer (IAO), Fishing Optimization Algorithm (CFOA), and the IAO-CFOA fusion algorithm are discussed. Experiments were conducted in a simulated complex fire environment that integrates dynamically spreading fire sources, random obstacles, and non-uniform wind interference.

[0147] From the basic performance analysis of path planning, A While the paths generated by the algorithm possess structural stability, their path length and task completion time are the longest among all compared algorithms, especially exhibiting sluggish performance in complex scenarios such as forest and industrial fires. The algorithm's inherent grid search mode struggles to effectively adapt to dynamic changes in the fire scene and lacks real-time response capabilities to sudden fires and obstacles. The IAO algorithm alone significantly shortens task completion time through dynamic information filtering and adaptive adjustment mechanisms. However, due to its limitations in group collaboration and global optimization, the total length of the paths planned by IAO is not optimal. In contrast, the CFOA algorithm alone demonstrates superior global exploration capabilities, but its convergence speed relies on population iteration and random search, thus creating a bottleneck in computational efficiency, resulting in a slightly slower task completion time compared to the IAO algorithm.

[0148] The IAO-CFOA fusion algorithm forms a complementary and collaborative mechanism. The IAO component, acting as the front-end perception and rapid response unit, processes environmental information in real time and makes preliminary dynamic adjustments to the path, effectively avoiding invalid search intervals that CFOA might encounter. The CFOA component, acting as the back-end optimization engine, performs in-depth development and fine-tuning based on the high-quality solution space provided by IAO, ultimately generating a global path that is superior in both path length and energy efficiency. Experimental data shows that this fusion strategy achieves optimal levels in both total path length and task completion time, achieving a balance between planning efficiency and solution quality.

[0149] The group collaboration mechanism within the fusion algorithm avoids path conflicts and uneven resource allocation that may occur with multiple drones, thereby improving task execution efficiency at the system level. For example... Figure 6 As shown, the UAV swarm using the IAO-CFOA fusion algorithm for collaborative path planning exhibits significantly higher firefighting support efficiency than other comparative groups. The IAO-CFOA fusion algorithm enables the overall planning of swarm collaborative behavior, allowing multiple UAVs to execute tasks in a highly coordinated manner, significantly shortening the overall firefighting time and improving task completion.

[0150] Example 2

[0151] Using Monte Carlo simulation on traditional A The proposed fusion algorithm was used to simulate the planning paths of the algorithm, the Information Acquisition Optimizer (IAO), the Fishing Optimization Algorithm (CFOA), and the IAO-CFOA fusion algorithm. The Monte Carlo simulation data are shown in Table 1. The prevention efficiency of the proposed fusion algorithm is significantly higher than that of the baseline method, with an overall improvement of 11%-30%.

[0152] The classification of northern and southern forest regions in the table is based on the "Classification of Forest Fire Risk Zones in China" standard issued by the National Forestry and Grassland Administration. The northern forest region is represented by the temperate natural forest region of Northeast China, while the southern forest region is represented by the subtropical planted forest regions of Southwest and South China. Forest restoration year data are taken from the "China's Forest and Grassland Resources and Ecological Status" published by the National Forestry and Grassland Administration. The restoration cycle of northern natural forests is approximately 40 years, influenced by vegetation type, climate conditions, and ecological restoration rate. Southern planted forests have a relatively simpler community structure and higher restoration efficiency, with a restoration cycle of approximately 15 years.

[0153] The mean fire area is a statistical result of the average burned area of ​​a single fire in the northern and southern forest areas over the past 10 years. The standard deviation is the fluctuation range of the fire area scenario simulated based on the IAO-CFOA fusion algorithm, combined with factors such as the complexity of forest terrain, vegetation flammability, fire spread patterns, and spatial heterogeneity of historical fires. It is a quantitative characterization of the uncertainty of fire spread. The prevention efficiency is the statistical mean of Monte Carlo simulations under a specified number of experiments, calculated using the indicator function method. The calculation formula is as follows:

[0154] ;

[0155] In the formula, Indicates the critical time for a fire to spiral out of control. Represented as the first The response time of the fusion algorithm in this simulation This is an indicator function.

[0156] The carbon emission factor is scientifically derived based on the carbon emission accounting standards in the IPCC report "2021 Climate Change: The Natural Science Basis". The accounting formula is as follows:

[0157] ;

[0158] in, This is the baseline data for forest carbon density in the North and South Forest Regions published by the IPCC. For the forest restoration cycle in the forest area, The fire spread area coefficient is 0.9 for natural forests and 1.2 for planted forests. CF=3.667 is the standard conversion coefficient from carbon to carbon dioxide. The standard deviation of the index is a comprehensive correction value that combines the spatial variation of regional carbon density, the fluctuation of fire scenarios, and the differences in vegetation types.

[0159] The carbon sequestration rate was derived using the IPCC-recommended baseline value for annual carbon sequestration rates in East Asian temperate and subtropical forests, and was further localized by considering the climate characteristics, vegetation growth rate, stand structure, and ecosystem carbon sequestration capacity of the northern and southern forest regions. Among these, southern plantations have shorter growth cycles and higher carbon sequestration efficiency, resulting in higher carbon sequestration rates than northern natural forests. The standard deviation of this indicator is a quantitative result of the differences in carbon sequestration capacity under different stand types and site conditions within the region.

[0160] Simulation parameters include the fire zone extent. (Unit: hectares), Regional carbon density (Unit: tons of carbon per hectare) and number of simulations (For example Second-rate).

[0161] Fire area The fire boundary can be calculated by combining the fire scene boundary detected by drones with GIS topographic data. In the simulation, it can be set as a random variable that follows a normal distribution, for example... , indicating an average area of ​​1000 hectares and a standard deviation of 200 hectares.

[0162] carbon density Based on forest type (north / south), for example, northern forest is... Tons of carbon per hectare, Southern forests are Tons of carbon per hectare.

[0163] The ecological benefits of fire rescue are assessed using an evaluation model, where the average net emission reduction is calculated for N simulations. and standard deviation And set a 95% confidence interval, the formula is:

[0164] Confidence interval = .

[0165] Table 1 Monte Carlo simulation data

[0166]

[0167] Although the disaster prevention effectiveness of forests in the south is significantly inferior to that in the north, this is mainly due to the predominantly planted forests in the south—areas prone to fires and with relatively high carbon exchange rates. Therefore, a multi-machine collaborative model should be adopted in southern forest areas to improve disaster prevention efficiency. Figure 7 This shows the distribution of net emission reduction probability of forest systems in different regions of China. Figure 8 This reveals the relationship between the disaster prevention efficiency of the environmental system and net emission reduction.

[0168] Therefore, this invention employs the aforementioned UAV path planning system based on forest fire intervention. It utilizes the IAO algorithm to improve the dynamic fusion of multi-sensor data, enhancing environmental perception accuracy, and uses the CFOA algorithm to compensate for path calculation errors in real time, reducing transmission errors. The IAO-CFOA fusion algorithm effectively overcomes the limitations of traditional UAV path planning algorithms in dynamic wildfire environments. The coupling strategy is superior to traditional algorithms, achieving high fire source coverage in dynamic scenarios, improving response speed and flight stability. At the same time, it further quantifies the ecological benefits of UAV intervention systems in different regions, providing a supplement to the dynamic research on forest carbon sequestration.

[0169] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the technical solutions of the present invention, and these modifications or equivalent substitutions cannot cause the modified technical solutions to deviate from the spirit and scope of the technical solutions of the present invention.

Claims

1. A drone path planning system based on forest fire intervention, characterized in that: It includes a data acquisition module, a cloud computing module, an intelligent control module, and an ecological assessment module; The data acquisition module uses a drone equipped with multiple sensor components to collect environmental data such as fire source coordinates, wind speed, and wind direction. After the environmental data is preprocessed by the airborne edge computing module, it is transmitted to the cloud computing module. The cloud computing module uses the improved YOLOv8 model to identify fire sources, and based on the identified fire sources, it uses the IAO-CFOA fusion algorithm to optimize the rescue path. The intelligent control module receives the rescue path from the cloud computing module, uses real-time differential positioning technology to correct the drone trajectory, and integrates a wind speed compensation algorithm for fire rescue. The ecological assessment module, based on the constructed assessment model, uses Monte Carlo simulation to evaluate the ecological benefits after fire rescue and quantifies net carbon emission reduction. The wind speed compensation algorithm includes flight speed correction and material conveying displacement correction; The formula for calculating corrected flight speed is: ； in, It is the expected speed. Wind speed; The formula for calculating the material conveying displacement correction is as follows: ； in, The time for release to the target; The evaluation model includes: The formula for setting the prevention and control efficiency is: ； in, To ensure accuracy, For dynamic correction coefficients, These are the discrete coefficients; The formula for calculating carbon loss due to fire is as follows: ； in, The fire area For carbon dioxide conversion rate, Carbon loss rate; The carbon sequestration gain achieved by setting preventative measures is given by the following formula: ； in, For the sequestration rate, For the recycling period, Represents the discrete coefficients; The net emission reduction is set using the following formula: 。 2. The UAV path planning system based on forest fire intervention according to claim 1, characterized in that: The multi-sensor components include a wide-angle camera, a lidar, and an infrared thermometer.

3. The UAV path planning system based on forest fire intervention according to claim 1, characterized in that: The cloud computing module includes a terrain construction unit, a fire source identification unit, and a path planning unit; The fire source identification unit uses the improved YOLOv8 model to identify fire sources and outputs a set of fire source coordinates. ; The path planning unit is based on the set of fire source coordinates output by the fire source identification unit. The final rescue route is generated through optimization using the IAO-CFOA fusion algorithm. ; The terrain construction unit is based on MATLAB to build a digital twin model, integrates GIS terrain data, constructs a terrain model, and deploys rescue routes.

4. The UAV path planning system based on forest fire intervention according to claim 3, characterized in that: The YOLOv8 model improvement process includes: Add deformable convolutional modules to the backbone network to dynamically adjust the receptive field of the convolutional kernels, using the following formula: ； in, Indicates the input feature map, Indicates the convolution kernel weights, Indicates the standard sampling location. Indicates the position offset; A convolutional attention module is embedded in the neck layer of the network, and its attention weights are calculated as follows: ； in, yes No. 1 eigenvector This represents the scaling function. Indicates the total number of features; Introducing the HIoU loss function, the HIoU loss function for bounding box regression is: ； in, It is the intersection, and it is a small constant. and These are predicted bounding boxes and ground truth bounding boxes. It is Euclidean distance. Indicates bounding box The length of the diagonal.

5. The UAV path planning system based on forest fire intervention according to claim 3, characterized in that: The route planning unit optimizes the route using the IAO-CFOA fusion algorithm to generate the final rescue route. The process includes: Based on the set of fire source coordinates output by the fire source identification unit Based on real-time environmental data, the IAO algorithm continuously adjusts path points through a weighted fusion model to generate an initial path solution. ; Fire source coordinate set output by the fire source identification unit Based on the UAV performance parameters, the CFOA algorithm is used to calculate compensation path solutions for dynamic environments and mission energy consumption. ; The IAO-CFOA fusion algorithm is used to jointly optimize the two algorithms and generate the final rescue path. The formula for the synergistic effect of the IAO-CFOA fusion algorithm is as follows: ； in, It is the synergy coefficient that balances the two algorithms.

6. The UAV path planning system based on forest fire intervention according to claim 5, characterized in that, The weighted fusion model dynamically fuses the input data from multiple sensors. The weights of each sensor are: ； in, Indicates the first Measurement variance of each sensor, Indicates the first Measurement variance of each sensor, It refers to the number of sensors; Fusion The parameters after that are: ； in, As the first Each sensor measures, For the first Sensor weights.

7. The UAV path planning system based on forest fire intervention according to claim 5, characterized in that: The CFOA algorithm is used to calculate the compensation path solution for dynamic environments and task energy consumption. for: Define the fitness function, and the fitness function formula is: ； in, Indicates the task time. This indicates energy consumption, used to evaluate and filter pathways. , The weighting coefficients are used to update the position of the solution in CFOA. ; During the iteration process, the solution is updated using the position update formula, which is: ； in, As a tendency factor, it controls the intensity with which an individual learns towards the current optimal solution. The disturbance factor controls the strength of the random disturbance; the random disturbance term... , For the updated path solution, For the current generation's path solution, This is the optimal path solution found in the entire population. Through continuous iterative updates, a compensation path solution is generated. .

8. The UAV path planning system based on forest fire intervention according to claim 1, characterized in that, The assessment of the ecological benefits following fire relief using Monte Carlo simulation includes: Set the fire zone extent, carbon density, and number of simulations; The ecological benefits after fire rescue are assessed based on the evaluation model. For N simulations, the mean value is calculated. and standard deviation And set a 95% confidence interval, the formula is: ； ； 。

Images