Unmanned aerial vehicle‑based data collection method in forest fire scenario

Abstract

The present invention relates to the technical fields of unmanned aerial vehicles and the Internet of Things, and in particular to an unmanned aerial vehicle‑based data collection method in a forest fire scenario, comprising the steps: acquiring unmanned aerial vehicle data and node data, and establishing a regional model and a forest fire model; establishing constraint conditions on the basis of the unmanned aerial vehicle data, the node data, the regional model and the forest fire model; further establishing an optimization problem on the basis of the constraint conditions; solving the optimization problem by means of an unmanned aerial vehicle path planning algorithm based on a double deep Q network, and using optimal parameters obtained by means of the solving as initial parameters of the double deep Q network to obtain a trained model; and deploying the model onto the unmanned aerial vehicle, the model providing an action instruction for the unmanned aerial vehicle at a next moment on the basis of real-time data captured by the unmanned aerial vehicle, and the unmanned aerial vehicle executing a data collection task on the basis of the action instruction. The present invention enhances the rationality of decision making, avoids collection of invalid nodes by unmanned aerial vehicles, and improves resource utilization; by using a novel communication scheduling method, the data transmission efficiency is improved, and the communication stability is ensured.

Description

A method for collecting drone data in forest fire scenarios Technical Field

[0001] This invention relates to the fields of drone and Internet of Things (IoT) technology, and specifically to a method for drone data collection in forest fire scenarios. Background Technology

[0002] Forest fires pose a significant risk to forest resource protection. Firefighters face serious threats to their lives in every forest fire fighting operation. Timely situational awareness at the fire scene can minimize this threat.

[0003] The rapid and continuous development of wireless communication technology and sensor performance has made the deployment and use of wireless sensors simpler and more reliable, driving the advancement of the Internet of Things (IoT) and wireless sensor networks. By strategically deploying wireless sensors, specific data can be collected quickly and efficiently in specific areas. This enables on-demand data collection, adapts to dynamically changing and complex environments, and saves energy consumption for data transmission. These characteristics free drone-assisted IoT and wireless sensor networks from network connectivity limitations; the deployment of device nodes only needs to consider application requirements and the geographical structure of the target area.

[0004] Existing technologies, such as CN2024119045509A, disclose an efficient multi-UAV patrol method for low-altitude environments based on terrain elevation maps. This method considers multi-UAV collaboration, directly collecting forest fire scene information during each UAV's patrol, and uniformly planning the initial paths of multiple UAVs by maintaining a global patrol information map. During the planning process of a single UAV, environmental obstacle information is incorporated to obtain a safe path for that individual UAV. However, this technology has the following problems:

[0005] First, the data and resource utilization efficiency is low. The proposed solution mainly relies on current UAV detection data and limited prior information for analysis and decision-making, failing to fully explore and utilize a large amount of historical fire data and relevant data from different forest environments. This results in relatively weak adaptability and generalization ability when facing new fire scenarios or complex situations, making it difficult to make optimal decisions quickly and accurately. For example, different regions have significant differences in forest vegetation, topography, and climate conditions, and this solution may require readjusting parameters and models for each new scenario, leading to low efficiency.

[0006] 2. The impact of wind speed on drones was not considered: When drones are performing data collection tasks, wind speed will limit their flight.

[0007] Therefore, there is an urgent need for a method for collecting drone data in forest fire scenarios. Summary of the Invention

[0008] In view of this, the present invention discloses a method for collecting drone data in a forest fire scenario to solve the above problems; comprising:

[0009] S1. Acquire drone data and node data;

[0010] S2. Establish a regional model based on node data, use cellular automata to describe the forest fire spread process and establish a forest fire model;

[0011] S3. Establish UAV flight constraints based on UAV data; establish obstacle avoidance constraints based on the region model; establish communication constraints based on failure nodes, UAV data, and node data.

[0012] S4. Based on the constraints of UAV flight, obstacle avoidance, and communication, establish an optimization problem.

[0013] S5. The UAV path planning algorithm based on dual-depth Q-network solves the optimization problem, obtains the optimal parameters of the dual-depth Q-network, and uses the optimal parameters as the initial parameters of the path planning algorithm to obtain the trained UAV path planning model.

[0014] S6. Deploy the trained model onto the drone. The drone captures real-time data and inputs it into the trained model. The model provides the drone with action instructions for the next moment. The drone then performs the data collection task according to the action instructions.

[0015] The beneficial effects of this invention include:

[0016] By acquiring environmental wind speed data from ground nodes and incorporating this factor into the flight constraints of the UAV, the stability and adaptability of the UAV during actual mission execution were increased. By considering the uncertainty caused by fires in the lifecycle of ground nodes, the rationality of decision-making was enhanced, and the collection of invalid nodes by the UAV was avoided, thus improving resource utilization. By adopting a new communication scheduling method, data transmission efficiency was improved, and communication stability was ensured. Attached Figure Description

[0017] Figure 1 is a schematic diagram of the system model of the present invention;

[0018] Figure 2 is a comparison of the convergence process after smoothing under different algorithms of the present invention;

[0019] Figure 3 is a comparison chart of the data collection rate of the UAV in each mission under different algorithms of the present invention;

[0020] Figure 4 is a comparison of the success rates of UAV safe path planning under different algorithms of the present invention; Detailed Implementation

[0021] To make the objectives, technical solutions, features, and advantages of the present invention clearer, the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0022] Figure 1 shows a schematic diagram of the scenario corresponding to this embodiment. The method for collecting drone data in a forest fire scenario described in this embodiment includes the following steps:

[0023] S1. Acquire drone data and node data.

[0024] S2. Establish regional and forest fire models based on node data. The forest fire model is used to calculate failed nodes.

[0025] Specifically, wind is a significant factor at forest fire sites. Wind speed affects the spread of forest fires and also influences the air density around the rotor blades of drones, leading to changes in motor power. Ambient wind speed Represented as ,in This represents the data from the drone's wind speed sensor, where n represents the nth time slot obtained by evenly dividing the maximum mission time. The uncertainty of wind speed is determined by the difference between meteorological data and the actual wind speed. The larger the difference, the greater the uncertainty and the faster the wind changes. The maximum uncertainty of wind speed is... The larger the value, the more it is obtained from the ground node.

[0026] Furthermore, the forest fire model uses cellular automata to describe the forest fire spread process.

[0027] Specifically, cellular automata divide the task area into closely adjacent square grids, each grid being a cell. The attributes of a cell are determined by factors such as the spatial terrain, vegetation, weather conditions, ignition time, and combustion state within the cell. The state of a cell can be divided into five stages: unburned, burning, fully burning, not extinguished, and fully extinguished, denoted as follows: .

[0028] The initial rate of wildfire spread The formula is:

[0029]

[0030] in, For temperature, Wind force rating For relative humidity, a, b, c, and d are empirical parameters with preferred values ​​of 0.03, 0.05, 0.01, and -0.3, respectively. Further considering the influence of combustible material type, terrain slope, and wind vector on the initial spread rate, the corrected forest fire spread rate is obtained as follows: ,in, This is the corrected wildfire spread rate; It is the correction factor for combustible materials; This is the terrain slope correction factor. The effects of different slope gradients on the rate of wildfire spread were described. This is the wind vector correction factor. The effects of different wind speeds and wind directions on the spread rate are described.

[0031] Furthermore, the coordinate index of each cell is defined as follows: and Define a cell The rate of wildfire spread is The distance between a cell and its adjacent cells is the side length L of the cell. The state transition rule for a cell is: when an unburned cell is ignited, Become ; After internal combustion time Later converted to , Ignition in the direction parallel to the fire line within a time limit ,in It is the rate of forest fire spread within a cell. Indicates the side length of a cell; Ignite everything around it ,when The state of the cell does not exist around it Time conversion ,time consuming ; experience Enter after time State. Set This indicates the area that the forest fire has spread to, from the point of entry. and the subsequent state's cell composition, if the node coordinates exist In the middle, then the node This is a failed node.

[0032] S3. Establish UAV flight constraints based on UAV data; establish obstacle avoidance constraints based on the region model; establish communication constraints based on failure nodes, UAV data, and node data.

[0033] Specifically, regarding the flight constraints of the drone: the spatial position of the drone is:

[0034]

[0035] in, Let be the flight altitude of the drone, and let the projected coordinates of the drone on the flight plane be:

[0036]

[0037] Drone heading Furthermore, the system has K wireless sensor nodes, of which the... The position of each node is Limited by the performance of the drone itself, there are constraints. and Constrained by the influence of wind speed in the regional model, there are the following limitations: .

[0038] Specifically, regarding obstacle avoidance constraints: the mountain terrain is represented as follows:

[0039]

[0040] in, This represents the geographic coordinates, where z is the altitude of that location. The terrain correction parameters control the overall height of the i-th mountain. and Let be the attenuation of mountain i along the x-axis and y-axis, respectively, controlling the slope of the mountain. During drone flight, it needs to avoid both known terrain obstacles and unknown environmental obstacles; therefore, the region of obstacles in the drone's flight altitude plane is represented by a set. ,in Indicates the first An obstacle, Right now The set of points whose altitude is higher than the drone's flight altitude; the obstacle avoidance constraints for the drone are... .

[0041] Specifically, regarding communication constraints: communication constraints include scheduling constraints and data transmission constraints. In forest fire scenarios, the channel between the UAV and nodes is mainly affected by vegetation; therefore, this invention employs a path loss model. The formula is:

[0042]

[0043]

[0044] in, Indicates carrier frequency, Indicates the elevation angle of the drone relative to the ground node; This indicates that the shadow fading follows a Gaussian distribution; This represents the distance between the drone and node k; the drone's spatial position is... The position of the kth node is ; and These are the elevation angle influence factor and the path loss index, respectively.

[0045] Furthermore, the node transmit power is defined as... The noise power of the drone receiver is Signal-to-noise ratio at the drone receiver for:

[0046]

[0047] Furthermore, the communication rate between the drone and the node for:

[0048]

[0049] Where B represents the channel bandwidth.

[0050] Furthermore, in order to support stable data stream transmission, the received signal-to-noise ratio at the drone location needs to be [higher / lower]. The data collection task begins at that time, among which The signal-to-noise ratio threshold is set manually, preferably in the range of -5dB to -10dB. In this embodiment, it is preferably -5dB. The UAV collects data from a maximum of one node at a time. The connection between the drone and the k-th node is represented by the scheduling constraint as follows: .

[0051] Specifically, the UAV prioritizes data transmission tasks with nodes that have good and stable communication channels. The scheduling strategy formula is as follows:

[0052]

[0053] Among them, activity nodes Defined as signal-to-noise ratio The node, This indicates that the signal-to-noise ratio is below the threshold. Nodes and failed nodes.

[0054] Furthermore, Indicates in time slot Time from node Total amount of data collected, superscript Used to distinguish and The data transmission constraint obtained from the amount of data collected by the drone from node k is expressed as: The total amount of data that the drone can collect during the data collection process across all nodes is The maximum mission time is uniformly discretized into N time slots, such that the position of the UAV in each time slot can be approximated as constant. The length of the nth time slot is then determined. In this embodiment, the number of time slots N is preferably 300.

[0055] S4. Based on the constraints of UAV flight, obstacle avoidance, and communication, establish an optimization problem.

[0056] Specifically, to maximize the total amount of data quickly collected from all nodes, the optimization problem is as follows:

[0057]

[0058]

[0059]

[0060]

[0061]

[0062]

[0063]

[0064]

[0065]

[0066] in, To constrain the flight speed performance of drones, To constrain the turning performance of the drone, Constrained by the uncertainty of environmental wind force. To avoid obstacles and constraints, For scheduling constraints, For data transmission constraints, This indicates the task time constraint.

[0067] S5. The UAV path planning algorithm based on dual-depth Q-network solves the optimization problem and obtains the optimal parameters of the dual-depth Q-network. The optimal parameters are used as the initial parameters of the path planning algorithm to obtain the trained UAV path planning model.

[0068] Specifically, the sequential decision problem is expressed using Markov decision processes, i.e., formalized, and using tuples. The decision-making process is explained in the following terms:

[0069] Representing the state space, the environment information vector. ;in, This is the state information vector of the UAV. , The formula for expressing the flight speed of a drone is: , The ground speed of the drone in a windless environment. The discount factor for the influence of wind speed is preferably in the range of 0.005 to 0.01, and in this embodiment, it is preferably 0.01; Vector of surrounding environment information detected by the drone; ,in This indicates the distance between the drone and the obstacle as detected by the sensor in a certain direction. The sensor has a maximum detection distance limit. This is a global node information vector. ; The time consumed for task execution; furthermore, Represents a single node Information vector.

[0070] The actions of the intelligent agent are limited by the mechanical performance of the drone, and can only take executable actions sampled from the set of legal actions. It also defines safety rules to cancel and penalize drones when their next move would lead them toward an obstacle.

[0071] It represents the state transition in the entire system.

[0072] This represents the reward function set according to the optimization problem.

[0073] A dual-deep Q-network reinforcement learning algorithm is used to solve the problem. The following settings are used during the solution process:

[0074] Data collection reward is ,in This represents a data collection scaling factor, used to reduce the amount of data collected to a suitable range. In this embodiment... Preferably 10 -6 , Indicates the n+1 time slot from node Total amount of data collected Indicates the number of nodes.

[0075] The remaining task time reward is , This represents the time scaling factor, used to reduce the time amount to a suitable range. In this embodiment... The preferred value is 0.1.

[0076] Dynamic obstacle avoidance penalty is , This represents the obstacle avoidance penalty coefficient, used to amplify the shortest detection distance to a suitable range. In this embodiment... The preferred value is 10. This represents the minimum distance between the drone and the obstacle detected by the sensors in all directions. This indicates the maximum detection distance of the drone from the sensor.

[0077] For known obstacles, there is a static obstacle collision penalty. , This represents the collision penalty value, in this embodiment. The preferred value is 1.

[0078] To reduce invalid actions, the penalty for invalid actions is... , This represents the penalty value for an invalid action, in this embodiment. The preferred value is 1.

[0079] The global reward function is .

[0080] Furthermore, the initial parameters for solving the Markov decision process are set as follows:

[0081]

[0082] Furthermore, the optimization problem is solved to obtain the optimal parameters of the dual-depth Q-network. The solution steps include:

[0083] Step 1: Set the signal-to-noise ratio threshold Maximum task time Maximum flight speed of drones and maximum steering angle .

[0084] Step 2: Initialize the replay buffer D and estimate the network speed. The target network is initialized by replicating the estimated network parameters. .

[0085] Step 3, from Calculate the legal action space.

[0086] Step 4: Initialize the experimental environment by setting all environmental parameters to their initial values.

[0087] Step 5: Obtain environmental information vectors ,Will Input into the network to obtain actions The intelligent agent interacts with the environment and receives rewards. and new environmental information vectors .

[0088] Step 6: Plot the tuples Add to replay cache D.

[0089] Step 7: Sample a certain number of tuples from D and calculate the gradient using the loss function to update the network parameters. Further, the target Q-value function in the dual-depth Q-network is as follows:

[0090]

[0091] in, The condition for determining the final state is: all The corresponding node signal disappears, or the task execution time... Greater than the maximum task time constraint .

[0092] Step 8, each time The next update will Copy the parameters in middle, In this embodiment, the threshold for the number of training iterations of the neural network is set manually. The preferred value is 600.

[0093] Step 9, Judgment Is this the final state? If yes, proceed to step 4; otherwise, proceed to step 5.

[0094] Step 10: If the data collection rate and path planning success rate reach the expected values, stop the training loop, obtain the optimal neural network parameters, and set the optimal parameters as the initial values ​​of each parameter in the neural network to obtain the trained model. The expected value for the data collection rate is preferably 98%, and the expected value for the path planning success rate is preferably 96%.

[0095] Furthermore, the optimal parameters are used as the initial parameters for the path planning algorithm to obtain a trained UAV path planning model.

[0096] S6. Deploy the trained model onto the drone. The drone captures real-time data and inputs it into the trained model. The model provides the drone with action instructions for the next moment. The drone then performs the data collection task according to the action instructions.

[0097] Furthermore, as shown in Figure 2, which illustrates the convergence process of the proposed algorithm and the comparative algorithm after smoothing, the proposed algorithm is a UAV path planning algorithm based on a dual-depth Q-network, and the comparative algorithm is the Dueling DQN algorithm. It can be seen that the reward value of the Dueling DQN algorithm fluctuates drastically after increasing and does not reach its maximum; the proposed algorithm shows a significant performance improvement and exhibits smaller fluctuations during convergence, demonstrating stability in random environments.

[0098] As shown in Figure 3, the data collection rate of the UAV in each mission is shown under different algorithms. Affected by environmental factors and the fire spread process, the collection rate of the Dueling DQN algorithm hovered around 90% after an initial increase. The UAV path planning algorithm based on dual-depth Q network adopted in this invention reached 100% after an initial increase and then stabilized at a level close to 100%, demonstrating the effectiveness of the algorithm proposed in this invention in data collection tasks.

[0099] As shown in Figure 4, the success rate of UAV planning safe paths under different algorithms can be seen. It can be seen that the success rate of the comparison algorithm and the algorithm used in this invention can reach 80% in the early stage, while the path planning success rate of the Dueling DQN algorithm continues to decrease and fluctuates drastically; the success rate of the algorithm used in this invention converges to over 95%, demonstrating strong adaptability to the complex environment of forest fires.

[0100] Finally, it should be noted that the above description only depicts some embodiments of the present invention. For those skilled in the art, various changes, modifications, substitutions, and variations can be conceived of these embodiments without departing from the principles and spirit of the present invention. The scope of protection of the present invention is defined by the appended claims and their equivalents, and all the above-mentioned behaviors should be covered within the scope of protection of the present invention.

Claims

1. A method for collecting data from drones in a forest fire scenario, characterized in that, include: S1. Acquire drone data and node data; S2. Establish a regional model and a forest fire model based on node data. The forest fire model is used to calculate failed nodes. The forest fire model uses cellular automata to describe the forest fire spread process. The cellular automata includes states: not burning, burning, completely burning, not extinguished, and completely extinguished, respectively represented as follows: The transition rules between states are as follows: When unburned cells are ignited, Become ； After internal combustion time Later converted to ， Ignition in the direction parallel to the fire line within a time limit ,in It is the rate of forest fire spread within a cell. Indicates the side length of a cell; Ignite everything around it ,when The state of the cell does not exist around it Time conversion ,time consuming ； experience Enter after time state; Using sets This indicates the area that the forest fire has spread to, from the point of entry. and the subsequent state's cell composition, if the node coordinates exist In the middle, then the node This is a failed node; S3. Establish UAV flight constraints based on UAV data; establish obstacle avoidance constraints based on the region model. Establish communication constraints based on failed nodes, UAV data, and node data; S4. Based on the constraints of UAV flight, obstacle avoidance, and communication, establish an optimization problem. The established optimization problems include: UAV flight speed performance constraints, UAV turning performance constraints, environmental wind uncertainty constraints, obstacle avoidance constraints, scheduling constraints, data transmission constraints, and task time constraints. S5. The UAV path planning algorithm based on dual-depth Q-network solves the optimization problem, obtains the optimal parameters of the dual-depth Q-network, and uses the optimal parameters as the initial parameters of the dual-depth Q-network to obtain the trained UAV path planning model. S6. Deploy the trained model onto the drone. The drone captures real-time data and inputs it into the trained model. The model provides the drone with action instructions for the next moment. The drone then performs the data collection task according to the action instructions.

2. The method for collecting drone data in a forest fire scenario according to claim 1, characterized in that, The formula for the communication scheduling strategy used in the scheduling constraints is as follows: in, Indicates drones and the first The connection of each node Indicates the first Signal-to-noise ratio of each node, active nodes Defined as signal-to-noise ratio The node, Indicates the signal-to-noise ratio threshold. This indicates nodes with a signal-to-noise ratio below the threshold and failed nodes.

3. The method for collecting unmanned aerial vehicle (UAV) data in a forest fire scenario according to claim 2, characterized in that, Signal-to-noise ratio threshold The value range is -5dB to -10dB.

4. The method for collecting drone data in a forest fire scenario according to claim 1, characterized in that, Discount factor for wind speed effect The value range is 0.005 ~ 0.01.

Images