A forest fire mobile node-oriented obstacle avoidance unmanned aerial vehicle data collection method

Abstract

The present application relates to unmanned aerial vehicle and internet of things technical field, especially in a kind of forest fire facing mobile node's data collection method of barrier unmanned aerial vehicle, comprising: the data of ground mobile node and the data of unmanned aerial vehicle node are obtained, and system model is established according to node data;According to system model, optimization problem is constructed;Deep reinforcement learning model is used to solve optimization problem;Intelligent agent for planning barrier path is obtained according to the result of solving;Intelligent agent is deployed to unmanned aerial vehicle, and intelligent agent gives the action instruction of next time according to the environmental data captured by unmanned aerial vehicle in real time, and unmanned aerial vehicle executes data collection task according to action instruction;The present application quantitatively analyzes the unit data value of mobile node according to forest fire spread, and more efficient data collection is realized;By establishing the safety distance constraint between unmanned aerial vehicle, the dynamic barrier problem with unmanned aerial vehicle is solved.

Description

Technical Field

[0001] This invention relates to the fields of drone and Internet of Things (IoT) technology, and specifically to a method for collecting obstacle avoidance data from a drone targeting mobile nodes during forest fires. Background Technology

[0002] At the scene of a forest fire, the safety of firefighters and the real-time changes in the spread of the fire are of paramount importance. Traditional data collection methods, such as wired network connections, multi-hop transmission via wireless network nodes, or reliance on fixed base stations, have numerous limitations. Wired connections are costly to deploy and cannot adapt to the dynamic changes of mobile nodes; multi-hop transmission via wireless network nodes consumes a large amount of energy, and mobile nodes also cause frequent changes in network topology; fixed base stations are limited by coverage area and cannot achieve full coverage in some remote or terrain-complex areas. Furthermore, when mobile nodes are frequently moving, traditional methods cannot guarantee stable data transmission, leading to data loss or delays.

[0003] The rapid development of crowdsourced sensing and the Internet of Things has made it possible to deploy a large number of mobile sensors at fire sites. These data collection nodes, benefiting from their mobility, can survive for extended periods within the fire zone, continuously collecting complete and constantly changing data. This provides a wealth of environmental information about forest fire scenes and vital sign information of firefighters, thus playing an irreplaceable role in ensuring the safety of personnel at forest fire sites. Using drones to assist in data collection further solves the problems of data collection and transmission, avoiding the energy consumption and random performance degradation associated with traditional multi-hop transmission.

[0004] However, the mobility of nodes means that the relative positions of different nodes to the fire site change during data collection tasks, and the value of their data also changes accordingly. At the same time, moving aerial nodes and other non-cooperative drones add dynamic obstacles to the flight environment.

[0005] The varying value of data from different nodes and the dynamic obstacle avoidance problem of drones present more challenges to drone-assisted data collection tasks. Summary of the Invention

[0006] In view of this, the present invention discloses a method for collecting obstacle avoidance data from a drone oriented towards moving nodes during forest fires, in order to solve the above-mentioned problems; including the following steps:

[0007] S1. Acquire data from ground mobile nodes and UAV nodes, and establish a system model based on the data from ground mobile nodes and UAV nodes;

[0008] S2. Construct an optimization problem based on the system model;

[0009] S3. Solve the optimization problem using a deep reinforcement learning model; obtain an agent for planning obstacle avoidance paths based on the solution results;

[0010] S4. Deploy the intelligent agent onto the drone. The drone acquires real-time environmental data. The intelligent agent provides the drone with action instructions for the next moment based on the real-time environmental data. The drone then performs data collection tasks according to the action instructions.

[0011] The beneficial effects of this invention include: by establishing a mobile node model, quantitative analysis of the unit data value of mobile nodes based on the spread of forest fires is achieved, enabling more efficient data collection, covering a wider range of data sources in a short time, greatly improving data integrity, and providing a solid and reliable data foundation for subsequent in-depth analysis and decision-making; by establishing safe distance constraints between drones based on the state vector relationship between the mission drone, node drones, and non-cooperative drones, the dynamic obstacle avoidance problem between drones is solved. Attached Figure Description

[0012] Figure 1 This is a flowchart illustrating the obstacle avoidance UAV data collection method of the present invention;

[0013] Figure 2 This is a schematic diagram of the system architecture in the obstacle avoidance UAV data collection method of the present invention;

[0014] Figure 3 This is a comparison chart of the convergence of reward values ​​of the agent during the training process under different algorithms in this embodiment of the invention;

[0015] Figure 4 This is a comparison chart of the total data collection rate of the mission UAV under different algorithms in the embodiments of the present invention;

[0016] Figure 5 This is a comparison chart of the data collection rates of the UAV to aerial nodes under different algorithms in this embodiment of the invention;

[0017] Figure 6 This is a comparison chart of the data collection rates of the mission UAV to ground mobile nodes under different algorithms in this embodiment of the invention;

[0018] Figure 7 This is a comparison chart of the data collection rates of high-value nodes by the mission drones under different algorithms in this embodiment of the invention;

[0019] Figure 8 This is a comparison chart of the success rates of safe trajectory planning for mission drones under different algorithms in embodiments of the present invention. Detailed Implementation

[0020] 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.

[0021] This embodiment includes a method for collecting obstacle avoidance data from a drone targeting mobile nodes during forest fires. The flowchart of the method is as follows: Figure 1 As shown, the steps include:

[0022] S1. Acquire data from ground mobile nodes and drone nodes, and establish a system model based on the data from ground mobile nodes and drone nodes.

[0023] Furthermore, the system model includes system architecture, fire zone model, mobile node model, and data collection task model.

[0024] Specifically, the system architecture is described as follows: Figure 2 As shown, the mission drone maintains a safe flight altitude, and its spatial position is represented as q[n] = (x[n], y[n], h u ), where h u Let q be the flight altitude of the drone. The projected coordinates of the drone on the flight plane are q. u [n] = (x[n], y[n]); the drone's ground velocity is v[n] = (v x [n],v y [n]); the drone's heading is Furthermore, the state vector of the mission UAV is

[0025] Furthermore, the system contains K wireless sensor nodes, which are divided into fast-moving aerial nodes and slow-moving ground nodes. The fast-moving aerial nodes include K1 drone nodes performing data collection tasks at the fire scene, while the slow-moving ground nodes include K2 unmanned vehicles and K3 firefighters on site. K = K1 + K2 + K3. The number of drones performing other tasks is denoted as L. These drones performing other tasks do not have direct scheduling communication with the task drones in this system, thus hindering their flight. When these drones approach the task drones, the task drones obtain basic information about these non-cooperative drones through onboard sensors. The state vectors of other non-cooperative drones at the fire scene are... q l [n] = (x l ,y l ) indicates the location of the non-cooperative drone, v l [n] represents the speed of the non-cooperative drone. Indicates the heading of a non-cooperative drone.

[0026] Specifically, the fire zone model is explained as follows: At a forest fire site, wind speed affects the spread of the fire and also the flight speed of drones. Ambient wind speed v w [n] is 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, and Δv w [n] represents the magnitude of wind speed uncertainty, 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 greater the potential value. Furthermore, the flight speed of the drone is v[n] = v u [n]+λv w [n], where v u [n] represents the ground velocity of the drone in a windless environment, λ is the discount factor for the effect of wind speed, and v u [n] represents the ground speed of the UAV in a windless environment, and λ is the discount factor for the influence of wind speed, preferably ranging from 0.005 to 0.01, with 0.01 being preferred in this embodiment. Due to limitations in the UAV's own performance, there is a constraint ||v u [n]||2≤v max , and

[0027] Furthermore, the impact of mountains and vegetation on topography is expressed as follows: Where (x, y) represents the xy-plane coordinates, z is the elevation of the location, and h i Let x represent the total height of the i-th mountain. i With y i The slope of the mountain along the x-axis and y-axis is controlled separately.

[0028] Furthermore, since the UAV needs to avoid known terrain obstacles and unknown environmental obstacles during flight, the regions of various obstacles in the UAV's flight altitude plane can be represented as a set D = {D1, D2, ... D}. j}, where D j Let D represent the j-th obstacle. j That is, z(x,y)>h u The set of points whose altitude is higher than the drone's flight altitude; the obstacle avoidance constraints for the drone are... The drone uses an onboard distance sensor to detect environmental information, and the obtained surrounding environmental information vector is S. o [n] = (d1[n], d2[n], ..., d8[n]), where d i[n], i∈{1,2,…,8} represents the distance between the drone and the obstacle detected by the sensor in a certain direction. The sensor has a maximum detection distance limit.

[0029] Furthermore, cellular automata are used to describe the forest fire spread process. The cellular automata divides 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 topography, vegetation, weather conditions, ignition time, and combustion state within the cell. The cell state can be divided into five stages: unburned, burning, fully burned, not extinguished, and fully extinguished, denoted as F0, F1, F2, F3, and F4, respectively. The initial rate of forest fire spread, R... F0 The formula is:

[0030] R F0 =a*T F +b*V F +c*(100-H F )+d

[0031] Among them, T F V represents temperature. F Wind speed rating, H F 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 R = R0. F0 *K f *K s *K w Where R is the corrected wildfire spread rate; K f It is the correction factor for combustible materials; K s K is the terrain slope correction factor. s The effect of different slope gradients on the rate of wildfire spread was described; K w K is the wind vector correction factor. w The effects of different wind speeds and wind directions on the spread rate are described.

[0032] Furthermore, the coordinate indices of each cell are defined as u and v, and the fire spread rate of cell (u,v) is defined as... The distance between a cell and its neighboring cells is the side length L of the cell. The state transition rule for a cell is: when an unburned cell is ignited, F0 changes to F1; F1 undergoes an internal combustion time... It then transforms into F2, igniting F0 in the direction parallel to the fire line within a time interval Δn, where R INThis represents the rate of forest fire spread within a cell, where L represents the cell's side length. F2 ignites all surrounding F0 states. When there are no more F0 states around the cell in state F2, it transitions to state F3, taking Δn time. After Δn time, F3 enters state F4. (Set) The area represented by the spread of the forest fire is composed of cells that have entered state F1 and thereafter. If the coordinates of node k are q k exist In this case, node k is the failed node.

[0033] Specifically, the mobile node model is explained as follows: Due to the different tasks they perform, different mobile nodes have different movement characteristics. Firefighters and unmanned vehicles, as ground nodes, have a relatively small range of movement during UAV data collection missions; unmanned vehicles approach the fire scene, while firefighters maintain a distance from it. UAVs, as aerial nodes, have a larger range of activity, and depending on the specific mission, the node UAVs may have different speeds and directions.

[0034] Furthermore, the spatial locations of the ground nodes (including K2 unmanned vehicle nodes and K3 firefighter nodes) are as follows: Since the ground node's movement speed is relatively low compared to the mission drone's flight speed, it is considered as a short-distance, slow linear motion; the ground node's speed is... Direction is The spatial positions of K1 node drones are speed is heading The node drones used for monitoring fire sites move in accordance with the changes in the fire line. That is, the speed of the node drones will change according to the task requirements based on the forest fire spread rate R in the fire area model. The node drones used for monitoring a key area will cruise over the area using a fixed trajectory. The cruise trajectory of the node drones will be different depending on the size of the area.

[0035] Furthermore, within time slot n, the locations of the primary ignition points A and B, which are closest to a certain node, are respectively (x... a ,y a ) and (x b ,y b ), and y b >y a The edge direction of the fire line formed by A and B is Indicative quantity of changes in the relative position of nodes and wildfires

[0036] Furthermore, define the unit data value p of a node. k [n] is as follows:

[0037]

[0038] Where, p k [n] represents the unit data value of node k in the nth time slot, β is the node movement trend attention coefficient, ranging from 1 to 2, preferably 1 in this embodiment, γ is the firefighter life safety coefficient, γ is 0 when node k is a drone node or unmanned vehicle node, and 2 when node is a firefighter node.

[0039] Furthermore, the node drone may adjust its working altitude at any time according to mission requirements, and the node drone's flight altitude is close to h. u This introduces new safety threats to the mission drone's flight. Furthermore, considering the presence of L non-cooperative drones in the same space, the safe distance constraint d between the mission drone and other drones... uk [n] is represented as:

[0040] d uk [n] = (x[n] - x k [n]) 2 +(y[n]-y k [n]) 2 +(h u -h k [n]) 2 ≥D u ,

[0041]

[0042] Among them, D u The calculation formula is:

[0043]

[0044] Among them, l mm Indicates the maximum external dimensions of the mission-specific UAV, l om d represents the maximum external dimensions of other drones. safe Indicates the safety redundancy distance, d safe The preferred length is 1.5m.

[0045] Specifically, the data collection task model is explained as follows: The data collection task model is used to establish communication constraints, which include scheduling constraints and data transmission constraints. In forest fire scenarios, the channel between the UAV and the ground mobile node is mainly affected by vegetation. The path loss model PL used in this invention... k The formula for [n] is:

[0046] PL k [n] = 20log 10 (f c )+10n1log10 (θ k [n])+10n2log 10 (d k [n])+X σ

[0047]

[0048] Among them, f c Represents the carrier frequency, θ k [n] represents the elevation angle of the UAV relative to the ground node; X σ This indicates shadow fading, which follows a Gaussian distribution; d k [n] represents the distance between the drone and node k; the drone's spatial position is (x[n], y[n], h). u The position of the kth node is (x k [n],y k [n],h k [n]); n1 and n2 are the elevation angle influence factor and path loss index, respectively.

[0049] Furthermore, let the node's transmit power be p0, and the UAV receiver's noise power be p. n The signal-to-noise ratio γ at the drone receiver n [n] is:

[0050] γ n [n] = p0 - PL k [n]-p n

[0051] Furthermore, the communication rate between the drone and the ground mobile node for:

[0052]

[0053] Where B represents the channel bandwidth.

[0054] The communication channel between the mission drone and the drone node can be considered as line-of-sight propagation, with a channel gain of [value missing]. Then, based on free-space propagation loss, the communication rate between the task-oriented UAV and the node UAV is: Where β0 is the channel gain per unit distance.

[0055] Furthermore, in order to support stable data stream transmission, the received signal-to-noise ratio γ at the drone needs to be [value missing]. k Data acquisition begins when [n] ≥ γ0, where γ0 is a manually set signal-to-noise ratio threshold, preferably ranging from -5dB to -10dB. In this embodiment, -5dB is preferred. The mission drone collects data from at most one node at a time. The connection between the drone and the k-th node is represented by the scheduling constraint as follows:

[0056] Furthermore, the UAV prioritizes data transmission tasks with nodes that have good and stable communication channels. The scheduling strategy formula is as follows:

[0057]

[0058] Where active nodes are represented by the signal-to-noise ratio γ. k Nodes with [n] ≥ γ0, and otherwise nodes with a signal-to-noise ratio below the threshold γ0 and failed nodes.

[0059] Furthermore, the data transmission constraint obtained from the amount of data collected by the drone from node k is expressed as: This represents the total amount of data collected from node k at time slot n, with the superscript L used for distinction. and Q K 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 time slot length of the nth time slot is δ[n] = T. max / N, in this embodiment, the number of time slots N is preferably 300.

[0060] S2. Construct an optimization problem based on the system model.

[0061] Specifically, the optimization problem is expressed as:

[0062]

[0063] subject to

[0064] c1:

[0065] C2:

[0066] C3:

[0067] C4:

[0068] C5:d uk [n] = (x[n] - x k [n]) 2 +(y[n]-y k [n]) 2 +(h u -h k [n])2 ≥D u ,

[0069]

[0070] C6:

[0071] C7:

[0072] C8:T≤T max

[0073] Among them, C1 is the UAV flight speed performance constraint, C2 is the UAV turning performance constraint, C3 is the environmental wind uncertainty constraint, C4 is the obstacle avoidance constraint, C5 is the safe distance constraint between UAVs, C6 is the scheduling constraint, C7 is the data transmission constraint, and C8 is the mission time constraint.

[0074] S3. Use a deep reinforcement learning model to solve the optimization problem; obtain an agent for planning obstacle avoidance paths based on the solution results.

[0075] Furthermore, the sequential decision problem is expressed using a Markov decision process, i.e., a formalized expression, using tuples. The decision-making process is explained in the following terms:

[0076] Representing the state space, the environmental information vector S[n] = (S u [n],S o [n],S s [n],S c [n],S t ); where S u [n] is the state information vector of the UAV. v[n] represents the flight speed of the drone, and the formula is v[n] = v u [n]+λv w [n], v u [n] represents the ground speed of the UAV in a windless environment, and λ is the discount factor for the influence of wind speed, preferably ranging from 0.005 to 0.01, with 0.01 being preferred in this embodiment; S o [n] = (d1[n], d2[n], ..., d8[n]), where d i [n], i∈{1,2,…,8} represents the distance between the drone and the obstacle detected by the sensor in a certain direction. The sensor has a maximum detection distance limit. S o [n] represents the environmental information vector obtained by the UAV through its onboard distance sensor; S s [n] represents the global node information vector. sk [n] represents the information vector of a single node k; S c [n] is the information vector of the non - cooperative UAV. Since the input of the neural network is the environmental information vector, and S c the change in the dimension of [n] will affect the dimension of the environmental information vector, so its dimension is restricted to C, then there is S c [n] = (S l [n]), l ∈ C, C < L. When the number of non - cooperative UAVs near the mission UAV is less than C, the corresponding input dimension is filled with 0; S t is the time consumed for task execution.

[0077] represents the action. The action of the agent is restricted by the mechanical performance of the UAV and can only take executable action pairs sampled from the set of legal actions and define safety rules. When the next action of the UAV will fly towards an obstacle, cancel this action and give a penalty.

[0078] represents the transition of the state in the whole system.

[0079] represents the reward function set according to the optimization problem.

[0080] γ represents the current value discount of future rewards, and its value range is [0, 1].

[0081] Use the double - deep Q - network reinforcement learning model to solve the optimization problem. When solving, set as follows:

[0082] The data collection reward is where α1 represents the data collection amount scaling coefficient, which is used to reduce the data collection amount to a suitable range. In this embodiment, α1 is preferably 10 -6 , p k [n] represents the unit data value of the k - th node, represents the total data collection amount collected from node k at the (n + 1) - th time slot.

[0083] The remaining task time reward is α2 represents the time amount scaling coefficient, which is used to reduce the time amount to a suitable range. In this embodiment, α2 is preferably 0.1.

[0084] The dynamic obstacle avoidance penalty is α3 represents the obstacle avoidance penalty coefficient, which is used to amplify the minimum detection distance to a suitable range. In this embodiment, α3 is preferably 10, d represents the minimum value of the distances between the UAV and the obstacles detected by the sensors in all directions, D s represents the longest detection distance of the UAV from the sensor.

[0085] For known obstacles, there is a static obstacle collision penalty. α4 represents the collision penalty value, and in this embodiment, α4 is preferably 1.

[0086] Safety penalties for multi-drone flight environments are α5 represents the safe distance maintenance coefficient of the drone, and in this embodiment, α5 is preferably 0.5.

[0087] To reduce invalid actions, the penalty for invalid actions is R. ia =-α6, where α6 represents the penalty value for invalid actions. In this embodiment, α6 is preferably 1.

[0088] The global reward function is R = R dc +R tk +R uo +R so +R uf +R ia .

[0089] Specifically, the experimental parameters for this embodiment are shown in the table below:

[0090]

[0091]

[0092] Furthermore, to address the problem of overestimating action value, the target Q-value function in the dual-depth Q-network is designed as follows:

[0093]

[0094] Where S[n+1] is the condition for determining the final state: all The corresponding node signal disappears, or the task execution time T is greater than the maximum task time constraint T. max .

[0095] Furthermore, the steps for solving the optimization problem using a deep reinforcement learning model are as follows:

[0096] Step 1: Set the signal-to-noise ratio threshold γ0 and the maximum task time T. max The maximum flight speed of the drone is v max and maximum steering angle

[0097] Step 2: Initialize the replay buffer D, estimate the network ξ, and initialize the target network ξ by copying the estimated network parameters. - .

[0098] Step 3, from v max , Calculate the legal action space.

[0099] Step 4: Initialize the experimental environment by setting all environmental parameters to their initial values.

[0100] Step 5: Decrease the learning rate linearly to improve the model's convergence performance.

[0101] Step 6: Obtain the environmental information vector S[n], and perform state normalization on S[n] to obtain S′[n].

[0102] Step 7: Input S′[n] into the estimation network to obtain action a[n]. The agent interacts with the environment and receives a reward. And the new environmental information vector S[n+1].

[0103] Step 8: Normalize S[n+1] and convert the tuples... Add to replay cache D.

[0104] Step 9: Sample from D, calculate the gradient using the loss function, and update the network parameters.

[0105] Step 10, every N r The next update copies the parameters in ξ to ξ. - N r In this embodiment, N is a threshold number of training iterations set manually for the neural network. r The preferred value is 600.

[0106] Step 11: Determine if S[n+1] is the final state; if yes, proceed to step 4; otherwise, proceed to step 5.

[0107] Step 12: 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 agent. The expected value for the data collection rate is preferably 99%, and the expected value for the path planning success rate is preferably 95%.

[0108] S4. Deploy the intelligent agent onto the drone. The drone acquires real-time environmental data. The intelligent agent provides the drone with action instructions for the next moment based on the real-time environmental data. The drone then performs data collection tasks according to the action instructions.

[0109] Furthermore, Figure 3The convergence process of the proposed algorithm and the comparison algorithm after smoothing is demonstrated. Specifically, the proposed algorithm is a UAV data collection obstacle avoidance trajectory planning algorithm based on dual-depth Q-network, while the comparison algorithm is the basic Dueling DQN algorithm. It can be seen that the reward value of the basic Dueling DQN algorithm increases during training and then fluctuates sharply without reaching its maximum. The proposed algorithm shows a significant performance improvement and exhibits smaller fluctuations during convergence, demonstrating stability in forest fire environments with dynamic obstacles.

[0110] Figure 4 The paper demonstrates the total data collection rate of the UAV from all nodes in each mission under different algorithms. Affected by environmental factors, fire spread, aerial nodes, and non-cooperative UAVs, the collection rate of the basic Dueling DQN algorithm initially rises and then hovers around 90%, before decreasing to around 75%. The algorithm proposed in this invention, however, reaches 100% after an initial rise and then stabilizes at a level close to 100%, demonstrating the effectiveness of the proposed algorithm in data collection tasks for moving nodes under the influence of dynamic obstacles.

[0111] Figure 5 The data collection rate of the UAV from the air node was shown under different algorithms. It can be seen that due to the existence of a good line-of-sight propagation path, the mission UAV achieved a 100% data collection rate under different algorithms. Moreover, the UAV had already achieved 100% data collection from the air node at the beginning of the training process.

[0112] Figure 6 The data collection rates of UAVs on ground nodes, including unmanned vehicles and firefighters, under different algorithms are shown. Compared with air nodes, ground nodes have greater communication channel loss. The mobility of nodes requires the task UAV to flexibly adjust its trajectory during the data collection process. It can be seen that the collection rate of the basic Dueling DQN algorithm hovers around 90% after an initial rise, and then decreases to about 70%. The algorithm proposed in this invention reaches 100% after an initial rise and then stabilizes at a level close to 100%.

[0113] Figure 7 The figure shows the data collection rates of drones for high-value nodes under different algorithms. The data in the top 30% of the data per unit value are high-value node data. It can be seen that the basic Dueling DQN algorithm fluctuates wildly at around 90% in the initial stage, and then decreases to 70%; the algorithm proposed in this invention converges quickly to around 100%, showing a preference for high-value data.

[0114] Figure 8The success rate of UAV trajectory planning under different algorithms is shown. It can be seen that although the same method of environmental perception is used, when faced with both static and dynamic obstacles in the environment, the trajectory planning success rate of the basic Dueling DQN algorithm increases in the initial stage and then decreases continuously and fluctuates drastically, which cannot guarantee the flight safety of the UAV itself. The success rate of the algorithm proposed in this invention converges to more than 95% after a slow increase, demonstrating the adaptability of the proposed algorithm to the complex flight environment of forest fires.

[0115] 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 obstacle avoidance data from a UAV oriented towards moving nodes during forest fires, characterized in that, include: S1. Acquire data from ground mobile nodes and UAV nodes, and establish a system model based on the data from ground mobile nodes and UAV nodes; Ground mobile nodes include unmanned vehicle nodes and firefighter nodes; the system model includes system architecture, fire area model, mobile node model, and data collection task model; the system architecture consists of... It consists of several wireless sensor nodes, and the wireless sensor nodes include One drone node, One unmanned vehicle node and A firefighter node, The fire zone model uses cellular automata to describe the forest fire spread process. The mobile node model calculates the unit data value of a node based on data from ground mobile nodes and drone nodes; S2. Construct an optimization problem based on the system model; the unit data value is used to construct the optimization problem; the formula for the unit data value is: ； in, Indicates the first Nodes in each time slot The value of a unit of data This represents the attention coefficient for node movement trends. An indicator of the change in the relative position of a node to a wildfire. Indicates the life safety factor of firefighters; S3. Solve the optimization problem using a deep reinforcement learning model; obtain an agent for planning obstacle avoidance paths based on the solution results; S4. Deploy the intelligent agent onto the drone. The drone acquires real-time environmental data. The intelligent agent provides the drone with action instructions for the next moment based on the real-time environmental data. The drone then performs data collection tasks according to the action instructions.

2. The UAV data collection obstacle avoidance trajectory planning method for mobile nodes according to claim 1, characterized in that, The constraints of the optimization problem include: UAV flight speed performance constraints, UAV turning performance constraints, environmental wind force uncertainty constraints, obstacle avoidance constraints, safe distance constraints between UAVs, scheduling constraints, data transmission constraints, and task time constraints.

3. The UAV data collection obstacle avoidance trajectory planning method for mobile nodes according to claim 1, characterized in that, A dual-deep Q-network reinforcement learning model is used to solve the optimization problem.

Images