Unmanned aerial vehicle cluster task planning method based on dubins direction angle model

Abstract

The application discloses a kind of based on dubins direction angle model unmanned aerial vehicle cluster task planning method, comprising: establishing dubins direction angle model;Target task point coordinate information is obtained, and dubins direction angle model is combined with multi-traveler problem, and the mathematical model of unmanned aerial vehicle cluster task planning that meets the shortest track requirement is established;Unmanned aerial vehicle cluster task planning mathematical model is optimized and solved, and the optimal scheme of unmanned aerial vehicle cluster task planning is output.The method effectively solves the unmanned aerial vehicle circling condition caused by the close distance of task point in the task planning process without increasing the amount of calculation, so that the unmanned aerial vehicle formation can complete team task safely and efficiently under the premise of the shortest voyage.

Description

Technical Field

[0001] This invention belongs to the field of UAV swarm collaborative search technology, specifically relating to a UAV swarm task planning method based on the Dubins orientation angle model. Background Technology

[0002] In recent years, unmanned aerial vehicles (UAVs) and related technologies have continuously developed and matured. Due to their low cost, maneuverability, and portability, they are widely used in collaborative reconnaissance, collaborative strikes, flight demonstrations, express delivery and logistics, and agricultural spraying. However, against the backdrop of continuous technological updates and complex application environments, UAV technology is gradually developing towards intelligence, networking, and swarming. Control systems based on multi-UAV (Unmanned Aerial Vehicle) swarms are currently a key research focus both domestically and internationally. UAV swarms overcome the low efficiency and survivability of individual UAVs, possessing advantages such as group stability and high efficiency, achieving multi-dimensional and large-area coverage. As a collective behavior, the most important characteristic of UAV swarming is its evolution from simple local rules to coordinated global behavior. Therefore, swarming will inevitably be the future trend of UAV development. Highly autonomous and intelligent swarms can rely on intelligent algorithms to complete tasks such as formation assembly and formation search.

[0003] Task planning is crucial in UAV swarm technology. Based on the different requirements of mission objectives, mission environment, and swarm characteristics, it involves rationally allocating mission objectives and planning swarm mission trajectories to enable the UAV swarm to complete all tasks efficiently and safely. Task planning can be viewed as a process of combining and allocating mission objectives with swarm characteristics, followed by optimization of the combined solution. Depending on the mission objectives and swarm attributes, task allocation models can be categorized into multiple traveling salesman problem models, vehicle routing models, mixed-integer linear programming models, and network optimization models, among others. The optimization process typically employs biomimetic intelligent algorithms, such as Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Ant Colony Optimization (ACO), Differential Evolution Algorithm (DE), and Group Search Optimizer (GSO), as well as their derivative optimization algorithms. Through iterative computation using group optimization or partial group optimization heuristics, the combined task can be quickly solved to obtain the desired optimal value.

[0004] Currently, significant progress has been made in the research of mission planning problems for unmanned aerial vehicle (UAV) swarms. For example, Ye et al. proposed an adaptive genetic algorithm with multi-type gene chromosome encoding strategies to solve the CMTAP model; Guang et al. proposed a heterogeneous UAV model considering UAV kinematic constraints; Liang et al. considered multiple complex constraints, including mission time constraints and UAV ammunition consumption, and proposed a multi-UAV cooperative multi-task allocation method based on discrete particle swarm optimization; Wang Jian et al. proposed a minimum risk model based on uncertainty theory with a confidence function as the objective for UAV mission allocation problems with uncertain target values.

[0005] However, most of the existing methods mentioned above suffer from high computational costs, and when task points are close together, drones are prone to detours, making it difficult for drone swarms to complete team tasks safely and efficiently. Furthermore, existing biomimetic intelligent algorithms generally have weak global search capabilities and are prone to getting trapped in local optima, impacting the efficiency of collaborative search in drone swarms. Summary of the Invention

[0006] To address the aforementioned problems in the existing technology, this invention provides a UAV swarm mission planning method based on the Dubins orientation angle model. The technical problem to be solved by this invention is achieved through the following technical solution:

[0007] A method for UAV swarm mission planning based on the Dubins orientation angle model includes:

[0008] Step 1: Establish the Dubins orientation angle model;

[0009] Step 2: Obtain the target task point coordinate information, and combine the Dubins orientation angle model with the multiple traveling salesman problem to establish a mathematical model for UAV swarm task planning that meets the shortest path requirement;

[0010] Step 3: Optimize and solve the mathematical model for UAV swarm mission planning, and output the optimal solution for UAV swarm mission planning.

[0011] The beneficial effects of this invention are:

[0012] 1. This invention uses Dubins curves to simulate the actual flight trajectory of UAVs. To address the azimuth angle selection problem, a Dubins azimuth angle model is proposed. Based on the distribution relationship between task points, a method for setting the azimuth angle that meets the requirements of the shortest path is given. Combined with the multiple traveling salesman problem, a mathematical model for UAV swarm task planning is established. Without increasing the amount of computation, it effectively solves the problem of UAVs flying around due to the close distance between task points during the task planning process, enabling UAV formations to complete team tasks safely and efficiently under the premise of the shortest flight distance.

[0013] 2. To address the problem that the global search capability of the whale algorithm is weak and prone to getting trapped in local optima, this invention proposes an improved whale algorithm based on extreme individual guidance. In the target hunting and prey searching stages, the population mutation is guided to move towards the global variable and the optimal individual. While ensuring iterative efficiency, the algorithm avoids getting trapped in local optima and improves the efficiency of UAV swarm collaborative search.

[0014] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a UAV collaborative task allocation model provided in an embodiment of the present invention;

[0016] Figure 2 A schematic diagram of the Dubins curve provided in an embodiment of the present invention;

[0017] Figure 3 A flowchart illustrating a UAV swarm mission planning method based on the Dubins orientation angle model provided in an embodiment of the present invention;

[0018] Figure 4 This is a schematic diagram of a scenario where the radius of flight of the target point and the target points before and after it does not intersect, provided in an embodiment of the present invention.

[0019] Figure 5 This is a schematic diagram of a scenario where the radius of flight of a target point intersects with that of the target point before and after it, as provided in an embodiment of the present invention.

[0020] Figure 6 This is a schematic diagram of a scenario where a target point is within the radius of a flight path of one target point before or after the target point, provided in an embodiment of the present invention.

[0021] Figure 7 A flowchart of the improved whale algorithm provided in this embodiment of the invention;

[0022] Figure 8 The flight path is planned for the UAV mission with a fixed heading angle in Experiment 1.

[0023] Figure 9 This is the flight path diagram for the UAV mission in Experiment 1, where the heading angle is towards the next target point.

[0024] Figure 10 This is a flight path diagram of the UAV mission planning using the DAM model of this invention in Experiment 1;

[0025] Figure 11 The overall flight path diagram for different UAV models in Experiment 1;

[0026] Figure 12 The flight path planning results for the UAV swarm mission in Experiment 2 are shown in the figure.

[0027] Figure 13-14 The graph shows the results of two different task planning methods when the WOA algorithm reaches local convergence in Experiment 3.

[0028] Figure 15 The diagram shows the UAV swarm task planning results of the PWOA algorithm in Experiment 3;

[0029] Figure 16 This is a comparison chart showing the number of iterations required for different algorithms to converge in Experiment 3.

[0030] Figure 17 This is a schematic diagram illustrating the iterative convergence process of UAV swarm task planning using different algorithms in Experiment 3. Detailed Implementation

[0031] The present invention will be further described in detail below with reference to specific embodiments, but the implementation of the present invention is not limited thereto.

[0032] Example 1

[0033] Before introducing the solution of this invention, let’s first introduce the UAV collaborative task allocation model.

[0034] In such Figure 1 Within the task area shown, N targets are irregularly distributed and need to be searched. There are currently M fixed-wing UAVs, fewer than the number of targets. The minimum turning radius of each UAV is R meters. Before the UAV swarm conducts search and reconnaissance of the task area, the targets within the task area need to be assigned. Through planning and calculation by the ground station, specific search targets and search paths are assigned to individual UAVs. Each target requires only one UAV to search. After searching all target points, the UAV swarm mission ends.

[0035] The model optimization objective is to optimize the shortest flight path of the UAV swarm. Considering the actual performance of the UAVs, the Dubins curve is used to simulate the shortest path planning between two target points, and the shortest turning radius of the UAV is used as the radius of the Dubins curve.

[0036] The Multiple Traveling Salesman Problem (MTSP) is an extension of the Traveling Salesman Problem (TSP), and both are NP-hard problems. In the MTSP, n couriers visit m delivery locations, where m > n. Each courier appears at each location only once, and the goal is to minimize the total distance traveled by all couriers after reaching all locations. Task planning for a drone swarm is analogous to the MTSP, with the drones acting as couriers. Each drone is assigned multiple search tasks, and the goal is to minimize the total distance traveled by all drones after completing their search tasks. Depending on the starting point, this problem can be categorized into several types. This invention primarily focuses on the problem of drone swarms taking off and landing from the same point.

[0037] Specifically, the total flight distance from the starting point of the drone swarm to the return of all drones is used as the objective of the MTSP model optimization. There are n target points, and the coordinates of all target points are represented by S = [(x1,y1),(x2,y2),…,(x...y1)...y2). n ,y n If we represent the distance between two points after the drone swarm task planning, then the distance between them can be expressed as:

[0038] d i,j =|x i -x j |+|y i -y j | (1)

[0039] The drones in this invention are all fixed-wing drones. Unlike rotary-wing drones, the trajectory change of a fixed-wing drone is a linear process. Therefore, the distance formula between two points needs to be supplemented with constraints that conform to the drone's motion trajectory. Here, a Dubins curve with a turning radius of 100 meters is used to simulate the actual trajectory change energy of the drone, that is:

[0040] d i,j =l+r+D (2)

[0041] Where D represents the straight line portion of the Dubins curve between two adjacent mission points in the mission planning trajectory, and l and r are the arc lengths between the two adjacent mission points and the straight line.

[0042] However, calculating the distance between two task points using the Dubins curve requires specific direction angles and coordinate parameters, both of which are indispensable. For example... Figure 2The diagram of the Dubins curves shown illustrates that, depending on the direction angle, the Dubins curves have six different structures: LSL, RSR, RSL, LSR, RLR, and LRL. It can be seen that different direction angles lead to different shortest paths between two points. In the subsequent optimization process of the model, a biomimetic intelligent algorithm is used to solve the model. The initial population includes the initial population of the UAV swarm (population size N) and the initial population of the target point to be searched (population size M). In the iterative calculation, the computational scale corresponding to a single iteration of the UAV is N×M. When the UAV faces different task assignments, the optimal direction angle θ between adjacent points keeps changing. Adding another initial population with a different direction angle θ (population size B) causes the computational scale to increase exponentially, resulting in a dramatic increase in computational scale and reduced computational efficiency.

[0043] To optimize the solution process, this invention proposes a Dubins Azimuth model (DAM). Based on the distribution relationship between task points, it provides a method for setting the azimuth angle that meets the requirements of the shortest path. Combined with the MTSP problem, a UAV swarm Dubins Multiple Traveling Salesman Problem (UAV-DMTSP) model is established. By solving the model, the optimal planning scheme is obtained.

[0044] For details, please see Figure 3 , Figure 3 This is a flowchart illustrating a UAV swarm mission planning method based on the Dubins orientation angle model provided in an embodiment of the present invention, which includes:

[0045] Step 1: Establish the Dubins orientation angle model.

[0046] In this embodiment, based on the distance between the target point and the two points before and after the task planning, a Dubins curve azimuth angle model can be constructed for different scenarios, and the corresponding shortest flight distance under different scenarios can be analyzed; then, the azimuth angle of the Dubins curve can be obtained based on the corresponding shortest flight distance under different scenarios.

[0047] The different scenarios include scenarios where the target point and the radius of flight of the preceding and following target points do not intersect, scenarios where the target point and the radius of flight of the preceding and following target points intersect, and scenarios where the target point is within the radius of flight of the preceding and following target points.

[0048] The following sections will provide a detailed introduction to these three scenarios.

[0049] 1. Scenarios where the radius of flight of the target point and the target points before and after it does not intersect.

[0050] Please see Figure 4 , Figure 4This is a schematic diagram of a scenario where the target point and the radius of flight of the preceding and following target points do not intersect, provided by an embodiment of the present invention. A, B, and C are the mission target points, D and E are the tangent points, R is the minimum radius of the Dubins curve, and α and β are the central angles corresponding to point C. a, b, and c are the distances between the three target points. Therefore, the total flight distance L between the three points after mission planning can be expressed as:

[0051]

[0052] Among them, L AD and L BE S represents the straight-line distance between points A and D, and between points B and E, respectively. 弧DE Let be the length of the arc between points D and E.

[0053] As can be seen from equation (3) above, the value of L is related to the values ​​of α and β, as shown in equation (4):

[0054]

[0055] From equation (4), it can be seen that when the value of α+β is smaller and α=β, the value of equation (3) is the smallest. Figure 4 It can be seen that when the center O of the circle lies on the angle bisector of angle C, the total distance L reaches its minimum value, which is:

[0056] 2. Scenarios where the radius of flight of the target point intersects with that of the target point before and after it.

[0057] Please see Figure 5 , Figure 5 This is a schematic diagram of a scenario where the radius of flight of a target point intersects with that of the preceding and following target points, provided by an embodiment of the present invention. A, B, and C are the target points; D, E, E', F, and G are the tangent points; R is the minimum radius of the Dubins curve; and α1, β1, α2, and β2 are the central angles corresponding to points A and B. a, b, and c are the distances between the three target points.

[0058] from Figure 5 The total distance L between the three points after mission planning can be obtained, as shown below:

[0059]

[0060] As shown in equation (5), the value of L is related to the values ​​of α1, β1, α2, and β2, as shown in equation (6):

[0061]

[0062] From equation (6), it can be seen that when the values ​​of α1+β1 and α2+β2 are smaller, and when α1=β1 and α2=β2, the value of equation (5) is the smallest.

[0063]

[0064] Combination Figure 5 It can be seen that the total distance L reaches its minimum when the center O1 is on the angle bisector of ∠A and the center O2 is on the angle bisector of ∠CBE. E'F is the tangent of circle 1 and circle 2.

[0065] 3. Scenarios where the target point is within the radius of the flight path of the preceding and following target points.

[0066] Please see Figure 6 , Figure 6 This is a schematic diagram of a scene where a target point is within the radius of the flight path of the preceding and following target points, provided by an embodiment of the present invention. A, B, and C are the target points, R is the minimum radius of the Dubins curve, and α, β, and θ are the central angles corresponding to points A and B. a, b, and c are the distances between the three target points.

[0067] from Figure 6 The total distance L between the three points after mission planning can be obtained, as shown below:

[0068]

[0069] As shown in equation (7), the value of L is related to the values ​​of α, β, and θ, as shown in equation (8):

[0070]

[0071] From equation (8), we can see that when the values ​​of α, β, and θ are smaller, and α = β, the value of equation (7) is the smallest, which is:

[0072] Comparing equations (5) and (8), it can be seen that when there is another target point within the radius of the target point, the method of taking the mission points on the same circle results in a shorter flight distance L.

[0073] By analyzing the models under the different conditions above, the general formula for the Dubins curve direction angle model can be obtained, and its expression is:

[0074]

[0075] Where ∠A represents the direction angle of the Dubins curve, and Let |AC| represent the vector from point A to point C and from point A to point B, and |AB| represent the lengths of AC and AB, respectively. The sign of the direction angle depends on the positional relationship between the target point and its adjacent points.

[0076] After obtaining the direction angle of the Dubins curve, it can be substituted into the Dubins curve distance calculation formula to obtain the total range. For details, please refer to existing related technologies; they will not be described in detail here.

[0077] Step 2: Obtain the target task point coordinates and combine the Dubins orientation angle model with the multiple traveling salesman problem to establish a mathematical model for UAV swarm task planning that meets the shortest path requirement.

[0078] Specifically, the Dubins heading angle model is combined with the multiple traveling salesman problem, and the shortest flight distance of the UAV swarm is used as the optimization objective to construct a mathematical model for UAV swarm mission planning. Its objective function can be expressed as follows:

[0079]

[0080] In equation (9) above, L all k represents the total distance traveled. i,j k represents whether the i-th drone has a planned mission at the j-th target point. i,j The value is either 0 or 1. The constraint function is as follows:

[0081]

[0082]

[0083]

[0084]

[0085] Equation (10) represents the number of targets assigned to each drone. This constraint ensures that each drone is assigned no less than k targets, where k is a custom value set by the model requirements. Equation (11) represents the total number of targets assigned to the drone cluster. This constraint ensures that all target points are assigned. Equation (12) represents the number of times a target point is assigned to a drone. This constraint ensures that each point is assigned only once. Equation (13) is the determination formula for the starting and ending points, which requires that the starting and ending points of each aircraft are the same point.

[0086] This invention uses Dubins curves to simulate the actual flight trajectory of UAVs. Addressing the azimuth angle selection problem, it proposes a Dubins azimuth angle model. Based on the distribution relationship between task points, it provides a method for setting azimuth angles that meets the shortest path requirement. Furthermore, by combining this with the multiple traveling salesman problem, a mathematical model for UAV swarm mission planning is established. Without increasing computational load, this effectively solves the problem of UAVs flying around each other due to close distances between task points during mission planning, enabling UAV formations to safely and efficiently complete team tasks with the shortest possible flight distance.

[0087] Step 3: Optimize and solve the mathematical model for UAV swarm mission planning, and output the optimal solution for UAV swarm mission planning.

[0088] Optionally, as an implementation method, this embodiment selects the Whale Optimization Algorithm (WOA) for optimization.

[0089] Generally, the WOOA algorithm consists of three modes: target encirclement, bubble encirclement, and prey search. Below is a brief introduction to the three modes of the WOOA algorithm.

[0090] 1. Target of the operation

[0091] Like other biomimetic algorithms, the whale algorithm generates an initial population of size N and dimension D. Before the search task begins, individual whales do not know the specific location of the target. Therefore, the individual with the highest fitness function in the current population is taken as the current global optimum, and all other individuals move towards this individual, thereby gradually encircling and capturing the target. The population update formula is shown below:

[0092] X t+1 =X tbest -A*D (14)

[0093] D = |C*X tbest -X t | (15)

[0094] In the above formula, X t+1 It is the (t+1)th generation population, X tbest Let X be the current global optimal individual position, A and C be coefficients, D be the distance between the t-th generation individual and the optimal individual position, and X be the current global optimal individual position. t For an individual in the t-th generation of the population, the coefficients A and C are represented as follows:

[0095] A = 2 * a * ba (16)

[0096] C = 2 * b (17)

[0097] In the above formula, b is a random number between [0,1], and a decreases linearly from 2 to 0 as the number of iterations increases.

[0098] 2. Bubble Encirclement

[0099] Bubble encirclement is a spiral movement technique used by whales during hunting. In the whale algorithm model, the position of an individual whale is also updated to simulate bubble encirclement, and the specific update process is as follows:

[0100] X t+1 =D circle *ecl *cos(2π*l)+X tbest (18)

[0101] D circle =|X tbest -X t | (19)

[0102] In the above formula, c is a constant and l is a random number between [-1, 1].

[0103] In the whale algorithm, both bubble encirclement and target encirclement are hunting methods used by whales and occur simultaneously. Therefore, to select the appropriate behavior during the whale algorithm's search process, we assume that both encirclement methods have a 50% probability of occurring. Thus, the mathematical expression of the encirclement process is as follows:

[0104]

[0105] 3. Searching for prey

[0106] The whale algorithm also randomly updates the location of individual whales during the search process, communicating with other whales. The random update process is as follows:

[0107] X t+1 =X rand -A*D rand (twenty one)

[0108] D rand =|C*X rand -X t | (22)

[0109] In the above formula, X rand This represents the location information of random individuals within the population. In the whale algorithm, the same coefficient A is used for both searching and capturing prey. Similar to bubble trapping, the range of values ​​for A is used to determine whether searching or capturing prey is triggered, as shown below:

[0110]

[0111] Because the WOA algorithm, like other swarm optimization biomimetic algorithms, suffers from the problem of easily getting trapped in local optima and insufficient global search capability. In the prey-hunting phase, the WOA algorithm becomes trapped in local optima because the value of A gradually approaches 0 with increasing iterations. Therefore, to enhance the global search capability and improve the convergence speed of the WOA algorithm, this embodiment proposes an improved whale algorithm (PWOA) guided by extreme individuals. In the target-hunting and prey-searching phases, it guides population mutation towards the global variable and the optimal individual, respectively, to optimize the solution of the UAV swarm mission planning mathematical model.

[0112] Specifically, during the target capture phase, information exchange is introduced between some random individuals and the current best individual. This allows the whale individual to move towards the global best individual while also converging towards other global variables, thereby improving the global search capability of the whale algorithm. The position update formula is as follows:

[0113] X t+1 =X tbest -A*D (24)

[0114] D=d*|C*X tbest -X t |+(1-d)*|C*X rand -X t | (25)

[0115] In the above formula, d is a random variable between [0,1].

[0116] During the prey-hunting phase, extreme value individuals are introduced to guide the whale group, causing it to move towards the global optimum as it progresses towards the random variable. This allows the group to enter the optimization process earlier in the random search phase, and the position update formula is as follows:

[0117] X t+1 =X rand -A*D rand (26)

[0118] D rand =d*|C*X rand -X t |+(1-d)*|X tbest -X t | (27)

[0119] Therefore, combined Figure 7 As shown, the optimization process using the improved whale algorithm in this embodiment can be described as follows:

[0120] a) Initialize the whale algorithm parameters, generate an initial population, set the size to N, the dimension to D, and the maximum number of iterations to T;

[0121] b) Calculate the adaptive parameters of individual whales and calculate the optimal individual X. tbest Location;

[0122] c) Determine the value of p. If p ≤ 0.5, proceed to step d); otherwise, proceed to step f.

[0123] d) Determine the value of A. If |A|≤1, proceed to step e); otherwise, proceed to step g.

[0124] e) After updating the global optimal individual position using the following formula, proceed to step h);

[0125] Xt+1 =X tbest -A*D;

[0126] D=d*|C*X tbest -X t |+(1-d)*|C*X rand -X t |;

[0127] Among them, X t+1 It is the (t+1)th generation population, X tbest Let X be the current globally optimal individual position, A and C be coefficients, and d be the guiding coefficient for the extreme value individual, taking the value as a random variable between [0, 1]. t For an individual in the t-th generation of the population, X rand This refers to the location information of random individuals within the population.

[0128] f) After updating the individual whale population locations using the following formula, proceed to step h);

[0129] X t+1 =D circze *e cl *cos(2π*l)+X tbest ;

[0130] D circze =|X tbest -X t |;

[0131] Where c is a constant and l is a random number between [-1, 1];

[0132] g) Update the individual whale population locations using the following formula, then proceed to step h);

[0133] X t+1 =X rand -A*D rand ;

[0134] D rand =d*|C*X rand -X t |+(1-d)*|X tbest -X t |;

[0135] h) Calculate the adaptive parameters of individual whales in the population, find the globally optimal individual, update the position of the optimal individual, and determine whether the maximum number of iterations has been reached. If so, the calculation ends and the optimal solution is output; otherwise, return to step c).

[0136] This embodiment addresses the problem of the whale algorithm's weak global search capability and its tendency to get trapped in local optima. It proposes an improved whale algorithm based on extreme individual guidance. In the target hunting and prey searching stages, the algorithm guides population mutation towards the global variables and the optimal individuals. While ensuring iterative efficiency, it avoids the algorithm getting trapped in local optima and improves the efficiency of UAV swarm collaborative search.

[0137] Example 2

[0138] The effectiveness of the proposed DAM model and the improved whale algorithm will be verified and explained through simulation experiments below.

[0139] Experiment 1: To verify the accuracy of the DAM model proposed in this invention, this embodiment simulated a 2000m*2000m search area, randomly distributing a certain number of task targets within the search area, and performed simulations on the UAV swarm task planning model using the DAM model of this invention and the UAV swarm task planning model using a fixed direction angle.

[0140] 1.1 Experimental Content and Conditions

[0141] A drone takes off from a location, plans its trajectory for six task points within a search area, and returns to its starting point after completing its task. The search efficiency of the three models is compared: the DAM model, the fixed azimuth model, and the azimuth model pointing towards the next task point.

[0142] Please refer to Tables 1-3. Table 1 shows the initial heading angle and position parameters of the UAV, Table 2 shows the position parameters of the mission target, and Table 3 shows the initial parameter settings for the WOA algorithm.

[0143] Table 1 Initial state and position parameters of the UAV

[0144] Location (x, y) within the region Direction angle θ Minimum turning radius drones (10，0) 90 100m

[0145] Table 2 Task Point Coordinates

[0146] Start (Stop) Target1 Target2 Target3 Target4 Target5 1000,0 200,200 400,500 500,600 1500,400 1550,350

[0147] Table 3 Initial parameter settings for the WOA algorithm

[0148] Population size a b C Maximum number of iterations 50 2->0 1 [0,2] 50

[0149] 1.2 Experimental Results and Analysis

[0150] Please see Figure 8-11 ,in, Figure 8 For UAV missions with fixed heading angles (0°, 90°, -90°), plan flight paths from... Figure 8As can be seen, the drone returns to the starting point after passing five target points. However, when the target points are close, the drone has to circle around to enter the circular trajectory of the target point. This causes the drone to perform unnecessary circling operations, increasing the flight range and reducing the search efficiency.

[0151] Figure 9 The image shows the flight path of a drone mission with a azimuth angle pointing towards the next target point. As can be seen from the image, circling operations still occurred between target points 2 and 3, and 4 and 5, which is the same situation as in the fixed azimuth angle model.

[0152] Figure 10 The image shows the flight path of a UAV mission planning using the DAM model of this invention. As can be seen from the image, the model does not have any unnecessary detours during the flight path planning process at close range points. The flight path is smooth and completes all target point search tasks.

[0153] Figure 11 The total flight path is planned for different UAV models. As can be seen from the figure, when facing the target point and with a fixed value, the total flight path of the UAV is much greater than that of the UAV with the DAM model. Moreover, the straight-line distance of the DAM model is not much greater than that of the target UAV. This also proves that the DAM model proposed in this invention can completely solve the path planning problem between close points under the DMTSP model. Under the same conditions, the total flight path is the shortest, which improves the efficiency of task search.

[0154] Experiment 2

[0155] 2.1 Experimental Content and Conditions

[0156] Building upon Experiment 1, this experiment increased the number of task points and drones to verify the scalability of the DAM model, and studied the task planning of the drone swarm. Within a 2000m*2000m task area, three drones took off from the same location, planned their routes to 29 task points within the search area, and returned to their starting point after completing their tasks.

[0157] The flight path data for the drone swarm mission planning is shown in Table 4 below.

[0158] Table 4. Flight path data for UAV swarm mission planning

[0159]

[0160] 2.2 Experimental Results and Analysis

[0161] Flight test results as follows Figure 12As shown, the three drones were assigned 13, 6, and 10 targets respectively through mission planning, and the search order was planned. Under the parameter settings of 30 m / s and a turning radius of 100 m, the total flight distance of the three drones was 174.8546 km. The shortest flight distance was 42.8364 km for drone No. 2, and the longest flight distance was 70.2387 km for drone No. 1. Drone No. 2 took the shortest time of 1427.88 seconds, and drone No. 1 took the longest time of 2341.29 seconds. The effect diagram shows that the turning points of the drones are consistent with the actual flight conditions of the drones and conform to the optimal route planning from the starting point to the return.

[0162] Experiment 3

[0163] To verify the performance of the improved whale algorithm proposed in this invention, based on Experiment 2 above, this experiment compared the drone swarm task planning of different algorithms. The improved whale algorithm (PWOA) of this invention, the existing whale algorithm (WOA), and the existing GA, DE, and QUATRE algorithms were used for optimization. The experimental results are shown below. Figure 13-17 As shown.

[0164] in, Figure 13-14 These are two different task planning results when the WOA algorithm reaches local convergence. Figure 15 The results of drone swarm task planning using the PWOA algorithm show that, through comparison, the PWOA algorithm can effectively reduce the probability of local convergence in the WOA algorithm.

[0165] Figure 16 The number of iterations required for the five algorithms to converge is given by... Figure 16 It can be seen that the PWOA algorithm requires the fewest iterations, 262.

[0166] Figure 17 The process of iterative convergence of UAV swarm task planning using different algorithms is demonstrated. Among them, the WOA and PWOA algorithms are in the target search stage when |A|>1. Their convergence speed is slower than that of the Quatre algorithm in the early stage. However, when |A|<1, the convergence speed of the WOA and PWOA algorithms is faster. As can be seen from the figure, the PWOA algorithm has the fastest convergence speed, which proves that the PWOA algorithm proposed in this invention is effective.

[0167] In summary, the DAM model proposed in this invention can effectively solve the problem of UAVs flying around each other due to the close proximity of task points during task planning; and through the results of UAV swarm task planning with different algorithms, it is proved that the PWOA algorithm can avoid getting trapped in local optima while ensuring iterative efficiency, thereby improving the efficiency of UAV swarm collaborative search.

[0168] The above description, in conjunction with specific preferred embodiments, provides a further detailed explanation of the present invention. It should not be construed that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art, various simple deductions or substitutions can be made without departing from the concept of the present invention, and all such modifications and substitutions should be considered within the scope of protection of the present invention.

Claims

1. A method for UAV swarm mission planning based on the Dubins orientation angle model, characterized in that, include: Step 1: Establish the Dubins orientation angle model; Step 2: Obtain the target task point coordinate information, and combine the Dubins orientation angle model with the multiple traveling salesman problem to establish a mathematical model for UAV swarm task planning that meets the shortest path requirement; Step 3: Optimize and solve the mathematical model for UAV swarm mission planning, and output the optimal solution for UAV swarm mission planning; wherein, Step 1 includes: Based on the target point and the distance between the two target points before and after the mission planning, Dubins curve direction angle models are constructed for different scenarios, and the corresponding shortest flight distances under different scenarios are analyzed. The heading angle of the Dubins curve is obtained based on the shortest flight distance corresponding to different scenarios; the Dubins curve heading angle model is expressed as follows: Where ∠A represents the direction angle of the Dubins curve, and Let A represent the vector from point A to point C and from point A to point B, where A, B, and C are the target points of the task, and point A is between points B and C. |AC| and |AB| represent the lengths of AC and AB, respectively. The different scenarios include scenarios where the target point has no intersection with the radius of flight of the preceding and following target points, scenarios where the target point has an intersection with the radius of flight of the preceding and following target points, and scenarios where the target point is within the radius of flight of the preceding and following target points.

2. The UAV swarm mission planning method based on the Dubins orientation angle model according to claim 1, characterized in that, For scenarios where the radius of flight of the target point does not intersect with that of the preceding and following target points, let A be the target point, and B and C be its preceding and following target points, respectively. Then, when the center O of the circle lies on the angle bisector of ∠C, the total flight distance L reaches its minimum value, which can be expressed as: Where a and b are the distances between points B and C, and between points A and C, respectively, and point A lies between points B and C; α and β are the central angles corresponding to point C; and R is the minimum radius of the Dubins curve.

3. The UAV swarm mission planning method based on the Dubins orientation angle model according to claim 1, characterized in that, For scenarios where the radius of flight of a target point intersects with that of the preceding and following target points, let A be the target point, and B and C be its preceding and following target points, respectively. When the center O1 of target point A is placed on its angle bisector, and the center O2 of the second target point B is placed on the angle bisector of the new angle formed by the tangent of the newly formed circle from point B and the previous point, the total flight distance L reaches its minimum value, expressed as: Where a, b, and c are the distances between points B and C, points A and C, and points A and B, respectively; α1 and β1 are the central angles corresponding to point A; α2 and β2 are the central angles corresponding to point B; and R is the minimum radius of the Dubins curve.

4. The UAV swarm mission planning method based on the Dubins orientation angle model according to claim 1, characterized in that, For scenarios where the target point is within the radius of the flight path of the preceding and following target points, the total flight distance L reaches its minimum when the values ​​of α, β, and θ are smaller, and α = β. The expression for this minimum is: Where a, b, and c are the distances between points B and C, A and C, and A and B, respectively; α, β, and θ are the central angles between points A and B; and R is the minimum radius of the Dubins curve.

5. The UAV swarm mission planning method based on the Dubins orientation angle model according to claim 1, characterized in that, In step 2, the objective function of the constructed UAV swarm mission planning mathematical model is expressed as: In the formula, L all k represents the total distance traveled. i,j The value is 0 or 1, representing whether the i-th UAV has a planned mission at the j-th target point, d i,j This represents the distance between two points after the drone swarm mission planning is completed. The objective function includes the following four constraints: First, ensure that each drone is assigned no fewer than k targets, where k is a custom value; Second, ensure that all target points are assigned; Third, ensure that each point is allocated only once; Fourth, it requires that the starting point and ending point of each aircraft be the same.

6. The UAV swarm mission planning method based on the Dubins orientation angle model according to claim 1, characterized in that, Step 3 includes: An improved whale algorithm based on extreme individual guidance is used to guide population mutation toward the global variable and the optimal individual during the target encirclement and prey search phases, respectively, in order to optimize the solution of the mathematical model for the UAV swarm mission planning.

7. The UAV swarm mission planning method based on the Dubins orientation angle model according to claim 6, characterized in that, An improved whale algorithm based on extreme individual guidance is used to guide population mutation towards the global variable and the optimal individual during the target encirclement and prey search phases, respectively, to optimize the solution of the UAV swarm mission planning mathematical model, including: a) Initialize the whale algorithm parameters, generate an initial population, set the size to N, the dimension to D, and the maximum number of iterations to T; b) Calculate the adaptive parameters of individual whales and calculate the optimal individual X. tbest Location; c) Determine the value of p. If p ≤ 0.5, proceed to step d); otherwise, proceed to step f. d) Determine the value of A. If |A|≤1, proceed to step e); otherwise, proceed to step g. e) After updating the global optimal individual position using the following formula, proceed to step h); X t+1 =X tbest -A*D; D=d*|C*X tbest -X t |+(1-d)*|C*X rand -X t |; Among them, X t+1 It is the (t+1)th generation population, X tbest Let X be the current globally optimal individual position, A and C be coefficients, and d be the guiding coefficient for the extreme value individual, taking the value as a random variable between [0,1]. t For an individual in the t-th generation of the population, X rand This refers to the location information of random individuals within the population. f) After updating the individual whale population locations using the following formula, proceed to step h); X t+1 =D circle *e cl *cos(2π*l)+X tbest ; D circle =|X tbest -X t |; Where c is a constant and l is a random number between [-1, 1]; g) Update the individual whale population locations using the following formula, then proceed to step h); X t+1 =X rand -A*D rand ; D rand =d*|C*X rand -X t |+(1-d)*|X tbest -X t |; h) Calculate the adaptive parameters of individual whales in the population, find the globally optimal individual, update the position of the optimal individual, and determine whether the maximum number of iterations has been reached. If so, the calculation ends and the optimal solution is output; otherwise, return to step c).

Images