Unmanned aerial vehicle group cooperative flight autonomous decision-making method based on improved particle swarm algorithm

By improving the particle swarm optimization algorithm and dynamic multi-swarm optimization strategy, the complex constraint optimization problem of UAV swarm cooperative flight was solved, realizing the three-dimensional trajectory planning and temporal and spatial coordination of UAV swarm, ensuring that the UAV swarm reaches the target point safely and efficiently.

CN119024692APending Publication Date: 2024-11-26SICHUAN UNIV

Patent Information

Application Number
CN202411113814.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-14
Publication Date
2024-11-26

AI Technical Summary

Technical Problem

Existing technologies struggle to effectively address the complex constraints of optimization during collaborative flight of drone swarms, particularly in three-dimensional path planning where they are prone to getting trapped in local optima and failing to achieve coordinated target strikes by multiple drones.

Method used

An improved particle swarm optimization algorithm is adopted to construct the cost function and constraints for UAV flight. Combined with a dynamic multi-swarm particle swarm optimization algorithm, the three-dimensional trajectory of the UAV swarm is planned through the update strategies of dominant group, inferior group and mixed group to meet the time and space cooperative constraints.

Benefits of technology

It improves the efficiency and accuracy of autonomous decision-making in the coordinated flight of UAV swarms, ensuring that UAV swarms can effectively avoid obstacles and reach the target point simultaneously, thus achieving efficient coordinated strikes.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119024692A_ABST
    Figure CN119024692A_ABST
Patent Text Reader

Abstract

The invention belongs to the technical field of unmanned aerial vehicle cluster cooperative autonomous flight, and particularly relates to an unmanned aerial vehicle cluster cooperative flight autonomous decision-making method based on an improved particle swarm algorithm. The problem of complex constraint optimization of unmanned aerial vehicle cluster collaborative flight path planning is solved, and for the complex optimization problem of collaborative flight path planning, the solving speed and accuracy of the optimization problem are improved by providing an improved particle swarm algorithm, establishing an autonomous decision problem model and giving related constraint conditions (such as terrain threats and meteorological threats). After the flight path planning of the cluster is completed, a multi-unmanned aerial vehicle time and space coordination constraint model is constructed, an alternative flight path group strategy is proposed to solve the time coordination problem, and meanwhile, an overall framework of a multi-unmanned aerial vehicle coordination flight path planning solution method is given.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention belongs to the technical field of cooperative autonomous flight of unmanned aerial vehicle swarms, and in particular relates to an autonomous decision-making method for cooperative flight of unmanned aerial vehicle swarms based on an improved particle swarm algorithm. Background Art

[0002] A drone is an aircraft operated by a remote control or autonomous control system, and is widely used in military, logistics, agriculture, photography, and surveying and mapping. Due to its flexibility and low cost, drones can enter areas that are difficult for humans to reach or dangerous when performing tasks, and achieve efficient data collection, monitoring, and transportation. With the advancement of technology, the automation and intelligence level of drones continues to improve, further expanding their application scope and practical value. In the modern military field, drones, as a new technology with huge threats and application potential, are highly valued by various countries. Compared with traditional manned combat, drone combat has the characteristics of low cost, no need for driver participation, high flexibility, small size, strong maneuverability, and the ability to flexibly operate in complex and dangerous environments and perform multiple tasks. With the continuous development of drone intelligent equipment, the future war pattern will be profoundly affected, and drone technology will also become a key research area for various countries.

[0003] With the widespread use of drones, the enemy's air defense system has become increasingly complex and complete, which has greatly restricted the penetration capability of drones. Simply improving the penetration capability of a single drone is no longer enough to cope with the current complex defense system. In this context, drone swarm collaborative operations are gaining increasing attention due to their stronger penetration capabilities. Through drone swarm collaborative operations, the overall combat effectiveness can be significantly enhanced, thereby breaking through the enemy's defense line more effectively. When multiple drones collaborate to explore, collaborative trajectory planning is an optimization problem with complex constraints such as terrain, threat area, time domain, airspace, and communication. Optimization problems with complex constraints are often difficult to solve, time-consuming, and prone to falling into local optimal solutions.

[0004] In order to effectively solve the optimization problem of rapid autonomous decision-making during coordinated flight of UAV swarms, which contains complex constraints, improve the coordinated combat capability between UAV swarms, and ensure efficient attack on the enemy, the study of the autonomous decision-making method of coordinated flight of unmanned swarms based on the improved particle swarm algorithm will provide strong technical support for solving the problem of rapid autonomous decision-making during coordinated flight of UAV swarms.

[0005] The current methods are:

[0006] 1) Three-dimensional path planning based on improved particle swarm algorithm "Xie Yonghong, Kong Yueping. Three-dimensional path planning based on improved particle swarm algorithm [J]. Computer Measurement and Control, 2022, 30(03): 179-182+191"; This method is easy to fall into the local optimal problem when the particle swarm algorithm is used to solve the three-dimensional path planning problem of UAVs in construction projects. A particle swarm optimization algorithm with mixed inertial traction is proposed. By introducing an adaptive initialization mechanism in the initial stage and the position relationship between the starting point and the target point, the initial population of the particle swarm is optimized; the linear decreasing inertia weight method is adopted to enhance the global search performance in the early stage and the local search performance in the later stage of the algorithm; the acceleration is introduced in the speed update with the help of the universal gravitation idea to enhance the search performance. The improved algorithm with and without the adaptive initialization mechanism is used for comparative experiments. The results verify that the mechanism is more conducive to improving the solution quality of the algorithm. However, its disadvantages are: (1) This method is for the three-dimensional path planning of a single UAV, and does not consider the three-dimensional path planning problem of a UAV cluster; (2) No UAV cluster coordination scheme is given to enable the UAV cluster to coordinate and strike the target point.

[0007] 2) Multi-UAV collaborative trajectory planning method based on PSO-HJ algorithm "Shan Wenzhao, Cui Naigang, et al. Multi-UAV collaborative trajectory planning method based on PSO-HJ algorithm [J]. Journal of Chinese Inertial Technology, 2020, 28(01): 122-128"; This method aims to solve the problems of large computational complexity and difficulty in convergence in multi-UAV collaborative trajectory solution, and proposes a multi-UAV time collaborative three-dimensional trajectory planning method based on the fusion of particle swarm optimization and Hook-Jeeves search algorithm. First, a single UAV trajectory planning solution model is established. Then, by introducing the Hooke-Jeeves search algorithm for particles with low fitness evaluation function values, the particle diversity is increased and the convergence of the trajectory planning algorithm is improved; the constraint violation function is introduced for particles that do not meet the constraints, and a new particle evaluation mechanism is proposed based on the comparison criterion to promote the particle search for the optimal solution located at the constraint boundary, thereby accelerating the computational efficiency of the trajectory planning algorithm. Finally, a multi-UAV time collaborative trajectory planning solution algorithm was designed. The PSO-HJ algorithm was used to solve the trajectory information of each UAV separately, and the collaborative trajectory planning was realized by coordinating the arrival time through the centralized trajectory planning layer of multiple UAVs. The disadvantages of this method are: (1) It does not consider the spatial constraints of the UAV cluster, that is, the possibility of cluster collision during flight; (2) When the time constraint is not met, no specific method for reselecting the trajectory group is given; (3) No solution for the suboptimal trajectory group is given. Summary of the invention

[0008] In response to the complex constraint optimization problem of autonomous decision-making in cooperative flight of drone clusters, the present invention established a constraint model for autonomous flight of drones, designed the corresponding objective function, and then established a time and space cooperative constraint model to significantly improve the efficiency of autonomous decision-making in cooperative flight of drone clusters.

[0009] The technical solution adopted by the present invention is:

[0010] The autonomous decision-making method for cooperative flight of UAV swarm based on improved particle swarm algorithm is characterized by comprising the following steps:

[0011] S1. Construct the cost function of three-dimensional path planning of UAVs based on the flight space and obstacle area of ​​the UAV swarm:

[0012] J i =α1J i,1 +α2J i,2 +α3J i,3

[0013]

[0014]

[0015] Among them, J i,1 is the cost function of the UAV track length, P i,j is the jth track node of the i-th UAV, expressed as a three-dimensional coordinate P i,j =(x i,j ,y i,j ,z i,j ), ||...|| means finding the distance between two points, n is the number of track nodes; J i,2 is the obstacle area cost function, is a function for calculating the risk of collision, d m =||P i,j ,Rhreat m || is the distance from the track point to the center of the obstacle, D is the diameter of the drone, R m is the radius of the obstacle, Rthreat m is the coordinate of the obstacle center point, m=1,2,...,M, M is the number of obstacle areas; J i,3 is the cost function of the UAV flight altitude, △z j is the height difference between the two track points; α1, α2, α3 are weight coefficients, and the sum is 1.

[0016] Set the following constraints for the cost function:

[0017] h min ≤z i ≤h max

[0018]

[0019] Among them, z i is the height coordinate, h min is the minimum flight altitude, h max is the maximum flight altitude; μ min is the minimum pitch angle, μ max is the maximum pitch angle; f1=(x i -x i-1 )(x i+1 -x i )+(y i -y i-1 )(y i+1 -y i ),

[0020] Used to calculate the yaw angle, β min is the minimum yaw angle, β max is the maximum yaw angle.

[0021] S2. Solve the constructed cost function based on the improved particle swarm algorithm. The specific method is as follows:

[0022] The cost function established above is used as the fitness function of the particle, as shown in the following formula. After randomly initializing the fitness function value, it is brought into the multi-swarm particle swarm optimization method for iteration and the fitness value and global optimal solution under each iteration number are recorded.

[0023] fit(x g )=α1J i,1 +α2J i,2 +α3J i,3

[0024] Among them, fit(x g ) is the function for calculating the fitness value, x g is the position vector of the particle in the D-dimensional search space.

[0025] The traditional particle swarm is divided into dominant swarm, inferior swarm and mixed swarm according to the fitness value of the current iteration. Dominant swarm particle update strategy:

[0026]

[0027] Among them, x g (t) and are the particle positions before and after the Levy flight update at the tth iteration, α is the step size control factor, which is generally 0.01. The small and large steps are used to fly to the original low-probability exploration area to make the search area more uniform; Levy(λ) is the random search path, stands for dot product; The particle position after the greedy algorithm updates.

[0028] Disadvantage group particle update strategy:

[0029]

[0030] x new =x×(1+N(0,1))

[0031] Among them, particle x i The best historical position is The global optimal position of the population is The combined particle is denoted as p mix (t), c1 and c2 are the self-learning factor and the global learning factor respectively, c3 is the mixed learning factor, r1, r2 and r3 are random numbers in the interval (0, 1) and are uniformly distributed; r and N(0, 1) are random numbers between [0, 1], x is the position of the particle, x new is the position of the particle after mutation, t max is the maximum number of iterations for the particle.

[0032] Mixed group particle update strategy:

[0033]

[0034] Among them, ω is the inertia weight, ω max is the initial inertia weight, ω min is the inertia weight when it iterates to the maximum, generally set to ω max =0.9,ω min =0.4.

[0035] By using the established cost function as the fitness function of the particle, randomly initializing the fitness function value and then bringing it into the multi-swarm particle swarm optimization method for iterative solution, the optimal solution of the particle swarm is obtained. Since each particle represents a three-dimensional track node of a UAV, the optimal three-dimensional track of the UAV is obtained.

[0036] S3, plan alternative flight paths for the drone cluster, and the planning process requires the spatial coordination C s :

[0037] In the process of alternative trajectory planning, m flight routes are planned in sequence for the n drones in the drone group according to step S2. In order to ensure the richness of the trajectory, the maximum number of repeated trajectory points R is set. max When the number of common track points between two tracks in the candidate track group is greater than R max When , one of the trajectories is replanned. It is necessary to ensure that the trajectories between different drones meet the spatial constraints:

[0038]

[0039] Among them, D s ≤d i,j ≤D c , d i,j is the distance between the i-th UAV and the j-th UAV, D s is the minimum anti-collision distance, D c is the maximum communication distance;

[0040] S4. Time coordination based on the planned alternative trajectory C t :

[0041] After planning all the alternative routes for the drone swarm, calculate the arrival time interval of each drone along the alternative route. v i ∈[v min ,v max ] is the speed range of the i-th UAV, l i,j is the length of the jth alternative track. Find the union of the time intervals of each drone AT i , to conduct temporal coordination of UAV clusters.

[0042]

[0043] in, It is the intersection of the time intervals of all drones in the cluster, and uav_num is the number of drones.

[0044] Check whether the intersection of the arrival time intervals of the drone cluster is an empty set. If the intersection is an empty set, it means that there is no common arrival time and the track needs to be reselected; if the intersection is not an empty set, an optional track group that meets the coordination constraints is obtained.

[0045] S5. Set the flight parameters of the UAV according to the obtained UAV three-dimensional path planning track group:

[0046] The minimum time set of the arrival time interval is selected as the arrival time (AT) and the corresponding track. According to the determined arrival time (AT), the required flight speed of each UAV is calculated.

[0047] Beneficial effects of the present invention:

[0048] (1) The present invention solves the complex constraint optimization problem of collaborative trajectory planning of UAV clusters. For the complex optimization problem of collaborative trajectory planning, an improved particle swarm algorithm is proposed, and an autonomous decision-making problem model is established. Relevant constraint conditions (such as terrain threats, meteorological threats, etc.) are given, thereby improving the speed and accuracy of solving the optimization problem.

[0049] (2) After the trajectory planning of the cluster is completed, a multi-UAV time and space coordination constraint model is constructed, and an alternative trajectory group strategy is proposed to solve the time coordination problem. At the same time, an overall framework of the multi-UAV collaborative trajectory planning solution is given. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 Schematic diagram of drone arrival time.

[0051] Figure 2 Schematic diagram of the combined particle principle.

[0052] Figure 3 Flowchart for collaborative trajectory planning for drone swarms.

[0053] Figure 4 Schematic diagram of the overall framework of collaborative trajectory planning for UAV swarms.

[0054] Figure 5 A collaborative trajectory planning view for a swarm of drones.

[0055] Figure 6 It is the iterative fitness change diagram of IDM-PSO and PSO. DETAILED DESCRIPTION

[0056] The technical solution of the present invention is described in detail below.

[0057] The present invention models the problem of autonomous decision-making for coordinated flight of drone swarms, which is to plan a reasonable trajectory of the drone swarm from the starting point to the target point, ensuring that the swarm can avoid threat areas (such as terrain threats, weather threats, etc.) during flight, and at the same time, the swarm must maintain coordination during flight and can reach the target point at the same time. This can be regarded as solving a multi-constraint function optimal problem.

[0058] Specifically include:

[0059] (1) Optimization indicators

[0060] Track length cost. The fuel consumption of a drone during flight is somewhat related to the length of the flight. Minimizing the track length is used as the optimization indicator, so the track length cost of the i-th drone can be expressed as follows:

[0061]

[0062] Among them, P j Represents the track node in the UAV flight trajectory. The UAV track is composed of n track nodes P j By connecting the track points, the starting point and the target point, a flyable path for the UAV can be obtained. i ,P i,1,...,P i,n ,E i}, each track node can be represented as P i,j =(x i,j ,y i,j ,z i,j ), ||...|| means finding the distance between two points.

[0063] Threat cost. In addition to the optimal track length, the planned trajectory must also ensure the safety of the drone during flight. In the combat space, there are obstacle areas that will damage the drone and make it impossible for the drone to effectively strike the enemy target point.

[0064] Considering the complexity of obstacle area modeling and the difficulty of obtaining real data, the present invention abstracts the obstacle environment and abstracts the obstacle area into a cylinder with a constant radius. The effective radius of the obstacle area is equal to the radius of the cylinder. Assuming there are M obstacle areas, in order for the drone to successfully avoid the obstacle area and achieve effective attack on the enemy, the following threat cost is defined:

[0065]

[0066] When the drone flies outside the obstacle area, its threat cost is zero; when the drone enters the obstacle area, the drone has a high possibility of destruction, and the threat cost of the path is set to infinite.

[0067] Flight altitude cost. In order to meet the safety and concealment requirements of the UAV during flight and reduce the probability of being detected by radar, a low-altitude flight strategy is adopted, which requires that the UAV should not have drastic altitude changes during flight. At the same time, frequent flight altitude changes will also increase energy consumption. Therefore, the following altitude cost function is defined:

[0068]

[0069] Among them, △z j is the height difference between the two track points.

[0070] Overall cost function. Combining the path trajectory cost, threat cost and height cost mentioned above, the cost function of the three-dimensional path planning of the UAV can be defined as follows:

[0071] J i =α1J i,1 +α2J i,2 +α3J i,3 (4)

[0072] Among them, α1, α2, α3 are weight coefficients, and the sum of them is 1.

[0073] (2) Constraints

[0074] Flight altitude constraint: To prevent the drones from being discovered during flight, the drone cluster of the present invention adopts a low-altitude flight strategy, so the flight altitude of the drones needs to be restricted.

[0075] h min ≤z i ≤h max (5)

[0076] Among them, h min is the minimum flight altitude, h max The maximum flight altitude.

[0077] Yaw angle constraint. Due to the limitations of the drone's own mechanical properties, its yaw angle should be within its maximum allowable range:

[0078]

[0079] Among them, β min is the minimum yaw angle, μ max is the maximum yaw angle.

[0080] Pitch angle constraint. Due to the limitations of the drone's own mechanical properties, its yaw angle should be within its maximum allowable range:

[0081]

[0082] Among them, μ min is the minimum pitch angle, μ max is the maximum pitch angle.

[0083] (3) Collaborative constraints

[0084] Time coordination constraint. Time coordination constraint specifically refers to the process of UAV swarm trajectory planning. In order to achieve a saturation strike on the enemy, the flight time of each UAV in the mission cycle should be reasonably coordinated so that the UAVs can arrive at the target point at the same time to strike the enemy. Therefore, it is necessary to constrain the arrival time (AT) of the UAVs.

[0085] Take three drones as an example. Figure 1 As shown, the part between the red lines is the time intersection AT when the drone reaches the target.

[0086] Let the speed of the i-th drone be v i ∈[v min ,v max ], each UAV generates k alternative tracks in advance, and the length of the jth alternative track is recorded as l i,j , then the time interval for the i-th UAV to reach the target point is The total time interval to reach the target point is:

[0087] AT i,j =AT i,1 ∪AT i,2 ∪...∪AT i,k (8)

[0088] Therefore, the time coordination is:

[0089]

[0090] Among them, uav_num is the number of drones.

[0091] Spatial coordination constraints. During the flight of a drone swarm, the communication interaction between drones must be considered, and any two drones must meet the set communication range. And for safety reasons, the real-time minimum distance between drones must be considered.

[0092] Therefore, the maximum communication distance and minimum anti-collision distance constraints that can be used between the two machines are set.

[0093] D s ≤d i,j ≤D c

[0094]

[0095] Among them, d i,j is the distance between the i-th UAV and the j-th UAV.

[0096] Solving the trajectory coordination problem based on dynamic multi-swarm particle swarm algorithm.

[0097] In the particle swarm optimization algorithm, each particle represents a solution. In the D-dimensional search space, the position vector of the i-th particle is x i =(x i1 ,x i2 ,...,x iD ), the velocity vector of the i-th particle is v i =(v i1 ,v i2 ,...,v iD ), after t+1 iterations, particle x i The best historical position is The global optimal position of the population is The specific update method is:

[0098]

[0099] Among them, ω is the inertia weight, which is a non-negative number that decreases linearly. maxis the initial inertia weight, ω min is the inertia weight when it iterates to the maximum, generally set to ω max =0.9,ω min =0.4; c1 and c2 are the self-learning factor and the global learning factor, respectively; r1 and r2 are random numbers in the interval (0,1) and are uniformly distributed; M is the population size (i.e., the number of particles in the population); D is the dimension of the particle; t max is the maximum number of iterations for the particle.

[0100] When solving some complex optimization problems, the PSO algorithm, like other algorithms, has the defects of being difficult to get rid of local optimality, low execution efficiency, and difficulty in balancing global search capabilities and local search capabilities. To solve these problems, the present invention introduces a dynamic multi-population particle swarm algorithm, which divides the entire population into multiple sub-populations, and then designs an information interaction mechanism between sub-populations to enable multiple sub-populations to execute in parallel while improving the performance of the algorithm. This parallelization strategy is also called a multi-population strategy.

[0101] (1) Dominant group particle update strategy

[0102] The main reason why the particle swarm algorithm converges slowly in the later stage of optimization is that it is difficult to get rid of the current local extreme value, resulting in a decrease in accuracy. In order to enhance the ability of the dominant group to jump out of the local optimum, Levy flight is introduced. Levy flight is carried out in the form of large and small step intervals, which can enhance the activity and jumping ability of particles, expand the search range, improve diversity, and avoid falling into the local optimum. The update formula is as follows:

[0103]

[0104] Among them, x g (t) and are the particle positions before and after the Levy flight update at the tth iteration, α is the step size control factor, which is generally 0.01. The large and small steps are used to fly to the original low probability exploration area to make the search area more uniform. Levy(λ) is the random search path, Represents dot product.

[0105] Levy flight helps particles get rid of local optimality, but the updated position may not be better. This paper introduces a greedy algorithm evaluation strategy, which only updates when the updated position is better, otherwise the original position is retained. The implementation process is as follows:

[0106]

[0107] in, is the particle position after the greedy algorithm updates, and f(x) represents the particle fitness function.

[0108] (2) Disadvantaged group particle update strategy

[0109] In the update of the "disadvantaged group", the concept of combined particles is introduced to change the traditional speed update formula. The combined particle is recorded as, and the dimension value of is randomly combined by the historical optimal values ​​of each particle, such as Figure 2 As shown. The update method of the disadvantaged group position and speed is obtained as follows:

[0110]

[0111] Among them, is the hybrid learning factor, and is a random number. As can be seen from the formula, the combined particle is composed of the current optimal individual particles, which not only inherits the excellent dimensions of each particle, but also has randomness, taking into account the diversity and excellence of particles.

[0112] In the "inferior group", since there is relatively little information about particles that are worth retaining, the population is far away from the high-quality solution to the problem. Therefore, after the update is completed, this paper also introduces a Gaussian mutation mechanism. This mutation mechanism can widely explore various possible solution areas in the entire search space, thereby effectively avoiding premature convergence and further enhancing the diversity of the sub-population. The specific mutation process is shown in the following formula:

[0113]

[0114] Among them, r and N(0,1) are random numbers between [0,1], x is the position of the particle, and x new is the position of the particle after mutation.

[0115] (3) Mixed group particle update strategy

[0116] The mixed group is between the dominant group and the inferior group. Since the differences between individuals are large in the early stage of the algorithm, the mixed group focuses on its cognitive part in the early stage, which can achieve multi-party communication; in the later stage, it strengthens the leading power of the global extreme value and encourages particles to gather around the optimal solution. Therefore, the learning factor introduces cosine and sine functions, so that the self-learning factor decreases monotonically and the population learning factor increases monotonically. The speed update formula of the mixed group is:

[0117]

[0118] Although the optimal result of single UAV trajectory planning based on the improved particle swarm algorithm can meet the various constraints of a single UAV, it cannot meet the time and space constraints in multi-UAV collaboration. Therefore, in order to meet the time and space constraints, multiple trajectories should be designed for each UAV in advance for selection. Based on the improved particle swarm algorithm, multiple flight routes are planned for UAVs by reasonably modifying parameters and constraints. The main purpose of this method is to obtain distributed trajectories in a larger range, rather than the global optimal trajectory.

[0119] In the process of alternative trajectory planning, m flight routes are planned in sequence for the n drones in the drone swarm. In order to ensure the richness of the trajectory, the maximum number of repeated trajectory points R is set. max When the number of common track points between two tracks in the candidate track group is greater than R max In addition, it is also necessary to ensure that the trajectories between different drones satisfy equation (10), i.e., the spatial constraint, to ensure the overall coordination of the drone swarm.

[0120] The specific steps of the alternative trajectory group planning method are:

[0121] 1) Initialize parameters and assign values ​​to m, n, and D c , D s , R max .

[0122] 2) Plan the trajectory of the first UAV based on the dynamic multi-swarm particle swarm algorithm proposed in the previous article.

[0123] 3) When i≠1, determine whether the distance between the current UAV and the previous UAV satisfies the spatial constraint of formula (10). If yes, continue to execute step 4), otherwise, re-execute step 2).

[0124] 4) Calculate the number of repeated track points between the current track of the i-th UAV and the previous track. If it is less than R max Then execute step 5) and record m=m-1, otherwise execute step 2) again.

[0125] 5) Calculate the number m of the tracks of the i-th UAV. When m=0, the track planning of the i-th UAV is completed, and i=i+1 is recorded, and continue to step 6), otherwise re-execute step 2).

[0126] 6) When i=n, ​​the alternative track group planning is completed and the result is output, otherwise step 2) is executed again.

[0127] After planning the alternative routes of the drone cluster, collaborative route planning is performed. The collaborative planning process is as follows: Figure 3 As shown:

[0128] 1) Plan a set of alternative trajectories for each UAV.

[0129] 2) Calculate the arrival time interval of each UAV when flying along the alternative trajectory.

[0130] 3) Check whether the intersection of the arrival time intervals of all drones is an empty set. If the intersection is an empty set, it means that there is no common arrival time, and you need to reselect the track and return to step 1); if the intersection is not an empty set, proceed to the next step.

[0131] 4) Select the minimum time set of the arrival time interval as the arrival time (AT).

[0132] 5) Based on the determined arrival time (AT), calculate the required flight speed for each UAV.

[0133] 6) Assign the calculated flight speed to each drone.

[0134] In summary, the framework of the UAV swarm collaborative trajectory planning method based on dynamic multi-population particle swarm algorithm (IDM-PSO) is as follows: Figure 4 shown.

[0135] The effect of the present invention is demonstrated through simulation below.

[0136] Five drones avoid three threat zones from their initial positions and arrive at the target point to attack at the same time. The initial position and terminal time coordinates of each drone are shown in Table 1, and the coordinates of the threat zone are shown in Table 2. The speed range of the drone is [200,280] m / s.

[0137] Table 1 The starting and ending locations of the five drones

[0138]

[0139] The optimal coordinated trajectory of the five UAVs is shown in Figure 5 As shown in the figure, the required trajectories do not cross the threat area and reach the target point, satisfying the constraints.

[0140] Table 3 shows the coordinated trajectory planning information of the drone swarm. The drone’s trajectory length is not completely the optimal trajectory, but the global optimal result after comprehensive consideration of combat requirements and spatiotemporal coordination constraints. Through the planned coordination time interval, the corresponding flight speed is set for each drone, so that the drone swarm can arrive at the target area synchronously at the same time or within a specified time period. Table 4 shows the speeds assigned to the drones.

[0141] Table 2 Coordinates and radius settings of the three threat zones

[0142]

[0143] Table 3 UAV cluster track coordination information

[0144]

[0145] Table 4. Speed ​​distribution of drone clusters

[0146]

[0147] The dynamic particle swarm algorithm is used to simulate the collaborative trajectory planning of five UAVs in three threat areas. The results show that the UAV trajectories can be planned and successfully avoid the threat areas, and the collaborative trajectories can reach the target area at the same time to attack the enemy.

[0148] At the same time, in order to compare the efficiency of the improved dynamic multi-population particle swarm optimization algorithm, this paper compares and analyzes the iterative fitness changes of the improved algorithm and the traditional algorithm under the same population size and iteration number parameter settings, such as Figure 6 As shown in the figure, it can be found that the convergence fitness of the improved particle swarm optimization algorithm is about 885, while the convergence fitness of the traditional algorithm is 924. Compared with the PSO algorithm, the IDM-PSO algorithm proposed in this paper can jump out of the local optimal solution and find the global optimal solution, with better convergence effect and convergence accuracy.

Claims

1. Autonomous decision-making method for cooperative flight of UAV swarms based on improved particle swarm algorithm, It is characterized in that The following steps are involved: S1. Construct the cost function of three-dimensional path planning of UAVs based on the flight space and obstacle area of ​​the UAV swarm: minJ i =a 1 J i,1 +a 2 J i,2 +a 3 J i,3 Among them, J i,1 is the cost function of the UAV track length, P i,j is the jth track node of the i-th UAV, expressed as a three-dimensional coordinate P i,j =(x i,j ,y i,j ,z i,j ), ||...|| means finding the distance between two points, n is the number of track nodes; J i,2 is the obstacle area cost function, is a function for calculating the risk of collision, d m =||P i,j ,Rhreat m || is the distance from the track point to the center of the obstacle, D is the diameter of the drone, R m is the radius of the obstacle, Rthreat m is the coordinate of the obstacle center point, m=1,2,...,M, M is the number of obstacle areas; J i,3 is the cost function of the UAV flight altitude, △z j is the height difference between the two track points; α 1 ,α 2 ,α 3 It is a weight coefficient, and the sum is 1; Set the following constraints for the cost function: h min ≤z i ≤h max Among them, z i is the height coordinate, h min is the minimum flight altitude, h max is the maximum flight altitude; μ min is the minimum pitch angle, μ max is the maximum pitch angle; f 1 =(x i -x i-1 )(x i+1 -x i )+(y i -y i-1 )(y i+1 -y i ), Used to calculate the yaw angle, β min is the minimum yaw angle, β max is the maximum yaw angle; S2. Solve the constructed cost function based on the improved particle swarm algorithm. The specific method is as follows: The established cost function is used as the fitness function of the particle. After randomly initializing the fitness function value, it is brought into the multi-swarm particle swarm optimization method for iteration and the fitness value and global optimal solution under each iteration number are recorded: fit(x g )=a 1 J i,1 +a 2 J i,2 +a 3 J i,3 Among them, fit(x g ) is the function for calculating the fitness value, x g is the position vector of the particle in the D-dimensional search space; The traditional particle swarm is divided into dominant group, inferior group and mixed group according to the fitness value of the current iteration; the dominant group particle update strategy is: Among them, x g (t) and are the particle positions before and after Levy flight update at the tth iteration, α is the step size control factor; Levy(λ) is the random search path, stands for dot product; The particle position after the greedy algorithm updates; Disadvantage group particle update strategy: x new =x×(1+N(0,1)) Among them, particle x i The best historical position is The global optimal position of the population is The combined particle is denoted as p mix (t), c 1 and c 2 They are the self-learning factor and the global learning factor, c 3 is the hybrid learning factor, r 1 、r 2 and r 3 is a random number in the interval (0,1) and is uniformly distributed; r and N(0,1) are random numbers between [0,1], x is the position of the particle, x new is the position of the particle after mutation, t max is the maximum number of iterations of the particle; Mixed group particle update strategy: Among them, ω is the inertia weight, ω max is the initial inertia weight, ω min is the inertia weight when the iteration reaches the maximum; The established cost function is used as the fitness function of the particle, and the fitness function value is randomly initialized and then brought into the multi-swarm particle swarm optimization method for iterative solution to obtain the optimal solution of the particle swarm. Since each particle represents a three-dimensional track node of a UAV, the optimal three-dimensional track of the UAV is obtained. S3, plan alternative flight paths for the drone cluster, and the planning process requires the spatial coordination C s : In the process of alternative trajectory planning, m flight routes are planned in sequence for n drones in the drone group according to step S2; in order to ensure the richness of the trajectory, the maximum repeated trajectory point R is set. max , when the number of common track points between two tracks in the candidate track group is greater than R max , one of the trajectories is replanned; it is necessary to ensure that the trajectories between different drones meet the spatial constraints: Among them, D s ≤d i,j ≤D c , d i,j is the distance between the i-th UAV and the j-th UAV, D s is the minimum anti-collision distance, D c is the maximum communication distance; S4. Time coordination based on the planned alternative trajectory C t : After planning all the alternative routes for the drone swarm, calculate the arrival time interval of each drone along the alternative route. v i ∈[v min ,v max ] is the speed range of the i-th UAV, l i,j is the length of the jth alternative track; find the union of the time intervals of each drone AT i , to conduct time coordination of drone clusters: in, is the intersection of the time intervals of all drones in the cluster, and uav_num is the number of drones; Check whether the intersection of the arrival time intervals of the drone cluster is an empty set. If the intersection is an empty set, it means that there is no common arrival time and the track needs to be reselected. If the intersection is not an empty set, an optional track group that meets the coordination constraints is obtained. S5. Set the flight parameters of the UAV according to the obtained UAV three-dimensional path planning track group: The minimum time set of the arrival time interval is selected as the arrival time and the corresponding track, and the required flight speed of each UAV is calculated based on the determined arrival time.

Citation Information

Patent Citations

  • Unmanned aerial vehicle flight path planning method based on improved quantum particle swarm optimization

    CN116430897A

  • Multi-unmanned aerial vehicle cooperative route planning method based on reconstruction hybrid dynamic strategy and multi-population difference algorithm

    CN117908580A

  • Method and system for planning autonomous homeward flight path of broken stock repair unmanned aerial vehicle

    CN118376242A

Cited By

  • Multi-vertical fixed-wing unmanned aerial vehicle flight path planning and intelligent obstacle avoidance method

    CN119690112A

  • Multi-unmanned aerial vehicle cooperative path planning method based on adaptive multi-population hybrid algorithm

    CN119828765A

  • Multi-uav cooperative path planning method based on adaptive multi-population hybrid algorithm

    CN119828765B

  • Unmanned aerial vehicle cluster flight path visual simulation method based on Unity3D

    CN120044813A

  • UAV three-dimensional path obstacle avoidance method based on multi-strategy fusion improved arithmetic optimization algorithm

    CN120631006A