A Three-Dimensional Cooperative Coverage Planning Method for UAV Swarms Based on Multi-Subgroup Gray Wolf Optimization

Abstract

This invention belongs to the field of autonomous control and path planning technology for unmanned aerial vehicle (UAV) swarms. Specifically, it discloses a three-dimensional collaborative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization. The method includes establishing a digital elevation map of a three-dimensional multi-source threat environment; preprocessing the environment using dynamic Poisson disk sampling and Voronoi diagrams; based on the preprocessed environment, performing balanced task allocation for the UAV swarm using a sector migration clustering strategy; based on the balanced task allocation results, performing global coverage path planning for the UAV swarm using an improved multi-subgroup gray wolf optimization algorithm, and optimizing local paths using an adaptive elastic window local greedy method; and performing UAV swarm communication and smooth path output based on differential communication to complete the three-dimensional collaborative coverage planning for the UAV swarm. This invention solves the problems of existing technologies, such as difficulty in reflecting the real operating environment, severe track overlap, and uneven task load.

Description

Technical Field

[0001] This invention belongs to the field of autonomous control and path planning technology for unmanned aerial vehicle (UAV) swarms, specifically involving a three-dimensional collaborative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization. Background Technology

[0002] With the development of artificial intelligence, automatic control, sensing, and communication technologies, drones, with their advantages of high mobility, fast response speed, low cost, and high safety, are widely used in environmental monitoring, aerial surveying, civil security, weather warning, power line inspection, and other fields, playing a significant socio-economic role. In recent years, mission environments have become increasingly complex, and single drones can hardly meet the operational needs of some scenarios. Therefore, the industry typically adopts drone swarm collaboration to improve mission efficiency.

[0003] Existing UAV swarm coverage planning methods are mostly based on two-dimensional spatial models or grid maps, which are difficult to describe real three-dimensional terrain features such as mountain undulations and canyon obstructions. In areas with drastic slope changes, this can easily lead to problems such as unreachable paths, reduced coverage, and increased energy consumption, posing flight safety hazards. Moreover, traditional methods do not consider the impact of terrain complexity on path planning and airspace allocation, making it difficult to support subsequent conflict detection and obstacle avoidance. Therefore, UAV swarm collaborative coverage path planning methods for real three-dimensional terrain still need further research. However, current technical means are mostly limited to two-dimensional no-fly zone settings or static penalty constraints, lacking three-dimensional avoidance mechanisms. This causes UAV tracks to frequently cross the boundaries of threat areas, posing safety hazards and reducing the overall coverage efficiency of the mission. At the same time, UAV swarm collaborative coverage planning needs to achieve multi-aircraft task allocation and track decoupling. Existing centralized or greedy allocation strategies are prone to uneven task load, track overlap, and low resource utilization. Furthermore, existing information sharing mechanisms, fitness function design, and global and local search coordination mechanisms are insufficient, affecting the overall efficiency of UAV swarm missions.

[0004] To address the technical problems of existing technologies, such as difficulty in reflecting the real operating environment, severe track overlap, and uneven task load, this invention proposes a three-dimensional collaborative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization. Summary of the Invention

[0005] The purpose of this invention is to solve the problems of existing technologies that fail to reflect the real operating environment, have severe track overlap, and have uneven task load. It proposes a three-dimensional collaborative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization.

[0006] The technical solution of this invention is: a three-dimensional cooperative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization, comprising the following steps:

[0007] Establish a three-dimensional digital elevation map of the multi-source threat environment;

[0008] Based on digital elevation maps, dynamic Poisson disk sampling and Voronoi diagrams are used for environmental preprocessing.

[0009] Based on the preprocessed environment, a clustering strategy using sector migration is used to achieve balanced task allocation for the UAV swarm.

[0010] Based on the balanced task allocation results, the improved multi-subgroup gray wolf optimization algorithm is used to plan the global coverage path of the UAV swarm and obtain the global path.

[0011] The multi-objective cost function is used as the evaluation index for global path selection of UAV swarm, and an adaptive elastic window and potential field obstacle avoidance strategy are designed to optimize the local path of UAV swarm.

[0012] UAV swarm communication is based on differential communication, and the global path of the UAV swarm after local path optimization is smoothly output to complete the three-dimensional collaborative coverage planning of the UAV swarm.

[0013] The beneficial effects of this invention are:

[0014] 1. By adopting a clustering strategy based on sector migration, the task area is dynamically subdivided, enabling each UAV to cover the entire sub-area during the planning phase and eliminating coverage blind spots.

[0015] 2. An improved multi-subgroup gray wolf optimization algorithm is used to generate a global skeleton track, combined with local shaping methods such as 2-Opt and curvature smoothing to reduce the length of the covered path and path overlap.

[0016] 3. By using sector migration clustering strategies and experience sharing mechanisms, the task allocation and capabilities of each UAV are matched, reducing path variance, avoiding overload and idleness, and achieving load balancing of the UAV cluster.

[0017] 4. A differential communication mechanism is adopted, which only exchanges incremental data, ensuring that the local task can be completed independently in the event of a single machine losing connection. After reconnection, information can be quickly synchronized, reducing the system communication load and improving system robustness. Attached Figure Description

[0018] Figure 1 The diagram shows a flowchart of a three-dimensional collaborative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization.

[0019] Figure 2 The diagram shown is a schematic of the multi-source threat environment provided in Example 2.

[0020] Figure 3 The diagram shown is a schematic diagram of the environmental pretreatment provided in Example 2.

[0021] Figure 4 The diagram shown is a schematic of the improved multi-subgroup gray wolf optimization global planning provided in Example 2.

[0022] Figure 5 The diagram shown is a schematic diagram of the drone swarm coverage trajectory provided in Example 2.

[0023] Figure 6 The figure shown is a schematic diagram of the fitness curve provided in Example 2.

[0024] Figure 7 The diagram shown is a schematic diagram of the total flight path length of the UAV cluster provided in Example 2.

[0025] Figure 8 The figure shown is a schematic diagram of the drone cluster coverage curve provided in Example 2. Detailed Implementation

[0026] Exemplary embodiments of the present invention will now be described in detail with reference to the accompanying drawings. It should be understood that the embodiments shown and described in the drawings are merely exemplary and are intended to illustrate the principles and spirit of the invention, and are not intended to limit the scope of the invention.

[0027] Example 1:

[0028] like Figure 1 As shown, a three-dimensional cooperative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization includes the following steps:

[0029] S1. Establish a three-dimensional digital elevation map of the multi-source threat environment;

[0030] Specifically, this invention considers baseline terrain, mountainous terrain, and threat areas in the environment for environmental modeling. The baseline environmental model is represented as follows:

[0031]

[0032] in, and These are the coordinates of the 3D environment projected onto the horizontal plane. coordinate points in the reference environment The corresponding elevation value; A constant coefficient to control for fluctuations in the baseline environment; Represents the sine function; Represents the cosine function;

[0033] For steep natural terrain in a 3D environment, an exponential function is used for modeling, specifically expressed as:

[0034]

[0035] in, Represents coordinate points in a mountainous environment The corresponding elevation value; It is the first Projected coordinates of the center point of each mountain peak; Terrain parameters for controlling altitude; and For the first The mountain peaks along shaft and The amount of attenuation of the shaft; This represents the total number of mountain peaks. This represents an exponential function.

[0036] In addition to mountainous terrain, the environment includes threat zones such as threat areas and radar jamming zones, which are represented as follows:

[0037]

[0038]

[0039] in, and These represent the effective range of the threat zone and the effective range of the radar jamming zone, respectively. This represents the distance between a point within the radar interference range and the interference source. and Projecting the threat center and radar jamming source onto Coordinates on the plane; and These are the threat zone radius and the effective radius of radar jamming, respectively. The vertical upper limit of the interception envelope or no-fly zone; To threaten regional security margins; Indicates the first One threat source index; Indicates the first Index of radar interference sources.

[0040] By overlaying environmental elements such as baseline topography, mountainous terrain, threat zones, and radar jamming zones, a digital elevation map of the three-dimensional environment of multi-source threats is created, represented as follows:

[0041]

[0042] in, A digital elevation map representing a three-dimensional environment model.

[0043] S2. Based on digital elevation maps, dynamic Poisson disk sampling and Voronoi diagrams are used for environmental preprocessing;

[0044] Specifically, feasible regions are sampled using dynamic Poisson disks. Sampling is performed, taking into account terrain slope and threat areas, to generate a seed point set with uniform and adaptive density. First, a digital elevation map of the 3D environment model is generated. Calculate the amplitude of terrain curvature :

[0045]

[0046] in, Indicates the partial derivative sign;

[0047] Set global baseline radius According to the curvature amplitude of the terrain The sampling radius is adaptively adjusted to obtain an adaptive sampling radius that takes into account the terrain curvature. Specifically, it is expressed as:

[0048]

[0049] in, This represents the terrain curvature adjustment coefficient; by designing an adaptive sampling radius that takes into account the terrain curvature, dense sampling is achieved in steep terrain areas and sparse sampling in flat areas, reflecting the actual terrain.

[0050] The adaptive sampling radius based on terrain curvature is used in the task area, i.e., the feasible target area. Initial point obtained by internal random sampling Let the sampling set Active expression Based on the dynamic minimum spacing from the active table Select a new seed point The expansion operation is performed point by point in sequence; dynamic minimum spacing. for:

[0051]

[0052] in, Indicates the repulsive force weight. Indicates the new seed point The repulsive distance to the nearest existing seed point Indicates the attractiveness weight, Indicates the new seed point The average Euclidean distance to the uncovered area;

[0053] Angle constraints are imposed on the selection of new seed points. ,in, Indicates the first The polar angle of each candidate direction. Represents the uniform distribution function. Indicates the first Half-width of the angular window for round sampling. Indicates the maximum number of sampling rounds. Representing pi, that is, the first round only occurs along the seed point. Sampling is performed within the forward small sector; if sampling fails, then... The increased exponent widens the angle window until the last round allows Sampling to obtain candidate points , Indicates the first Each candidate point is relative to the new seed point. The sampling radius; Indicates the first The polar angle of each candidate direction; Represents the sine function; Represents the cosine function. Window inspection of overly dense seed clusters; characterization of overly dense clusters, retaining only potential values. The largest seed point. Potential value. The calculation formula is expressed as follows:

[0054]

[0055] in, This represents the neighborhood set of the current seed point set. Indicates the nearest seed point, This represents the average Euclidean distance from the current seed point to the uncovered region. express The maximum value, Represents the seed point. This indicates the calculation of overflow control quantity.

[0056] Furthermore, after preprocessing the task region using Poisson disk sampling, a seed point set is obtained. Based on this, construct the unit weights. The Voronoi elements of the task region are used to obtain weighted three-dimensional Voronoi elements. .in, This represents a weighted 3D Voronoi element; Indicates the first The spatial coordinates of the seed points Represents any point in space. Indicates the first The spatial coordinates of the seed points Indicates the first The weights corresponding to each seed point Indicates the first The weights corresponding to each seed point This represents the three-dimensional task environment.

[0057] Subsequently, the target Voronoi element is segmented using Delaunay triangulation to obtain the triangular set after segmentation:

[0058] ,

[0059] in, Represents a triangular unit. Represents a set The first in One vertex, Represents a set The first in One vertex, Represents a set The first in One vertex, Represents the set of edges in a triangulation; Indicates the first A weighted 3D Voronoi unit; thus, the task region is segmented into a region consisting of weighted 3D Voronoi units and A three-dimensional environment skeleton was constructed to complete the environment preprocessing. Finally, the task area was divided into sections... and The three-dimensional environment skeleton is composed of these elements.

[0060] S3. Based on the preprocessed environment, a clustering strategy using sector migration is used to achieve balanced task allocation for the UAV swarm;

[0061] Specifically, the clustering strategy for drone swarms based on sector migration is mainly divided into two stages, which include:

[0062] In the first stage, the task area is initially divided by polar sector to achieve approximately uniform partitioning.

[0063] Specifically, using base projection As the extreme point, the task point Mapped to polar coordinates This is an automatic angular domain established based on the distribution of poles and task points. Indicates the index number of the task point; polar coordinates are:

[0064]

[0065] in, Represents the modulo function. Represents the arctangent function; angular width of each sector. , This represents the number of partitions, the first... Each sector Represented as:

[0066]

[0067] In the second stage, capacity migration operations are performed on the partitioned clusters to achieve capacity balancing.

[0068] Specifically, for overloaded or underloaded sectors, capacity balancing is achieved by migrating task points along neighbor boundaries until each cluster meets capacity constraints. (Upper capacity limit for each sector.) ,in, Indicates theoretical capacity, Indicates task points. Extreme capacity; The curvature weight is the topographic curvature. This represents the clustering float.

[0069] For overloaded sectors For overloaded sectors Perform edge migration operations and define neighbor sector indexes. and adjacent sector gap The migrated sectors are represented as follows:

[0070]

[0071] in, Indicates the sector number of the adjacent sector on the left. Indicates the sector number to the right. Indicates a gap in the adjacent sector on the left. Indicates a gap in the adjacent sector on the right. This represents the set of task points within the left adjacent sector. This represents the set of task points within the adjacent sector to the right;

[0072] Will The task point is from the overloaded sector. The sector can be migrated from the edge to the target cluster. If both neighboring clusters are full, the remaining excess task points are moved to the cluster with the smallest global capacity. The above operation is repeated to ensure the continuity of the corner domains between clusters until all sectors are clustered.

[0073] S4. Based on the balanced task allocation results, the improved multi-subgroup gray wolf optimization algorithm is used to plan the global coverage path of the UAV swarm to obtain the global path. The specific steps include:

[0074] S41. Perform population initialization;

[0075] The embodiments of the present invention employ a population initialization method that combines chaotic perturbation and random sequences to ensure uniform population distribution and global convergence sensitivity, as specifically described below:

[0076]

[0077]

[0078] in, It is a chaotic sequence. For length is A chaotic sequence; It is a random sequence. It is an independent uniform random sequence. The permutation yields the first... One index, This function represents the function that returns the sorted index. Indicates the sequence according to After rearrangement The solution corresponding to each element. This represents the solution sequence obtained after rearrangement. Represents a sequence according to The first random permutation obtained The solution corresponding to each element. This represents the sequence of solutions obtained after random permutation.

[0079] S42. Calculate the fitness function;

[0080] fitness function Global sequence of capacity-balanced clusters balancing path length and mission point coverage completeness The calculation can be expressed as:

[0081]

[0082]

[0083]

[0084] in, Represents the fitness function. Represents the global sequence of capacity-balanced clusters Weighing path length, Indicates the completeness of task point coverage. The coverage radius is determined by the overall planning. Indicates the first The number of global path segments assigned to each drone. Represents the first path in the global path sequence. One task point, Represents the first path in the global path sequence. One task point, Indicates the first The drone's path returns to the index of the starting point. Indicates the first The set of task points to be handled by the drone. Represents the Euclidean distance function. This is the penalty coefficient for not covering the area. Indicates the first A global path segment.

[0085] S43. Calculate the adaptive convergence factor;

[0086] In this embodiment, the gray wolf population convergence factor The calculation employs a combination of mechanisms including linear decreasing, convergence adjustment, and periodic perturbation, specifically as follows:

[0087] Design No. Normalized path variance of a generation of gray wolf population Specifically, it is expressed as:

[0088]

[0089] in, For the first Only the gray wolf in the first The path length of the generation, For population size, Indicates the first The mean path length of the gray wolf population.

[0090] According to the Normalized path variance of a generation of gray wolf population Calculate the dynamic minimum convergence factor ,in, As the absolute lower limit, The variance sensitivity coefficient, Represents the dynamic minimum convergence factor Maximum allowed value;

[0091] Based on the dynamic minimum convergence factor Calculate the adaptive convergence factor Specifically, it is expressed as follows:

[0092]

[0093] in, The current iteration and the total number of iterations; It is an exponential decay rate; This indicates that the dynamic lower bound is the dynamic minimum convergence factor. Preventing later Frequent jitter; linear terms Provides a trend from global to local; exponential suppression term. To avoid premature convergence; cosine perturbation term It periodically escapes local optima.

[0094] S44. Based on the fitness function and adaptive convergence factor Population renewal includes the following steps:

[0095] Design No. Subgroup of task points and its population ,in, A solution sequence for a gray wolf population. , This indicates the size of the gray wolf population.

[0096] According to fitness function right Sort the results to obtain the optimal solution for the gray wolf pack. Optimal solution and suboptimal solutions As The Gray Wolf Sequence of the Generation The leadership sequence and candidate solutions of the gray wolf population are represented as follows:

[0097]

[0098]

[0099] in, The four candidate solutions constructed include , , and , Indicates the first The target sequence in the next iteration update. Population discrete update operator; , and This represents the solution sequence of the gray wolf population.

[0100] By ensuring during the iterative update process of the population From get This is the current sequence of optimal solutions. Repeat the above process until the iteration termination condition is met, finally obtaining the global programming sequence for each subgroup.

[0101] Multi-neighborhood discrete update operator, i.e. population discrete update operator Represented as:

[0102]

[0103] in, It is a random probability distribution; To update the length of the inserted fragment, where, For leadership sequence with sequence The length of the leadership difference set, Insert the segment length into the selected leadership difference set sequence; Indicates the insertion position; For sequence Sub-field index; This function inserts a specified object into a list at a specified position by inserting the resulting fragment of the leadership difference set. Location Then, the entire leadership sequence is moved at once, achieving rapid convergence of the current sequence; By fine-tuning the sequence, path crossing within the population can be reduced.

[0104] S45. Cross-subgroup periodic sharing mechanism, specifically:

[0105] During the population discrete update process, at fixed intervals... Perform an optimal fragment sharing update across subgroups to accelerate global convergence while maintaining self-consistent convergence in each subgroup.

[0106] First, through , obtained the group First-class priority sequence Its experience fragments ,in, For the length of the empirical segment, This represents the total population size. Further, a shared deduplication operation is performed on the target sequence, retaining only the elements in the empirical fragment that belong to the target sequence, resulting in a new empirical sequence. .Will A new target sequence is obtained by randomly inserting it into the target sequence. ,use Update its subgroups to the th This provides superior gene fragments to the gray wolf population. Finally, through... For the final sequence of each subgroup Perform reversal and optimization operations to remove redundant cross edges and improve local convergence accuracy.

[0107] S5. The multi-objective cost function is used as the evaluation index for global path selection of UAV swarm, and an adaptive elastic window and potential field obstacle avoidance strategy are designed to optimize the local path of UAV swarm.

[0108] In this embodiment, a refined cost function is employed during the local planning phase. Specifically, the UAV... The weighted composite cost function is also known as the multi-objective cost function. It is expressed as follows:

[0109]

[0110]

[0111]

[0112]

[0113]

[0114]

[0115] in, Indicates drone Multi-objective cost function; These represent the costs of level flight, climb / descent, energy consumption, boundary penalty, and turning, respectively. The weights are respectively for level flight cost, climb / descent cost, energy consumption cost, boundary penalty cost, and turning cost; and These are the coordinates of the 3D environment projected onto the horizontal plane. Indicates the current altitude of the drone. Indicates the height of the target point. and Indicates the horizontal coordinates of the target point. This indicates the current horizontal distance between the drone and the boundary of the mission area. This indicates the calculation of overflow control quantity. and This indicates the current horizontal coordinates of the drone. and This represents the maximum boundary value of the horizontal plane of the task area. This represents the difference between the current heading and the target azimuth. This indicates the current heading angle of the drone. Represents the normalization function. Represents the arctangent function. and This indicates the horizontal coordinates of the turning task point.

[0116] The design of an adaptive elastic window specifically involves: designing global paths for each UAV to generate. The flexible window is defined as:

[0117]

[0118] in, and These are the starting and ending indices of the flexible window, respectively. This represents the index of the flexible window. The flexible window can adaptively update during scrolling and has a length of [value missing]. Specifically, it can be represented as follows:

[0119]

[0120] in, This indicates that the window is updating to the minimum number of overlapping elements; Indicates drone Remaining available task points; Indicates the window sparsity threshold. Indicates the first A drone at all times The amount of change in the width of the flexible window. This indicates the maximum allowable value for the width variation of the flexible window.

[0121] The specific design of the potential field obstacle avoidance strategy is as follows: In this invention, the target heading angle of the UAV is determined by the target attraction angle. Boundary repulsion vector Inter-machine repulsive vector In summary, These are the weighting coefficients for each item, and the heading angle. Represented as:

[0122]

[0123] The determination of whether there are obstacles obstructing the path of the drone when flying to the next task point is based on the heading angle is expressed as follows:

[0124]

[0125] in, At the current flight altitude, For path parameterization factors; for The terrain elevation at that location; For vertical safety margin; and drones Current location and its target location.

[0126] If obstacles obstruct the path, a side-flight candidate point is selected as the relay target, and a bypass maneuver is performed. The side-flight candidate points are represented as follows:

[0127]

[0128] in, Candidate points for side-flight; Side-flight distance; This indicates the initial side-flight offset; Indicates additional deviation; Indicates drone Current position planar projection coordinates; Indicates the target's projected coordinates; This is the lateral flight direction control vector, perpendicular to the current flight tangent; subscript This indicates the candidate point number for side-flight maneuvers.

[0129] If there are no available mission points during side-flying operations, the drone is determined to have entered a difficult maze, and the RTT* algorithm is used to extricate it from the maze.

[0130] S6. Based on differential communication, perform UAV swarm communication and smoothly output the global path of the UAV swarm after local path optimization to complete the 3D collaborative coverage planning of the UAV swarm. This includes the following steps:

[0131] S61. Design a multi-machine collaborative sharing mechanism for UAV swarm communication;

[0132] This invention employs a multi-machine collaborative sharing mechanism based on pairwise communication. When any two UAVs enter the communication radius and meet the communication interval condition, they only exchange incremental information about newly added coverage task points within their respective clusters. After multiple iterations, the system can achieve global information synchronization. The multi-machine communication formula can be expressed as:

[0133]

[0134] in, For drones and drones The distance between them; For drones and drones Communication interval; Indicates the reliable communication radius; Indicates the communication cooling period; Indicates drone and Communication flags;

[0135] Specifically, differential communication only exchanges incremental information about the other party's task point set, and the UAV after communication synchronization... The covered point set can be represented as:

[0136]

[0137] in, Indicates the drone after communication At any moment The point set has been covered. Indicates the drone before communication At any moment The point set has been covered. Indicates the relationship with the first The first drone to communicate A drone at all times The task point set has been covered. Indicates the first The set of task points corresponding to the task cluster of a drone.

[0138] S62. Smoothly output the planned path for the drone swarm;

[0139] This invention smooths the obtained path points using cubic B-spline curves, and designs discrete path points as follows: First, the arc length is parameterized, and then for each path point... Independent construction of natural splines Specifically, it can be expressed as:

[0140]

[0141] Furthermore, continuous space curves can be obtained. .in, Represents a natural spline. Represents the control point coefficient. Describes the cubic basis functions of B-Spline. Representing natural spline functions In parameter value The function value at time, Indicates the known The parameter values ​​for each data point Indicates the first The target value for a known data point.

[0142] Example 2:

[0143] Based on Example 1, in order to verify the effectiveness of the UAV swarm 3D cooperative coverage planning method based on multi-subgroup gray wolf optimization proposed in this invention, experimental verification was conducted, and the main parameter settings are shown in Table 1:

[0144] Table 1 Main Parameter Settings

[0145]

[0146] For example, this embodiment provides a task scenario, such as Figure 2As shown, the mission scenario is as follows: In a mountainous terrain containing two radar jamming sources and two threat zones, a swarm of multiple rotary-wing UAVs enters the target area. The only known information is the geographical location and boundary information of the target area, as well as the locations of several key targets within the area. The mission objective is for the UAV swarm to cover the entire area as quickly as possible with minimal path overlap.

[0147] Based on the above task scenario, the experimental results are shown below. Figure 3 , Figure 4 , Figure 5 , Figure 6 , Figure 7 and Figure 8 This verifies the effectiveness of the method of the present invention. For example... Figure 3 As shown, the dynamic Poisson disk sampling and Voronoi diagram preprocessing method used in this invention can reflect the actual environmental characteristics and achieve balanced, sparse, and feasible partitioning of the task area. Figure 4 This paper demonstrates the global path distribution of a drone swarm, showing that the global paths traverse unreachable areas while meeting load balancing requirements. This verifies the proposed balanced clustering method based on sector migration and the improved global planning method for multi-subswarm gray wolf optimization. Combined with... Figure 6 As shown, this invention employs a local planning strategy that combines global planning with a flexible window local greedy algorithm and potential field obstacle avoidance, as well as a UAV swarm communication mechanism based on differential communication, to enable each UAV to avoid threat areas in a three-dimensional mountainous environment, complete the coverage task, and return to the base along a path with less overlap. Figure 7 and Figure 8 The paper further presents a comparison of the proposed method with traditional methods in terms of fitness curves, total trajectory length curves, and coverage curves. This shows that the proposed method improves the algorithm's convergence speed and reduces its cost in multi-UAV coverage optimization, generating a stable convergence curve. It can complete the area coverage task in a short time and reduce the total flight mileage. To further quantitatively compare the planning effects of each method, relevant comparison results are shown in Table 2.

[0148] Table 2 Comparison results of each method

[0149]

[0150] The example data in Table 2 shows that the method of the present invention reduces the average mileage and running time of the UAV swarm, improves the coordination among individual UAVs, and achieves load balancing within the UAV swarm. In summary, the experimental results of this embodiment verify the effectiveness of the proposed three-dimensional cooperative coverage planning method for UAV swarms based on multi-subgroup gray wolf optimization.

[0151] Example 3:

[0152] Based on Example 1, this embodiment of the invention provides a UAV swarm 3D cooperative coverage planning system based on multi-subgroup gray wolf optimization, which can be used to implement the UAV swarm 3D cooperative coverage planning method based on multi-subgroup gray wolf optimization as described in the previous embodiment. The system includes:

[0153] The model building module is used to create digital elevation maps of three-dimensional multi-source threat environments;

[0154] The preprocessing module is used for environmental preprocessing based on digital elevation maps, employing dynamic Poisson disk sampling and Voronoi diagrams.

[0155] The task allocation module is used to perform balanced task allocation for UAV swarms based on the preprocessed environment and using a sector migration clustering strategy.

[0156] The path planning module is used to perform global coverage path planning for UAV swarms based on the balanced task allocation results and using an improved multi-subgroup gray wolf optimization algorithm to obtain the global path.

[0157] The local path optimization module is used to optimize the local path of the UAV swarm based on the adaptive elastic window local greedy method and the potential field obstacle avoidance strategy.

[0158] The smooth output module is used for UAV swarm communication based on differential communication, and to smoothly output the global path of the UAV swarm after local path optimization, thus completing the three-dimensional collaborative coverage planning of the UAV swarm.

[0159] According to embodiments of the present invention, the present invention also provides an electronic device, a readable storage medium, and a computer program product.

[0160] In an exemplary embodiment, the electronic device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization as described in Embodiment 1 above.

[0161] In an exemplary embodiment, the readable storage medium may be a non-transitory computer-readable storage medium storing computer instructions for causing the computer to execute the UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization as described in Embodiment 1 above.

[0162] In an exemplary embodiment, the computer program product includes a computer program that, when executed by a processor, implements the UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization as described in Embodiment 1 above.

[0163] The program code used to implement the methods of the present invention can be written in any combination of one or more programming languages. This program code can be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing device, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code can be executed entirely on the machine, partially on the machine, as a standalone software package partially on the machine and partially on a remote machine, or entirely on a remote machine or server.

[0164] In the context of this invention, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can include, but are not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0165] To provide interaction with a user, the systems and techniques described herein can be implemented on a computer having: a display device for displaying information to the user (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor); and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the computer. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0166] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as a data server), or computing systems that include middleware components (e.g., an application server), or computing systems that include frontend components (e.g., a user computer with a graphical user interface or web browser through which a user can interact with embodiments of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., a communication network). Examples of communication networks include local area networks (LANs), wide area networks (WANs), and the Internet.

[0167] Computer systems can include clients and servers. Clients and servers are generally located far apart and typically interact via communication networks. Client-server relationships are created by computer programs running on the respective computers and having a client-server relationship with each other. Servers can be cloud servers, servers in distributed systems, or servers incorporating blockchain technology.

[0168] Those skilled in the art will recognize that the embodiments described herein are intended to help the reader understand the principles of the invention, and should be understood that the scope of protection of the invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific modifications and combinations based on the technical teachings disclosed in this invention without departing from the spirit of the invention, and these modifications and combinations are still within the scope of protection of this invention.

Claims

1. A method for three-dimensional cooperative coverage planning of UAV swarms based on multi-subgroup gray wolf optimization, characterized in that, Includes the following steps: Establish a three-dimensional digital elevation map of the multi-source threat environment; Based on digital elevation maps, dynamic Poisson disk sampling and Voronoi diagrams are used for environmental preprocessing. Based on the preprocessed environment, a clustering strategy using sector migration is used to achieve balanced task allocation for the UAV swarm. Based on the balanced task allocation results, the improved multi-subgroup gray wolf optimization algorithm is used to plan the global coverage path of the UAV swarm and obtain the global path. The multi-objective cost function is used as the evaluation index for global path selection of UAV swarm, and an adaptive elastic window and potential field obstacle avoidance strategy are designed to optimize the local path of UAV swarm. Based on differential communication, UAV swarm communication is carried out, and the global path of the UAV swarm after local path optimization is smoothly output to complete the three-dimensional collaborative coverage planning of the UAV swarm. Based on the balanced task allocation results, an improved multi-subgroup gray wolf optimization algorithm is used for global coverage path planning of UAV swarms, specifically including the following steps: Population initialization is performed using a fusion method of chaotic perturbation and random sequence. The fitness function is calculated using the following formula: in, Represents the fitness function. Represents the global sequence of capacity-balanced clusters Weighing path length, Indicates the completeness of task point coverage. The coverage radius is determined by the overall planning. Indicates the first The number of global path segments assigned to each drone. Represents the first path in the global path sequence. One task point, Represents the first path in the global path sequence. One task point, Indicates the first The drone's path returns to the index of the starting point. Indicates the first The set of task points to be handled by the drone. Represents the Euclidean distance function. This is the penalty coefficient for not covering the area. Indicates the first A global path segment; The adaptive convergence factor is calculated using the following formula: in, Indicates the adaptive convergence factor; Indicates the current iteration round; Indicates the total number of iterations; It is an exponential decay rate; This represents the maximum value of the adaptive convergence factor; Indicated to prevent later The dynamic minimum convergence factor with frequent jitter; linear term Provides a trend from global to local; exponential suppression term. To avoid premature convergence, The natural constant; cosine perturbation term Periodically escapes local optima. Pi; Represents the cosine function; Population updates are performed based on the fitness function and adaptive convergence factor, specifically including the following steps: Design No. Each subgroup of task points and its corresponding population ,in, A solution sequence for the gray wolf population. , Indicates the size of the gray wolf population; According to the fitness function Sort the results to obtain the optimal solution for the gray wolf pack. Optimal Solution and suboptimal solutions As the first Leadership sequence in the next iteration update The leadership sequence and candidate solutions for the gray wolf population are: in, The four candidate solutions constructed include , , and , Indicates the first The target sequence in the next iteration update. Population discrete update operator; , and The solution sequence represents the gray wolf population; By ensuring during the iterative update process of the population ,from get This is the sequence of optimal solutions in the current iteration. Indicates the first The optimal solution sequence obtained from the next iteration update The corresponding fitness function, Indicates the first The optimal solution sequence obtained from the next iteration update The corresponding fitness function is used; the iteration is repeated until the iteration termination condition is met, and the global planning sequence of each subgroup is obtained. By performing inversion and optimization operations on the global planning sequence of each subgroup using 2-Opt, redundant cross edges are removed, and the global coverage path planning of the UAV swarm is completed.

2. The UAV swarm 3D cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 1, characterized in that, The formula for representing the digital elevation map of the three-dimensional multi-source threat environment is as follows: in, Digital elevation map model representing a three-dimensional multi-source threat environment; and These are the coordinates of the 3D environment projected onto the horizontal plane. Represents coordinate points in a mountainous environment The corresponding elevation value; Indicates the effective range of the threat zone; Indicates the effective range of the radar jamming zone; Indicates the maximum value; Indicates the first One threat source index; Indicates the first Index of radar interference sources; Indicates the first Projected coordinates of the center point of each mountain peak; Terrain parameters for controlling altitude; No. The mountain peaks along The amount of attenuation of the shaft; Indicates the first The mountain peaks along The amount of attenuation of the shaft; This represents the total number of mountain peaks. Represents an exponential function; The vertical upper limit of the interception envelope or no-fly zone; To threaten regional security margins; Indicates the projection of the threat center on Coordinates on the plane; Indicates the radius of the threat zone; This indicates the distance between a point within the radar interference range and the radar interference source. Indicates the projection of the radar interference source on Coordinates on the plane; This indicates the effective radius of radar jamming.

3. The UAV swarm 3D cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 1, characterized in that, Environmental preprocessing is performed using dynamic Poisson disk sampling and Voronoi diagrams, specifically including the following steps: Based on a digital elevation map of a three-dimensional multi-source threat environment, the task area is sampled by dynamic Poisson disk sampling to generate a seed point set with uniform and adaptive density. Based on the seed point set, Voronoi elements and element weights are constructed for the task region, resulting in weighted 3D Voronoi elements: in, This represents a weighted 3D Voronoi element; Indicates the first The spatial coordinates of the seed points Represents any point in space. Indicates the first The spatial coordinates of the seed points Indicates the first The weights corresponding to each seed point Indicates the first The weights corresponding to each seed point Represents a three-dimensional task environment; The three-dimensional Voronoi elements are segmented using Delaunay triangulation to obtain the triangular sets after segmentation. : ， in, Represents a triangular unit. Represents a set The first in One vertex, Represents a set The first in One vertex, Represents a set The first in One vertex, Represents the set of edges in a triangulation; Indicates the first A weighted 3D Voronoi unit; thus, the task region is segmented into a region consisting of weighted 3D Voronoi units and A three-dimensional environmental skeleton is formed to complete environmental preprocessing.

4. The UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 3, characterized in that, A digital elevation map based on a 3D multi-source threat environment is generated by sampling the task area using dynamic Poisson disk sampling to produce a seed point set with uniform and adaptive density. The specific steps include: Calculate the terrain curvature amplitude based on a three-dimensional digital elevation map of a multi-source threat environment. : in, Indicates the sign of the partial derivative; and These are the coordinates of the 3D environment projected onto the horizontal plane. Digital elevation map model representing a three-dimensional multi-source threat environment; Set global baseline radius According to the curvature amplitude of the terrain The sampling radius is adaptively adjusted to obtain an adaptive sampling radius that takes into account the terrain curvature. : in, Indicates the terrain curvature adjustment coefficient; An initial set of points is obtained by randomly sampling within the task area based on an adaptive sampling radius according to the terrain curvature. Let the sampling set Active expression Based on the dynamic minimum spacing from the active table Select a new seed point The expansion operation is performed point by point in sequence; dynamic minimum spacing. for: in, Indicates the repulsive force weight. Indicates the new seed point The repulsive distance to the nearest existing seed point Indicates the attractiveness weight. Indicates the new seed point The average Euclidean distance to the uncovered area; An angle constraint is imposed on the selection of new seed points, and the angle window is expanded exponentially with each sampling round until the last round allows 180° sampling, thus obtaining candidate seed points. , Indicates the first Each candidate point is relative to the new seed point. The sampling radius; Indicates the first The polar angle of each candidate direction; Represents the sine function; Represents the cosine function; by The window checks for overly dense seed point clusters among the candidate seed points, performs feature processing on the overly dense seed point clusters, retains the candidate seed point with the largest potential value, and constructs a seed point set.

5. The UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 4, characterized in that, Potential The calculation formula is expressed as follows: in, This represents the neighborhood set of the current seed point set. Indicates the nearest seed point, This represents the average Euclidean distance from the current seed point to the uncovered region. express The maximum value, Represents the seed point. This indicates the calculation of overflow control quantity.

6. The UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 1, characterized in that, Based on the preprocessed environment, a clustering strategy using sector migration is employed to achieve balanced task allocation for the UAV swarm, specifically including the following steps: Based on the preprocessed environment, the task region is initially divided into multiple sectors using polar angle sectors, achieving an approximately uniform partitioning of the task region; Each sector Represented as: in, Indicates the index number of the task point; Indicates the angle width of each sector. , This represents the automatic angular domain representation established based on the distribution of poles and task points. This parameter represents the number of partitions. Indicates the task point The polar coordinates of the task point obtained by mapping to the polar coordinate system. , Represents the modulo function; Represents the arctangent function; Indicates the extreme point; Based on the approximately uniform partitioning of the task region, a capacity migration operation is performed on the partitioned seed point clusters to achieve capacity balancing. Specifically: Overloaded sectors To perform edge migration operations, the sectors that can be migrated are: in, Indicates the sector number of the adjacent sector on the left. Indicates the sector number to the right. Indicates a gap in the adjacent sector on the left. Indicates a gap in the adjacent sector on the right. This represents the set of task points within the left adjacent sector. This represents the set of task points within the adjacent sector to the right; Represents the modulo function; This indicates the maximum capacity of each sector. ,in, Indicates theoretical capacity, Indicates task points. Extreme capacity; The curvature weight is the topographic curvature. This represents the clustering float. Will The task point is from the overloaded sector. The edge is migrated to a migrateable sector. If all neighboring sectors are full, the remaining excess task points are migrated to the cluster with the smallest global capacity. ; Repeatedly apply overloaded sectors Edge migration operations are performed to ensure inter-cluster corner domain continuity until all sectors are clustered, thus achieving balanced task allocation for the UAV swarm.

7. The UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 1, characterized in that, During the iterative update process of the population, an optimal fragment sharing update is performed across subgroups at fixed intervals, specifically: Extract the first Subgroup No. The empirical fragment of the first-best sequence updated in the next iteration; Perform a shared deduplication operation on the target sequence, retaining the elements in the experience fragment that belong to the target sequence, to obtain a new experience sequence; The new experience sequence is randomly inserted into the target sequence to obtain a new target sequence. The new target sequence is then used to process the first... The subgroups are updated for the first subgroup. The next iteration provides superior gene fragments for the gray wolf population.

8. The UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 1, characterized in that, The multi-objective cost function is specifically as follows: in, Indicates drone Multi-objective cost function; These represent the costs of level flight, climb / descent, energy consumption, boundary penalty, and turning, respectively. The weights are respectively for level flight cost, climb / descent cost, energy consumption cost, boundary penalty cost, and turning cost; and These are the coordinates of the 3D environment projected onto the horizontal plane. Indicates the current altitude of the drone. Indicates the height of the target point. and Indicates the horizontal coordinates of the target point. This indicates the current horizontal distance between the drone and the boundary of the mission area. This indicates the calculation of overflow control quantity. and This indicates the current horizontal coordinates of the drone. and This represents the maximum boundary value of the horizontal plane of the task area. This represents the difference between the current heading and the target azimuth. This indicates the current heading angle of the drone. Represents the normalization function. Represents the arctangent function. and Indicates the horizontal coordinates of the turning point; The adaptive elastic window is used to limit the UAV's selectable heading search range and search resolution in local planning. The adaptive elastic window is adaptively updated during the sliding process, specifically as follows: in, This indicates that the window is updating to the minimum number of overlapping elements; Indicates drone Remaining available task points; Indicates the window sparsity threshold. Indicates the first A drone at all times The amount of change in the width of the flexible window. This indicates the maximum allowable value for the width variation of the flexible window; The potential field obstacle avoidance strategy is as follows: calculate the heading angle based on the target attraction angle, boundary repulsion vector, and inter-machine repulsion vector; determine whether there are obstacles blocking the path of the UAV when flying to the next task point based on the heading angle; if there are no obstacles blocking the path, perform the flight operation according to the original path; if there are obstacles blocking the path, select a candidate point for side flight as a relay target and perform a bypass operation; if there are no selectable task points during the side flight operation, determine that the UAV has entered a difficult maze and use the RTT* algorithm to escape the maze. The formula for describing the candidate points for side-flight is: in, Candidate points for side-flight; Side-flight distance; Indicates the initial side flight offset; Indicates additional deviation; Indicates drone Current position planar projection coordinates; Indicates the target's projected coordinates; This is the lateral flight direction control vector, perpendicular to the current flight tangent; subscript This indicates the candidate point number for side-flight maneuvers.

9. The UAV swarm three-dimensional cooperative coverage planning method based on multi-subgroup gray wolf optimization according to claim 1, characterized in that, The specific method for UAV swarm communication based on differential communication is as follows: When any two drones enter the communication radius and meet the communication interval condition, they only exchange incremental information about newly added coverage task points within their respective clusters. Global information synchronization is achieved through multiple iterations. The multi-drone communication formula is expressed as: in, For drones and drones The distance between them; For drones and drones Communication interval; Indicates the reliable communication radius; Indicates the communication cooling period; Indicates drone and Communication flags; drones after communication synchronization The covered point set is represented as: in, Indicates the drone after communication At any moment The point set has been covered. Indicates the drone before communication At any moment The point set has been covered. Indicates the relationship with the first The first drone to communicate A drone at all times The task point set has been covered. Indicates the first The set of task points corresponding to the task cluster of a drone; The global path of the UAV swarm, after local path optimization, is smoothly output using cubic B-spline curves, specifically as follows: Design a set of discrete path points and parameterize the arc length of each path point; Based on the parameterized path point arc length, natural splines are independently constructed for each path point, thereby obtaining a continuous spatial curve and completing the path smoothing output; The formula for constructing natural splines is: in, Represents a natural spline. Represents the control point coefficient. Describes the cubic basis functions of B-Spline. Representing natural spline functions In parameter value The function value at time, Indicates the known The parameter values ​​for each data point Indicates the first The target value for a known data point.

Images