A distribution network unmanned aerial vehicle autonomous patrol track planning method and system

Abstract

The application discloses a kind of distribution network unmanned plane autonomous patrol track planning method and system, wherein the method comprises: establishing plane coordinate map;Connecting starting point and target point obtains direct track, determines ideal track based on the intersection position of direct track and obstacle, divides plane coordinate map and determines sampling point distribution area;Randomly generate sampling point based on sampling point distribution area, establish gravitational field with starting point and ideal track as center, establish repulsive force field with obstacle as center to guide the search direction of RRT algorithm, establish random tree from target point Reverse planning unmanned plane patrol optimal track.This method introduces the gravitational effect of starting point and ideal track, the repulsive force effect of obstacle is the direction of RRT algorithm guidance, to solve the meaningless track search caused by pure random of RRT algorithm, the problem that artificial potential field exists and produces resultant force 0 and when obstacle is very close to target point excessive obstacle avoidance, realizes the feasibility and high efficiency of unmanned plane patrol optimal track planning.

Description

Technical Field

[0001] This invention relates to the field of flight path planning, specifically to a method and system for autonomous inspection flight path planning of power distribution network unmanned aerial vehicles (UAVs). Background Technology

[0002] With the rapid development of drone technology, more and more patrol drones are being used. Patrol drones have advantages such as large patrol range, short patrol time, high efficiency, high patrol safety, and high flexibility. However, when operating in the field, drones inevitably encounter environmental obstacles such as trees, utility poles, flocks of birds, houses, and hillsides, which may prevent the drones from reaching the designated locations and thus fail to achieve their patrol objectives.

[0003] Therefore, it is of great significance to study how to design a dynamic autonomous patrol trajectory planning method for UAVs to achieve efficient operation of patrol UAVs in the field while ensuring the safety of UAV operation.

[0004] Because the trajectory planning range of patrol drones is large and the environmental complexity varies greatly across different patrol missions, trajectory planning algorithms are required to possess environmental adaptability, very fast convergence speed, and computational accuracy. On the one hand, most existing trajectory planning algorithms are suitable for solving static trajectory planning problems. On the other hand, traditional trajectory planning algorithms suffer from various problems. For example, the lack of directionality in the RRT algorithm leads to many invalid searches and slow planning speed; the artificial potential field method suffers from problems such as zero net force resulting in unchanged heading and excessive obstacle avoidance when obstacles are very close to the target point; and intelligent optimization algorithms such as genetic algorithms, particle swarm optimization, and artificial bee colony optimization suffer from problems such as being prone to getting trapped in local optima and low search efficiency. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention provides a method and system for autonomous inspection trajectory planning of power distribution network UAVs. The purpose is to solve the problems in the prior art of UAV trajectory planning, such as the lack of directionality of the RRT algorithm, the generation of zero resultant force by the artificial potential field method, and excessive obstacle avoidance when the obstacle is very close to the target point.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for autonomous patrol trajectory planning of power distribution network unmanned aerial vehicles (UAVs), comprising:

[0008] A planar coordinate map is established based on the starting point, target point, and obstacle positions of the UAV trajectory planning, wherein the straight line connecting the starting point and the target point is the x-axis, and the straight line perpendicular to the x-axis is the y-axis.

[0009] A direct track is obtained by connecting the starting point and the target point. An ideal track is determined based on the intersection of the direct track and the obstacle. The planar coordinate map is segmented based on the ideal track, and the sampling point distribution area is determined.

[0010] Based on the sampling point distribution area, sampling points are randomly generated. An artificial potential field is constructed by establishing an gravitational field centered on the starting point and the ideal trajectory, and a repulsive field centered on the obstacle, to guide the search direction of the RRT algorithm. A random tree is built in reverse from the target point to plan the optimal trajectory for UAV patrol.

[0011] Preferably, determining the ideal trajectory based on the intersection position of the direct trajectory and the obstacle includes:

[0012] The maximum and minimum points of the y-axis coordinates of the obstacles intersecting the direct flight path are recorded as the obstacle tangent points;

[0013] Select one of the two obstacle tangents on each intersecting obstacle as the ideal track point such that the connecting line obtained by sequentially connecting the target point, each ideal track point, and the starting point from the target point in the direction of the starting point is the shortest, and the shortest connecting line is the ideal track.

[0014] Preferably, the step of segmenting the planar coordinate map and determining the sampling point distribution area based on the ideal flight path includes:

[0015] The planar coordinate map is divided into multiple segments using the selected obstacle tangent point as the segment endpoint;

[0016] Based on the segmentation, the sampling point distribution area of ​​the UAV in each segment of the planar coordinate map is determined sequentially from the target point towards the starting point:

[0017] Within each segment of the planar coordinate map, if the ideal trajectory of this segment passes through the x-axis or the two segment endpoints are the maximum and minimum points of the y-axis coordinates of the obstacles, then the areas within the upper and lower boundaries of the two obstacles closest to the ideal trajectory in the upper and lower regions of the planar coordinate map of this segment are selected as the sampling point distribution area in the planar coordinate map of this segment.

[0018] Within each segment of the planar coordinate map, if both endpoints of the segment are the minimum points of the y-axis coordinates of the obstacle, then the area below the ideal flight path within the planar coordinate map of this segment is selected as the sampling point distribution area within the planar coordinate map of this segment.

[0019] Within each segment of the planar coordinate map, if both endpoints of this segment are the maximum points of the y-axis coordinates of the obstacle, then the area above the ideal flight path within this segment of the planar coordinate map is selected as the sampling point distribution area within this segment of the planar coordinate map.

[0020] Preferably, establishing a gravitational field centered on the ideal trajectory includes:

[0021] The ideal trajectory exerts an attractive force on the optimal trajectory for UAV inspection, denoted as the ideal trajectory gravity. The ideal trajectory gravity is always perpendicular to the ideal trajectory and points towards the ideal trajectory, causing the optimal trajectory for UAV inspection to tend towards the ideal trajectory.

[0022] The ideal trajectory gravitational field function of the ideal trajectory gravity is:

[0023]

[0024] in, Let K be the gravitational field function of the ideal trajectory. a Let d(x,l) be the gravitational constant of the ideal trajectory, and let d(x,l) represent the distance between the current position x of the UAV and the position l of the ideal trajectory.

[0025] Preferably, the construction of the artificial potential field by establishing the gravitational field and repulsive field includes:

[0026] The net force F acting on the drone in the artificial potential field M The target gravity generated by the target gravitational field centered at the starting point The ideal trajectory gravity generated by the ideal gravitational field centered on the ideal trajectory The repulsive force F generated by the repulsive field centered on the obstacle rep sum;

[0027] The node step size of the RRT algorithm under the artificial potential field depends on randomness and the guiding effect of the artificial potential field. The node generation formula is as follows:

[0028]

[0029] Where, x new Let F be a new node generated by the random tree of the RRT algorithm under the artificial potential field, where C is a fixed step size based on randomness, and F is a new node generated by the random tree. M Let be the resultant force exerted on the UAV in the artificial potential field, || denote the magnitude of the vector, α be the weight of the random factor, β be the weight of the artificial potential field factor, and α < β.

[0030] Preferably, the fixed step size of the node generation step size of the RRT algorithm under the artificial potential field, based on randomness, is adapted to the complexity of the environment in which the UAV is located; specifically:

[0031] The complexity of the environment in which the UAV is located is calculated based on the total area and dispersion of obstacles in the aforementioned planar coordinate map;

[0032] A fixed step size based on randomness is determined based on the complexity of the environment in which the UAV is located and the distance between the starting point and the target point.

[0033] Preferably, after building the random tree in reverse from the target point, the method further includes:

[0034] Two nodes separated by one node on the current optimal trajectory are denoted as an optimized node group, wherein the node closer to the starting point in the optimized node group is denoted as the starting point, and the node closer to the target point in the optimized node group is denoted as the ending point;

[0035] From the starting point to the target point, sequentially check whether the direct connection between two nodes in the optimized node group intersects with an obstacle and whether the minor angle of the intersection of the connecting line with the previous flight path is greater than the maximum turning angle of the UAV;

[0036] If the nodes do not intersect and the minor angle of the intersection of the connecting line and the previous track is greater than the maximum turning angle of the UAV, then the endpoint of the optimized node group in this group is deleted, and the starting point in the optimized node group is directly connected to the endpoint of the optimized node group in the previous group, until all nodes on the current optimal track are checked to obtain the optimized track.

[0037] The optimal UAV patrol trajectory is obtained by smoothing the optimized trajectory using a cubic B-spline interpolation algorithm.

[0038] Secondly, this invention provides an autonomous patrol trajectory planning system for power distribution network unmanned aerial vehicles (UAVs), comprising:

[0039] The planar coordinate map creation module is used to create a planar coordinate map representing the starting point, target point, and obstacle positions of the UAV trajectory planning, with the straight line connecting the starting point and the target point as the x-axis and the straight line perpendicular to the x-axis as the y-axis.

[0040] The sampling point layout module is used to connect the starting point and the target point to obtain a direct track, determine an ideal track based on the intersection position of the direct track and the obstacle, and segment the planar coordinate map based on the ideal track and determine the sampling point layout.

[0041] The artificial potential field construction module is used to construct an artificial potential field by establishing a gravitational field centered on the starting point and the ideal trajectory, and by establishing a repulsive field centered on the obstacle.

[0042] The optimal trajectory planning module is used to plan the optimal trajectory for UAV patrol based on the RRT algorithm and guided by an artificial potential field. It builds a random tree from the target point in reverse to plan the optimal trajectory for the UAV patrol.

[0043] Thirdly, the present invention provides an electronic device, comprising:

[0044] Memory, used to store executable instructions;

[0045] The processor, when running the executable instructions stored in the memory, implements the autonomous patrol trajectory planning method for power distribution network UAVs as described above.

[0046] Fourthly, the present invention provides a computer-readable storage medium storing executable instructions, characterized in that, when the executable instructions are executed by a processor, they implement a method for autonomous inspection trajectory planning of a power distribution network unmanned aerial vehicle as described above.

[0047] The present invention provides a method and system for autonomous inspection trajectory planning of power distribution network unmanned aerial vehicles (UAVs), which has the following beneficial effects:

[0048] 1. This application addresses the undirected nature of the RRT algorithm by employing an artificial potential field to guide the trajectory search direction and introducing an ideal trajectory gravity. The ideal trajectory, as the shortest trajectory that avoids obstacles, serves as a guide, quickly converging the trajectory to be planned to the vicinity of the ideal trajectory, thereby improving the efficiency of trajectory search and avoiding some meaningless trajectory search processes.

[0049] 2. This application takes into account that there are a large number of meaningless sampling points in the early stage of the existing RRT algorithm. Based on the mathematical graph approach, the ideal trajectory is determined, and the location of sampling points is filtered to reduce the number of meaningless sampling points generated and improve the trajectory planning efficiency.

[0050] 3. Based on the large environmental changes under different UAV inspection missions, this application improves the fixed step size in the existing technology by introducing the randomness control of the RRT algorithm and the potential force guidance of the artificial potential field. It takes into account the area and degree of aggregation of obstacles in the environment where the inspection UAV is located, and is better suited to the needs of different inspection missions.

[0051] 4. By performing a reverse search from the target point to the starting point, the previously searched track information can still be used as a reference when the environment changes dynamically, reducing the amount of re-search required, thereby quickly finding the optimal drone patrol track to reach the target point. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating the autonomous inspection trajectory planning method for power distribution network drones according to the present invention.

[0053] Figure 2 This is a schematic diagram illustrating the principle of the present invention of segmenting a planar coordinate map and determining the sampling point distribution area based on an ideal flight path;

[0054] Figure 3 This is a structural block diagram of an autonomous inspection trajectory planning system for power distribution network drones according to the present invention. Detailed Implementation

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] To facilitate understanding of this embodiment, a detailed description of the autonomous inspection trajectory planning method for power distribution network unmanned aerial vehicles (UAVs) disclosed in this embodiment of the invention will be provided first.

[0057] This invention provides a method for autonomous inspection trajectory planning by a power distribution network unmanned aerial vehicle (UAV), such as... Figure 1 As shown, it includes:

[0058] A planar coordinate map is established based on the starting point, target point, and obstacle positions of the UAV trajectory planning, wherein the straight line connecting the starting point and the target point is the x-axis, and the straight line perpendicular to the x-axis is the y-axis.

[0059] A direct track is obtained by connecting the starting point and the target point. An ideal track is determined based on the intersection of the direct track and the obstacle. The planar coordinate map is segmented based on the ideal track, and the sampling point distribution area is determined.

[0060] Based on the sampling point distribution area, sampling points are randomly generated. An artificial potential field is constructed by establishing an gravitational field centered on the starting point and the ideal trajectory, and a repulsive field centered on the obstacle, to guide the search direction of the RRT algorithm. A random tree is built in reverse from the target point to plan the optimal trajectory for UAV patrol.

[0061] This application abstracts the environment of the UAV into a coordinate-based planar map by establishing a planar coordinate system. It determines the ideal trajectory based on the intersection of the direct path and obstacles, filtering where sampling points are generated to reduce the number of meaningless sampling points. Furthermore, it establishes an artificial potential field to address the lack of guidance in the RRT algorithm, introducing an ideal trajectory into the artificial potential field to guide the planned trajectory towards the optimal trajectory, enabling the RRT algorithm to achieve a better trajectory effect from the outset. By performing a reverse search from the target point to the starting point, the previously searched trajectory information can still be used as a reference when the environment changes dynamically, reducing the amount of re-searching required and thus quickly finding the optimal UAV patrol trajectory to reach the target point.

[0062] This application leverages the complementary advantages of artificial potential fields and the RRT algorithm. The attractive and repulsive forces of the artificial potential field provide the RRT algorithm with trajectory directionality, while the stochastic growth of the RRT algorithm effectively avoids the problems of not changing the course when the force of the artificial potential field is zero and excessive obstacle avoidance. Furthermore, some existing improvements to the artificial potential field method decompose obstacle repulsive forces and reconstruct the resultant force based on velocity and direction through classical mechanics analysis. However, this method involves a large computational load, requiring the calculation of the UAV's velocity and resultant force at every moment, which is unsuitable for real-time trajectory planning. Therefore, this application uses the RRT algorithm instead of classical mechanics analysis, reducing the need for manual analysis.

[0063] This application considers the trajectory planning problem of patrol drones in both static and dynamic environments. The challenge in dynamic environments lies in the fact that obstacles may move onto the previously planned path. Furthermore, since this application focuses on drone patrols, special circumstances may arise where the drone needs to change its original flight path or new obstacles may appear on the current optimal trajectory. The search mode employed in this application is a reverse search from the target point to the starting point. When obstacles change dynamically, the previously searched path information can still be used as a reference, reducing the amount of re-searching required.

[0064] In a forward search, "optimal" means that the path from the starting point to the target point is optimal, but the path from the current point of the UAV to the target point may not be optimal. If a reverse search is used, the path from the target point to the current point of the UAV is optimal and can be used as a reference for a new path.

[0065] Specifically, a local replanning strategy is adopted, replanning only the planned but unexecuted trajectories within a certain range of the current location, thereby reducing the computational load. First, continuous trajectories to be executed within a certain radius of the current location are searched. Nodes whose minimum distance to a new obstacle is within the UAV obstacle warning range are considered dangerous nodes. The method of this application is used to replan the trajectories by taking the two ends of the dangerous nodes as the starting and target points.

[0066] refer to Figure 2 In the diagram, polygons represent obstacles, point S is the starting point, and point G is the target point. Determining the ideal trajectory based on the intersection of the direct trajectory and the obstacle includes:

[0067] The maximum and minimum points of the y-axis coordinates of obstacles intersecting the direct flight path are recorded as the obstacle tangency points; for example... Figure 2 As shown, the maximum point of the y-axis coordinate of obstacle A is y1, and the minimum point is y2; the maximum point of the y-axis coordinate of obstacle A is y3, and the minimum point is y4.

[0068] Among the two tangent points of each intersecting obstacle, one is selected as the ideal waypoint, such that the shortest connecting line obtained by sequentially connecting the target point, each ideal waypoint, and the starting point from the target point towards the starting point is the ideal waypoint. Figure 2 The middle G-y3-y2-S segment.

[0069] It should be understood that the direct path is the shortest path from the target point to the starting point. However, since the direct path passes through obstacles, it is necessary to find a shortest path that does not cross obstacles as the ideal path to guide the planned path. This is similar to finding the tangent point of an obstacle at a given point, as described in [reference needed]. Figure 2 Specifically as follows:

[0070] The step of segmenting the planar coordinate map and determining the sampling point distribution area based on the ideal flight path includes:

[0071] The planar coordinate map is divided into multiple segments using the selected obstacle tangent points as segment endpoints; for example... Figure 2 As shown in the figure, the dashed lines parallel to the y-axis represent segments.

[0072] Based on the segmentation, the sampling point distribution area of ​​the UAV in each segment of the planar coordinate map is determined sequentially from the target point towards the starting point:

[0073] Within each segment of the planar coordinate map, if the ideal trajectory of this segment passes through the maximum and minimum points of the y-axis coordinates of obstacles, respectively, the areas within the upper and lower boundaries of the two obstacles closest to the ideal trajectory within the upper and lower regions of the planar coordinate map of this segment are selected as the sampling point distribution area within the planar coordinate map of this segment; Reference Figure 2 The middle section is y2y3.

[0074] Within each segment of the planar coordinate map, if both endpoints of this segment are the minimum points of the y-axis coordinates of obstacles, then the area below the ideal flight path within this segment of the planar coordinate map is selected as the sampling point distribution area within this segment of the planar coordinate map; (Reference) Figure 2 Middle Sy2 section.

[0075] Within each segment of the planar coordinate map, if both endpoints of this segment are the maximum points of the y-axis coordinates of obstacles, then the area above the ideal flight path within this segment of the planar coordinate map is selected as the sampling point distribution area within this segment of the planar coordinate map; (Reference) Figure 2 Gy3 segment.

[0076] It should be noted that when one of the segment endpoints is the starting point or the target point, it is only necessary to determine the location of the other endpoint of the plane coordinate map according to the above steps to determine the sampling point layout area within the plane coordinate map of that segment. For example, if one segment endpoint is the target point and the other endpoint is the maximum point of the y-axis coordinate of the obstacle, then the sampling point layout area within the plane coordinate map of that segment is the area above the ideal flight path.

[0077] The establishment of a gravitational field centered on the ideal trajectory includes:

[0078] The ideal trajectory exerts an attractive force on the optimal trajectory for UAV inspection, denoted as the ideal trajectory gravity. The ideal trajectory gravity is always perpendicular to the ideal trajectory and points towards the ideal trajectory, causing the optimal trajectory for UAV inspection to tend towards the ideal trajectory.

[0079] The ideal trajectory gravitational field function of the ideal trajectory gravity is:

[0080]

[0081] in, Let K be the gravitational field function of the ideal trajectory. a Let d(x,l) be the gravitational constant of the ideal trajectory, and let d(x,l) represent the distance between the current position x of the UAV and the position l of the ideal trajectory.

[0082] The gravity of the ideal trajectory is measured by the distance between the UAV's location and the ideal trajectory; the closer the distance, the stronger the gravity of the ideal trajectory.

[0083] The ideal trajectory, as the shortest trajectory to avoid all obstacles, is subject to an attractive force (perpendicular to the ideal trajectory and directed towards it) that pulls the planned trajectory toward the ideal trajectory, making it as close as possible to the ideal trajectory without deviating too much, thus achieving the shortest possible trajectory.

[0084] The target gravitational field function of the target gravitational field centered at the starting point is:

[0085]

[0086] in, Let K be the gravitational field function of the target gravitational field centered at the starting point. b Let d(x,x) be the target gravitational constant of the target gravitational field. g d′ represents the distance between the current position x of the UAV and the ideal flight path l. g Thresholds are defined for different ranges of influence of the target gravitational field.

[0087] It should be noted that the gravitational force exerted by the UAV on its ideal trajectory is less than the gravitational force generated by the target's gravitational field centered at the starting point, therefore Ka Less than K b .

[0088]

[0089] U rep Let K be the repulsive field function of the repulsive field centered on the obstacle. r Let d be the repulsive constant. min Let d'0 be the vector pointing from the current position of the drone to the nearest point of the obstacle, and let d'0 be the radius of the repulsive field.

[0090] Based on the above-mentioned ideal trajectory gravitational field function, target gravitational field function, and repulsive field function, the forces acting on the UAV under various potential fields are calculated. The existing artificial potential field method can be used here, and this application does not limit it.

[0091] Preferably, the construction of the artificial potential field by establishing the gravitational field and repulsive field includes:

[0092] The net force F acting on the drone in the artificial potential field M The target gravity generated by the target gravitational field centered at the starting point The ideal trajectory gravity generated by the ideal gravitational field centered on the ideal trajectory The repulsive force F generated by the repulsive field centered on the obstacle rep sum;

[0093] The node step size of the RRT algorithm under the artificial potential field depends on randomness and the guiding effect of the artificial potential field. The node generation formula is as follows:

[0094]

[0095] Where, x new Let F be a new node generated by the random tree of the RRT algorithm under the artificial potential field, where C is a fixed step size based on randomness, and F is a new node generated by the random tree. M Let be the resultant force exerted on the UAV in the artificial potential field, || denote the magnitude of the vector, α be the weight of the random factor, β be the weight of the artificial potential field factor, and α > β.

[0096] Unlike existing RRT algorithms that use a constant step size for random tree node generation, this application considers an RRT algorithm under the aforementioned artificial potential field. The generation of new nodes incorporates both the random effect of the conventional RRT algorithm and the guiding effect of the artificial potential field. Therefore, the step size of the new node is not fixed and depends primarily on the potential field. When the node is close to the starting point / ideal trajectory, the potential field is mainly generated by the target gravity / ideal trajectory gravity, causing the node to tend to be generated at the starting point / ideal trajectory. At this point, the gravity is strong, and the net force is large, resulting in a larger growth step size towards the starting point / ideal trajectory, which is beneficial for the random tree to quickly approach the starting point / ideal trajectory. When the node is close to an obstacle, the potential field is mainly generated by the repulsive force of the obstacle, causing the node to move away from the obstacle. At this point, the repulsive force is strong, and the net force is small, resulting in a smaller growth step size towards the obstacle region, which is beneficial for the random tree to approach but effectively avoid obstacles.

[0097] In this application, the sampling points are filtered by sampling point distribution area, which reduces a large number of sampling points. This reduces the randomness of RRT search in the entire map space. Therefore, the random tree needs to be able to reach any position in the sampling point distribution area as much as possible, and should not miss effective tracks due to excessive intervention of artificial potential field. Therefore, the weight of random factor should be set to be greater than the weight of artificial potential field factor, that is, α>β.

[0098] Conventional RRT algorithms use a constant step size for random tree growth, resulting in poor adaptability between complex and simple maps. In simple maps with few obstacles, the undirected nature of the algorithm can lead to reverse search. Although this application adds an artificial potential field to provide direction, the limited number of obstacles and their weak repulsive force make it difficult to quickly plan an optimal path. In complex maps with many obstacles, while the addition of an artificial potential field allows for step size variation, randomness remains a crucial factor in the RRT algorithm. Furthermore, excessively small or large repulsive or gravitational forces are simply due to proximity to the center of the repulsive or gravitational potential field, rather than adaptation to changes in the map environment. For example, multiple obstacles in close proximity can generate excessive repulsive force, causing the UAV to over-avoid obstacles. With a fixed step size, the same resultant force will produce the same step size variation in both simple and complex environments, impacting the time cost of path planning. Therefore, using a variable step size can better adapt to optimal path planning under varying environmental complexity.

[0099] The fixed step size for node generation in the RRT algorithm under the artificial potential field, based on randomness, adapts to the complexity of the environment in which the UAV is located; specifically:

[0100] The complexity of the environment in which the UAV is located is calculated based on the total area and density of obstacles in the aforementioned planar coordinate map;

[0101] A fixed step size based on randomness is determined based on the complexity of the environment in which the UAV is located and the distance between the starting point and the target point. The greater the complexity of the environment, that is, the more numerous and concentrated the obstacles in the map, the smaller the step size is required.

[0102] As a feasible implementation method, the complexity of the environment in which the drone operates can be determined by the following formula:

[0103]

[0104] Where μ represents the complexity of the environment in which the UAV is located, m and n are the weights of the total area and density of obstacles, respectively, and S obs S represents the total area of ​​obstacles in a planar coordinate map. map SQ represents the total area of ​​the planar coordinate map. obs Standard grid number SQ st The number of grid cells occupied by the lower obstacle.

[0105] As a feasible implementation method, the fixed step size C based on randomness can be determined by the following formula:

[0106]

[0107] Where d() represents the distance between the two points, x g Let x be the target point. s This is the starting point.

[0108] It should be noted that in this application, an artificial potential field is introduced to guide the direction and the step size is variable in the existing RRT algorithm, while other steps adopt the existing RRT algorithm.

[0109] Because the current optimal trajectory contains some node redundancy, it results in a large number of detours, increasing the cost of the UAV trajectory length and causing frequent maneuvers. Some nodes with a certain number of intervals can be directly connected without affecting the overall trajectory. Deleting the nodes between these nodes reduces the number of meaningless nodes on the trajectory, thus reducing the cost of the trajectory length.

[0110] Furthermore, since the currently obtained optimized trajectory is a polygonal trajectory, and this application is used for UAV trajectory planning, in reality, UAVs, due to inertia and physical constraints, have a maximum turning angle that prevents them from instantly changing large directions. Therefore, some turns on the current optimized trajectory cannot be completed by the UAV. Thus, these points need to be removed, and the optimized trajectory needs to be smoothed to meet the UAV's requirements for trajectory tracking flight. Specifically, the feasibility, effectiveness, and safety of the optimal trajectory are improved through node optimization and curve smoothing.

[0111] Two nodes separated by one node on the current optimal trajectory are denoted as an optimized node group, wherein the node closer to the starting point in the optimized node group is denoted as the starting point, and the node closer to the target point in the optimized node group is denoted as the ending point;

[0112] From the starting point to the target point, sequentially check whether the direct connection between two nodes in the optimized node group intersects with an obstacle and whether the minor angle of the intersection of the connecting line and the previous track is greater than the maximum turning angle of the UAV; the maximum turning angle of the UAV is the minor angle formed by the current track of the UAV and the track after the turn.

[0113] If the nodes do not intersect and the minor angle of the intersection of the connecting line and the previous track is greater than the maximum turning angle of the UAV, then the endpoint of the optimized node group in this group is deleted, and the starting point in the optimized node group is directly connected to the endpoint of the optimized node group in the previous group, until all nodes on the current optimal track are checked to obtain the optimized track.

[0114] The optimal UAV patrol trajectory is obtained by smoothing the optimized trajectory using a cubic B-spline interpolation algorithm.

[0115] This invention also provides an autonomous patrol trajectory planning system for power distribution network unmanned aerial vehicles (UAVs), such as... Figure 3 As shown, it includes:

[0116] The planar coordinate map creation module is used to create a planar coordinate map representing the starting point, target point, and obstacle positions of the UAV trajectory planning, with the straight line connecting the starting point and the target point as the x-axis and the straight line perpendicular to the x-axis as the y-axis.

[0117] The sampling point layout module is used to connect the starting point and the target point to obtain a direct track, determine an ideal track based on the intersection position of the direct track and the obstacle, and segment the planar coordinate map based on the ideal track and determine the sampling point layout.

[0118] The artificial potential field construction module is used to construct an artificial potential field by establishing a gravitational field centered on the starting point and the ideal trajectory, and by establishing a repulsive field centered on the obstacle.

[0119] The optimal trajectory planning module is used to plan the optimal trajectory for UAV patrol based on the RRT algorithm and guided by an artificial potential field. It builds a random tree from the target point in reverse to plan the optimal trajectory for the UAV patrol.

[0120] The present invention also provides an electronic device, comprising: a memory for storing executable instructions; and a processor for implementing, when running the executable instructions stored in the memory, a method for autonomous patrol trajectory planning of a power distribution network UAV as described above in this application.

[0121] The present invention also provides a computer-readable storage medium storing executable instructions, which, when executed by a processor, implement the autonomous inspection trajectory planning method for power distribution network unmanned aerial vehicles as described above.

[0122] This invention is not limited to the specific embodiments described above. Any modifications made by those skilled in the art based on the above concept without creative effort are within the scope of protection of this invention.

Claims

1. A method for autonomous patrol trajectory planning of power distribution network unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: A planar coordinate map is established based on the starting point, target point, and obstacle positions of the UAV trajectory planning, wherein the straight line connecting the starting point and the target point is the x-axis, and the straight line perpendicular to the x-axis is the y-axis. A direct track is obtained by connecting the starting point and the target point. An ideal track is determined based on the intersection of the direct track and the obstacle. The planar coordinate map is segmented based on the ideal track, and the sampling point distribution area is determined. The determination of the ideal trajectory based on the intersection position of the direct trajectory and the obstacle includes: The maximum and minimum points of the y-axis coordinates of the obstacles intersecting the direct flight path are recorded as the obstacle tangent points; Select one of the two obstacle tangents on each intersecting obstacle as the ideal track point such that the connecting line obtained by sequentially connecting the target point, each of the ideal track points and the starting point from the target point to the starting point is the shortest, and the shortest connecting line is the ideal track. The step of segmenting the planar coordinate map and determining the sampling point distribution area based on the ideal flight path includes: The planar coordinate map is divided into multiple segments using the selected obstacle tangent point as the segment endpoint; Based on the segmentation, the sampling point distribution area of ​​the UAV in each segment of the planar coordinate map is determined sequentially from the target point towards the starting point: Within each segment of the planar coordinate map, if the ideal trajectory of this segment passes through the x-axis or the two segment endpoints are the maximum and minimum points of the y-axis coordinates of the obstacles, then the areas within the upper and lower boundaries of the two obstacles closest to the ideal trajectory in the upper and lower regions of the planar coordinate map of this segment are selected as the sampling point distribution area in the planar coordinate map of this segment. Within each segment of the planar coordinate map, if both endpoints of the segment are the minimum points of the y-axis coordinates of the obstacle, then the area below the ideal flight path within the planar coordinate map of this segment is selected as the sampling point distribution area within the planar coordinate map of this segment. Within each segment of the planar coordinate map, if both endpoints of this segment are the maximum points of the y-axis coordinates of the obstacle, then the area above the ideal flight path within this segment of the planar coordinate map is selected as the sampling point distribution area within this segment of the planar coordinate map. Based on the sampling point distribution area, sampling points are randomly generated. An artificial potential field is constructed by establishing an gravitational field centered on the starting point and the ideal trajectory, and a repulsive field centered on the obstacle, to guide the search direction of the RRT algorithm. A random tree is built in reverse from the target point to plan the optimal trajectory for UAV patrol.

2. The method for autonomous inspection trajectory planning of power distribution network UAVs according to claim 1, characterized in that, The establishment of a gravitational field centered on the ideal trajectory includes: The ideal trajectory exerts an attractive force on the optimal trajectory for UAV inspection, denoted as the ideal trajectory gravity. The ideal trajectory gravity is always perpendicular to the ideal trajectory and points towards the ideal trajectory, causing the optimal trajectory for UAV inspection to tend towards the ideal trajectory. The ideal trajectory gravitational field function of the ideal trajectory gravity is: ； in, Let be the gravitational field function of the ideal trajectory. The gravitational constant of the ideal trajectory is... Indicates the current location of the drone and the ideal trajectory The distance between two locations.

3. The method for autonomous inspection trajectory planning of power distribution network UAVs according to claim 2, characterized in that, The construction of an artificial potential field by establishing gravitational and repulsive fields includes: The net force acting on the drone in the artificial potential field The target gravity generated by the target gravitational field centered at the starting point The ideal trajectory gravity generated by the ideal gravitational field centered on the ideal trajectory The repulsive force generated by the repulsive field centered on the obstacle sum; The node step size of the RRT algorithm under the artificial potential field depends on randomness and the guiding effect of the artificial potential field. The node generation formula is as follows: ； in, The new node generated by the random tree of the RRT algorithm under the artificial potential field. For a fixed step size based on randomness, Let be the net force acting on the UAV in the artificial potential field. Represents the magnitude of a vector. Weights for random factors, For the weights of artificial potential field factors, and .

4. The method for autonomous inspection trajectory planning of power distribution network UAVs according to claim 3, characterized in that, The fixed step size for node generation in the RRT algorithm under the artificial potential field, based on randomness, adapts to the complexity of the environment in which the UAV is located; specifically: The complexity of the environment in which the UAV is located is calculated based on the total area and dispersion of obstacles in the aforementioned planar coordinate map; A fixed step size based on randomness is determined based on the complexity of the environment in which the UAV is located and the distance between the starting point and the target point.

5. The method for autonomous inspection trajectory planning of power distribution network UAVs according to claim 1, characterized in that, After building the random tree backwards from the target point, it also includes: Two nodes separated by one node on the current optimal trajectory are denoted as an optimized node group, wherein the node closer to the starting point in the optimized node group is denoted as the starting point, and the node closer to the target point in the optimized node group is denoted as the ending point; From the starting point to the target point, sequentially check whether the direct connection between two nodes in the optimized node group intersects with an obstacle and whether the minor angle of the intersection of the connecting line with the previous flight path is greater than the maximum turning angle of the UAV; If the nodes do not intersect and the minor angle of the intersection of the connecting line and the previous track is greater than the maximum turning angle of the UAV, then the endpoint of the optimized node group in this group is deleted, and the starting point in the optimized node group is directly connected to the endpoint of the previous optimized node group, until all nodes on the current optimal track are checked to obtain the optimized track. The optimal UAV patrol trajectory is obtained by smoothing the optimized trajectory using a cubic B-spline interpolation algorithm.

6. A system for autonomous patrol trajectory planning of power distribution network unmanned aerial vehicles (UAVs), characterized in that, include: The planar coordinate map creation module is used to create a planar coordinate map representing the starting point, target point, and obstacle positions of the UAV trajectory planning, with the straight line connecting the starting point and the target point as the x-axis and the straight line perpendicular to the x-axis as the y-axis. The sampling point layout module is used to connect the starting point and the target point to obtain a direct track, determine an ideal track based on the intersection position of the direct track and the obstacle, and segment the planar coordinate map based on the ideal track and determine the sampling point layout. The determination of the ideal trajectory based on the intersection position of the direct trajectory and the obstacle includes: The maximum and minimum points of the y-axis coordinates of the obstacles intersecting the direct flight path are recorded as the obstacle tangent points; Select one of the two obstacle tangents on each intersecting obstacle as the ideal track point such that the connecting line obtained by sequentially connecting the target point, each of the ideal track points and the starting point from the target point to the starting point is the shortest, and the shortest connecting line is the ideal track. The step of segmenting the planar coordinate map and determining the sampling point distribution area based on the ideal flight path includes: The planar coordinate map is divided into multiple segments using the selected obstacle tangent point as the segment endpoint; Based on the segmentation, the sampling point distribution area of ​​the UAV in each segment of the planar coordinate map is determined sequentially from the target point towards the starting point: Within each segment of the planar coordinate map, if the ideal trajectory of this segment passes through the x-axis or the two segment endpoints are the maximum and minimum points of the y-axis coordinates of the obstacles, then the areas within the upper and lower boundaries of the two obstacles closest to the ideal trajectory in the upper and lower regions of the planar coordinate map of this segment are selected as the sampling point distribution area in the planar coordinate map of this segment. Within each segment of the planar coordinate map, if both endpoints of the segment are the minimum points of the y-axis coordinates of the obstacle, then the area below the ideal flight path within the planar coordinate map of this segment is selected as the sampling point distribution area within the planar coordinate map of this segment. Within each segment of the planar coordinate map, if both endpoints of this segment are the maximum points of the y-axis coordinates of the obstacle, then the area above the ideal flight path within this segment of the planar coordinate map is selected as the sampling point distribution area within this segment of the planar coordinate map. The artificial potential field construction module is used to construct an artificial potential field by establishing a gravitational field centered on the starting point and the ideal trajectory, and by establishing a repulsive field centered on the obstacle. The optimal trajectory planning module is used to plan the optimal trajectory for UAV patrol based on the RRT algorithm and guided by an artificial potential field. It builds a random tree from the target point in reverse to plan the optimal trajectory for the UAV patrol.

7. An electronic device, characterized in that, The electronic device includes: Memory, used to store executable instructions; The processor, when running the executable instructions stored in the memory, implements the autonomous patrol trajectory planning method for power distribution network UAVs as described in any one of claims 1 to 5.

8. A computer-readable storage medium storing executable instructions, characterized in that, When the executable instructions are executed by the processor, they implement the autonomous patrol trajectory planning method for power distribution unmanned aerial vehicles as described in any one of claims 1 to 5.

Images